2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* Simple NUMA memory policy for the Linux kernel.
|
|
|
|
*
|
|
|
|
* Copyright 2003,2004 Andi Kleen, SuSE Labs.
|
2005-10-29 19:16:59 -06:00
|
|
|
* (C) Copyright 2005 Christoph Lameter, Silicon Graphics, Inc.
|
2005-04-16 16:20:36 -06:00
|
|
|
* Subject to the GNU Public License, version 2.
|
|
|
|
*
|
|
|
|
* NUMA policy allows the user to give hints in which node(s) memory should
|
|
|
|
* be allocated.
|
|
|
|
*
|
|
|
|
* Support four policies per VMA and per process:
|
|
|
|
*
|
|
|
|
* The VMA policy has priority over the process policy for a page fault.
|
|
|
|
*
|
|
|
|
* interleave Allocate memory interleaved over a set of nodes,
|
|
|
|
* with normal fallback if it fails.
|
|
|
|
* For VMA based allocations this interleaves based on the
|
|
|
|
* offset into the backing object or offset into the mapping
|
|
|
|
* for anonymous memory. For process policy an process counter
|
|
|
|
* is used.
|
2005-10-29 19:16:59 -06:00
|
|
|
*
|
2005-04-16 16:20:36 -06:00
|
|
|
* bind Only allocate memory on a specific set of nodes,
|
|
|
|
* no fallback.
|
2005-10-29 19:16:59 -06:00
|
|
|
* FIXME: memory is allocated starting with the first node
|
|
|
|
* to the last. It would be better if bind would truly restrict
|
|
|
|
* the allocation to memory nodes instead
|
|
|
|
*
|
2005-04-16 16:20:36 -06:00
|
|
|
* preferred Try a specific node first before normal fallback.
|
2013-02-22 17:35:36 -07:00
|
|
|
* As a special case NUMA_NO_NODE here means do the allocation
|
2005-04-16 16:20:36 -06:00
|
|
|
* on the local CPU. This is normally identical to default,
|
|
|
|
* but useful to set in a VMA when you have a non default
|
|
|
|
* process policy.
|
2005-10-29 19:16:59 -06:00
|
|
|
*
|
2005-04-16 16:20:36 -06:00
|
|
|
* default Allocate on the local node first, or when on a VMA
|
|
|
|
* use the process policy. This is what Linux always did
|
|
|
|
* in a NUMA aware kernel and still does by, ahem, default.
|
|
|
|
*
|
|
|
|
* The process policy is applied for most non interrupt memory allocations
|
|
|
|
* in that process' context. Interrupts ignore the policies and always
|
|
|
|
* try to allocate on the local CPU. The VMA policy is only applied for memory
|
|
|
|
* allocations for a VMA in the VM.
|
|
|
|
*
|
|
|
|
* Currently there are a few corner cases in swapping where the policy
|
|
|
|
* is not applied, but the majority should be handled. When process policy
|
|
|
|
* is used it is not remembered over swap outs/swap ins.
|
|
|
|
*
|
|
|
|
* Only the highest zone in the zone hierarchy gets policied. Allocations
|
|
|
|
* requesting a lower zone just use default policy. This implies that
|
|
|
|
* on systems with highmem kernel lowmem allocation don't get policied.
|
|
|
|
* Same with GFP_DMA allocations.
|
|
|
|
*
|
|
|
|
* For shmfs/tmpfs/hugetlbfs shared memory the policy is shared between
|
|
|
|
* all users and remembered even when nobody has memory mapped.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* Notebook:
|
|
|
|
fix mmap readahead to honour policy and enable policy for any page cache
|
|
|
|
object
|
|
|
|
statistics for bigpages
|
|
|
|
global policy for page cache? currently it uses process policy. Requires
|
|
|
|
first item above.
|
|
|
|
handle mremap for shared memory (currently ignored for the policy)
|
|
|
|
grows down?
|
|
|
|
make bind policy root only? It can trigger oom much faster and the
|
|
|
|
kernel is not always grateful with that.
|
|
|
|
*/
|
|
|
|
|
2014-06-06 15:38:30 -06:00
|
|
|
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
#include <linux/mempolicy.h>
|
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/highmem.h>
|
|
|
|
#include <linux/hugetlb.h>
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <linux/nodemask.h>
|
|
|
|
#include <linux/cpuset.h>
|
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/string.h>
|
2011-10-16 00:01:52 -06:00
|
|
|
#include <linux/export.h>
|
2007-10-19 00:40:14 -06:00
|
|
|
#include <linux/nsproxy.h>
|
2005-04-16 16:20:36 -06:00
|
|
|
#include <linux/interrupt.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/compat.h>
|
2006-01-08 02:00:50 -07:00
|
|
|
#include <linux/swap.h>
|
2006-01-08 02:01:02 -07:00
|
|
|
#include <linux/seq_file.h>
|
|
|
|
#include <linux/proc_fs.h>
|
2006-03-22 01:09:12 -07:00
|
|
|
#include <linux/migrate.h>
|
ksm: memory hotremove migration only
The previous patch enables page migration of ksm pages, but that soon gets
into trouble: not surprising, since we're using the ksm page lock to lock
operations on its stable_node, but page migration switches the page whose
lock is to be used for that. Another layer of locking would fix it, but
do we need that yet?
Do we actually need page migration of ksm pages? Yes, memory hotremove
needs to offline sections of memory: and since we stopped allocating ksm
pages with GFP_HIGHUSER, they will tend to be GFP_HIGHUSER_MOVABLE
candidates for migration.
But KSM is currently unconscious of NUMA issues, happily merging pages
from different NUMA nodes: at present the rule must be, not to use
MADV_MERGEABLE where you care about NUMA. So no, NUMA page migration of
ksm pages does not make sense yet.
So, to complete support for ksm swapping we need to make hotremove safe.
ksm_memory_callback() take ksm_thread_mutex when MEM_GOING_OFFLINE and
release it when MEM_OFFLINE or MEM_CANCEL_OFFLINE. But if mapped pages
are freed before migration reaches them, stable_nodes may be left still
pointing to struct pages which have been removed from the system: the
stable_node needs to identify a page by pfn rather than page pointer, then
it can safely prune them when MEM_OFFLINE.
And make NUMA migration skip PageKsm pages where it skips PageReserved.
But it's only when we reach unmap_and_move() that the page lock is taken
and we can be sure that raised pagecount has prevented a PageAnon from
being upgraded: so add offlining arg to migrate_pages(), to migrate ksm
page when offlining (has sufficient locking) but reject it otherwise.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Izik Eidus <ieidus@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Chris Wright <chrisw@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-14 18:59:33 -07:00
|
|
|
#include <linux/ksm.h>
|
2006-06-23 03:03:53 -06:00
|
|
|
#include <linux/rmap.h>
|
2006-06-23 03:04:02 -06:00
|
|
|
#include <linux/security.h>
|
2007-10-16 02:26:26 -06:00
|
|
|
#include <linux/syscalls.h>
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
#include <linux/ctype.h>
|
2009-12-14 18:58:11 -07:00
|
|
|
#include <linux/mm_inline.h>
|
2012-10-25 06:16:32 -06:00
|
|
|
#include <linux/mmu_notifier.h>
|
2014-06-06 15:38:30 -06:00
|
|
|
#include <linux/printk.h>
|
2006-01-08 02:00:50 -07:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
#include <asm/tlbflush.h>
|
|
|
|
#include <asm/uaccess.h>
|
|
|
|
|
vmscan: move isolate_lru_page() to vmscan.c
On large memory systems, the VM can spend way too much time scanning
through pages that it cannot (or should not) evict from memory. Not only
does it use up CPU time, but it also provokes lock contention and can
leave large systems under memory presure in a catatonic state.
This patch series improves VM scalability by:
1) putting filesystem backed, swap backed and unevictable pages
onto their own LRUs, so the system only scans the pages that it
can/should evict from memory
2) switching to two handed clock replacement for the anonymous LRUs,
so the number of pages that need to be scanned when the system
starts swapping is bound to a reasonable number
3) keeping unevictable pages off the LRU completely, so the
VM does not waste CPU time scanning them. ramfs, ramdisk,
SHM_LOCKED shared memory segments and mlock()ed VMA pages
are keept on the unevictable list.
This patch:
isolate_lru_page logically belongs to be in vmscan.c than migrate.c.
It is tough, because we don't need that function without memory migration
so there is a valid argument to have it in migrate.c. However a
subsequent patch needs to make use of it in the core mm, so we can happily
move it to vmscan.c.
Also, make the function a little more generic by not requiring that it
adds an isolated page to a given list. Callers can do that.
Note that we now have '__isolate_lru_page()', that does
something quite different, visible outside of vmscan.c
for use with memory controller. Methinks we need to
rationalize these names/purposes. --lts
[akpm@linux-foundation.org: fix mm/memory_hotplug.c build]
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-10-18 21:26:09 -06:00
|
|
|
#include "internal.h"
|
|
|
|
|
2006-01-08 02:01:01 -07:00
|
|
|
/* Internal flags */
|
2006-01-08 02:00:50 -07:00
|
|
|
#define MPOL_MF_DISCONTIG_OK (MPOL_MF_INTERNAL << 0) /* Skip checks for continuous vmas */
|
2006-01-08 02:01:01 -07:00
|
|
|
#define MPOL_MF_INVERT (MPOL_MF_INTERNAL << 1) /* Invert check for nodemask */
|
2006-01-08 02:00:50 -07:00
|
|
|
|
2006-03-22 01:08:13 -07:00
|
|
|
static struct kmem_cache *policy_cache;
|
|
|
|
static struct kmem_cache *sn_cache;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
/* Highest zone. An specific allocation for a zone below that is not
|
|
|
|
policied. */
|
2007-02-10 02:43:07 -07:00
|
|
|
enum zone_type policy_zone = 0;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
/*
|
|
|
|
* run-time system-wide default policy => local allocation
|
|
|
|
*/
|
2011-10-31 18:09:23 -06:00
|
|
|
static struct mempolicy default_policy = {
|
2005-04-16 16:20:36 -06:00
|
|
|
.refcnt = ATOMIC_INIT(1), /* never free it */
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
.mode = MPOL_PREFERRED,
|
2008-04-28 03:13:21 -06:00
|
|
|
.flags = MPOL_F_LOCAL,
|
2005-04-16 16:20:36 -06:00
|
|
|
};
|
|
|
|
|
2012-11-02 12:19:13 -06:00
|
|
|
static struct mempolicy preferred_node_policy[MAX_NUMNODES];
|
|
|
|
|
2014-10-09 16:27:50 -06:00
|
|
|
struct mempolicy *get_task_policy(struct task_struct *p)
|
2012-11-02 12:19:13 -06:00
|
|
|
{
|
|
|
|
struct mempolicy *pol = p->mempolicy;
|
2014-10-09 16:27:43 -06:00
|
|
|
int node;
|
2012-11-02 12:19:13 -06:00
|
|
|
|
2014-10-09 16:27:43 -06:00
|
|
|
if (pol)
|
|
|
|
return pol;
|
2012-11-02 12:19:13 -06:00
|
|
|
|
2014-10-09 16:27:43 -06:00
|
|
|
node = numa_node_id();
|
|
|
|
if (node != NUMA_NO_NODE) {
|
|
|
|
pol = &preferred_node_policy[node];
|
|
|
|
/* preferred_node_policy is not initialised early in boot */
|
|
|
|
if (pol->mode)
|
|
|
|
return pol;
|
2012-11-02 12:19:13 -06:00
|
|
|
}
|
|
|
|
|
2014-10-09 16:27:43 -06:00
|
|
|
return &default_policy;
|
2012-11-02 12:19:13 -06:00
|
|
|
}
|
|
|
|
|
2008-04-28 03:12:33 -06:00
|
|
|
static const struct mempolicy_operations {
|
|
|
|
int (*create)(struct mempolicy *pol, const nodemask_t *nodes);
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
/*
|
|
|
|
* If read-side task has no lock to protect task->mempolicy, write-side
|
|
|
|
* task will rebind the task->mempolicy by two step. The first step is
|
|
|
|
* setting all the newly nodes, and the second step is cleaning all the
|
|
|
|
* disallowed nodes. In this way, we can avoid finding no node to alloc
|
|
|
|
* page.
|
|
|
|
* If we have a lock to protect task->mempolicy in read-side, we do
|
|
|
|
* rebind directly.
|
|
|
|
*
|
|
|
|
* step:
|
|
|
|
* MPOL_REBIND_ONCE - do rebind work at once
|
|
|
|
* MPOL_REBIND_STEP1 - set all the newly nodes
|
|
|
|
* MPOL_REBIND_STEP2 - clean all the disallowed nodes
|
|
|
|
*/
|
|
|
|
void (*rebind)(struct mempolicy *pol, const nodemask_t *nodes,
|
|
|
|
enum mpol_rebind_step step);
|
2008-04-28 03:12:33 -06:00
|
|
|
} mpol_ops[MPOL_MAX];
|
|
|
|
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
static inline int mpol_store_user_nodemask(const struct mempolicy *pol)
|
|
|
|
{
|
2010-05-24 15:31:59 -06:00
|
|
|
return pol->flags & MPOL_MODE_FLAGS;
|
mempolicy: add MPOL_F_RELATIVE_NODES flag
Adds another optional mode flag, MPOL_F_RELATIVE_NODES, that specifies
nodemasks passed via set_mempolicy() or mbind() should be considered relative
to the current task's mems_allowed.
When the mempolicy is created, the passed nodemask is folded and mapped onto
the current task's mems_allowed. For example, consider a task using
set_mempolicy() to pass MPOL_INTERLEAVE | MPOL_F_RELATIVE_NODES with a
nodemask of 1-3. If current's mems_allowed is 4-7, the effected nodemask is
5-7 (the second, third, and fourth node of mems_allowed).
If the same task is attached to a cpuset, the mempolicy nodemask is rebound
each time the mems are changed. Some possible rebinds and results are:
mems result
1-3 1-3
1-7 2-4
1,5-6 1,5-6
1,5-7 5-7
Likewise, the zonelist built for MPOL_BIND acts on the set of zones assigned
to the resultant nodemask from the relative remap.
In the MPOL_PREFERRED case, the preferred node is remapped from the currently
effected nodemask to the relative nodemask.
This mempolicy mode flag was conceived of by Paul Jackson <pj@sgi.com>.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:30 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
static void mpol_relative_nodemask(nodemask_t *ret, const nodemask_t *orig,
|
|
|
|
const nodemask_t *rel)
|
|
|
|
{
|
|
|
|
nodemask_t tmp;
|
|
|
|
nodes_fold(tmp, *orig, nodes_weight(*rel));
|
|
|
|
nodes_onto(*ret, tmp, *rel);
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
}
|
|
|
|
|
2008-04-28 03:12:33 -06:00
|
|
|
static int mpol_new_interleave(struct mempolicy *pol, const nodemask_t *nodes)
|
|
|
|
{
|
|
|
|
if (nodes_empty(*nodes))
|
|
|
|
return -EINVAL;
|
|
|
|
pol->v.nodes = *nodes;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int mpol_new_preferred(struct mempolicy *pol, const nodemask_t *nodes)
|
|
|
|
{
|
|
|
|
if (!nodes)
|
2008-04-28 03:13:21 -06:00
|
|
|
pol->flags |= MPOL_F_LOCAL; /* local allocation */
|
2008-04-28 03:12:33 -06:00
|
|
|
else if (nodes_empty(*nodes))
|
|
|
|
return -EINVAL; /* no allowed nodes */
|
|
|
|
else
|
|
|
|
pol->v.preferred_node = first_node(*nodes);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int mpol_new_bind(struct mempolicy *pol, const nodemask_t *nodes)
|
|
|
|
{
|
2014-12-18 17:17:09 -07:00
|
|
|
if (nodes_empty(*nodes))
|
2008-04-28 03:12:33 -06:00
|
|
|
return -EINVAL;
|
|
|
|
pol->v.nodes = *nodes;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
/*
|
|
|
|
* mpol_set_nodemask is called after mpol_new() to set up the nodemask, if
|
|
|
|
* any, for the new policy. mpol_new() has already validated the nodes
|
|
|
|
* parameter with respect to the policy mode and flags. But, we need to
|
|
|
|
* handle an empty nodemask with MPOL_PREFERRED here.
|
|
|
|
*
|
|
|
|
* Must be called holding task's alloc_lock to protect task's mems_allowed
|
|
|
|
* and mempolicy. May also be called holding the mmap_semaphore for write.
|
|
|
|
*/
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
static int mpol_set_nodemask(struct mempolicy *pol,
|
|
|
|
const nodemask_t *nodes, struct nodemask_scratch *nsc)
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
/* if mode is MPOL_DEFAULT, pol is NULL. This is right. */
|
|
|
|
if (pol == NULL)
|
|
|
|
return 0;
|
2012-12-12 14:51:33 -07:00
|
|
|
/* Check N_MEMORY */
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
nodes_and(nsc->mask1,
|
2012-12-12 14:51:33 -07:00
|
|
|
cpuset_current_mems_allowed, node_states[N_MEMORY]);
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
|
|
|
|
VM_BUG_ON(!nodes);
|
|
|
|
if (pol->mode == MPOL_PREFERRED && nodes_empty(*nodes))
|
|
|
|
nodes = NULL; /* explicit local allocation */
|
|
|
|
else {
|
|
|
|
if (pol->flags & MPOL_F_RELATIVE_NODES)
|
2014-12-18 17:17:09 -07:00
|
|
|
mpol_relative_nodemask(&nsc->mask2, nodes, &nsc->mask1);
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
else
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
nodes_and(nsc->mask2, *nodes, nsc->mask1);
|
|
|
|
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
if (mpol_store_user_nodemask(pol))
|
|
|
|
pol->w.user_nodemask = *nodes;
|
|
|
|
else
|
|
|
|
pol->w.cpuset_mems_allowed =
|
|
|
|
cpuset_current_mems_allowed;
|
|
|
|
}
|
|
|
|
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
if (nodes)
|
|
|
|
ret = mpol_ops[pol->mode].create(pol, &nsc->mask2);
|
|
|
|
else
|
|
|
|
ret = mpol_ops[pol->mode].create(pol, NULL);
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This function just creates a new policy, does some check and simple
|
|
|
|
* initialization. You must invoke mpol_set_nodemask() to set nodes.
|
|
|
|
*/
|
2008-04-28 03:12:25 -06:00
|
|
|
static struct mempolicy *mpol_new(unsigned short mode, unsigned short flags,
|
|
|
|
nodemask_t *nodes)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct mempolicy *policy;
|
|
|
|
|
2008-04-28 03:12:25 -06:00
|
|
|
pr_debug("setting mode %d flags %d nodes[0] %lx\n",
|
2013-02-22 17:35:36 -07:00
|
|
|
mode, flags, nodes ? nodes_addr(*nodes)[0] : NUMA_NO_NODE);
|
2007-07-16 00:38:16 -06:00
|
|
|
|
2008-04-28 03:12:34 -06:00
|
|
|
if (mode == MPOL_DEFAULT) {
|
|
|
|
if (nodes && !nodes_empty(*nodes))
|
2008-04-28 03:12:33 -06:00
|
|
|
return ERR_PTR(-EINVAL);
|
2012-10-25 06:16:29 -06:00
|
|
|
return NULL;
|
2008-04-28 03:12:33 -06:00
|
|
|
}
|
2008-04-28 03:12:34 -06:00
|
|
|
VM_BUG_ON(!nodes);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* MPOL_PREFERRED cannot be used with MPOL_F_STATIC_NODES or
|
|
|
|
* MPOL_F_RELATIVE_NODES if the nodemask is empty (local allocation).
|
|
|
|
* All other modes require a valid pointer to a non-empty nodemask.
|
|
|
|
*/
|
|
|
|
if (mode == MPOL_PREFERRED) {
|
|
|
|
if (nodes_empty(*nodes)) {
|
|
|
|
if (((flags & MPOL_F_STATIC_NODES) ||
|
|
|
|
(flags & MPOL_F_RELATIVE_NODES)))
|
|
|
|
return ERR_PTR(-EINVAL);
|
|
|
|
}
|
2012-10-25 06:16:28 -06:00
|
|
|
} else if (mode == MPOL_LOCAL) {
|
|
|
|
if (!nodes_empty(*nodes))
|
|
|
|
return ERR_PTR(-EINVAL);
|
|
|
|
mode = MPOL_PREFERRED;
|
2008-04-28 03:12:34 -06:00
|
|
|
} else if (nodes_empty(*nodes))
|
|
|
|
return ERR_PTR(-EINVAL);
|
2005-04-16 16:20:36 -06:00
|
|
|
policy = kmem_cache_alloc(policy_cache, GFP_KERNEL);
|
|
|
|
if (!policy)
|
|
|
|
return ERR_PTR(-ENOMEM);
|
|
|
|
atomic_set(&policy->refcnt, 1);
|
2008-04-28 03:13:12 -06:00
|
|
|
policy->mode = mode;
|
2008-04-28 03:12:34 -06:00
|
|
|
policy->flags = flags;
|
2008-04-28 03:12:33 -06:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
return policy;
|
2008-04-28 03:12:33 -06:00
|
|
|
}
|
|
|
|
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
/* Slow path of a mpol destructor. */
|
|
|
|
void __mpol_put(struct mempolicy *p)
|
|
|
|
{
|
|
|
|
if (!atomic_dec_and_test(&p->refcnt))
|
|
|
|
return;
|
|
|
|
kmem_cache_free(policy_cache, p);
|
|
|
|
}
|
|
|
|
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
static void mpol_rebind_default(struct mempolicy *pol, const nodemask_t *nodes,
|
|
|
|
enum mpol_rebind_step step)
|
2008-04-28 03:12:33 -06:00
|
|
|
{
|
|
|
|
}
|
|
|
|
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
/*
|
|
|
|
* step:
|
|
|
|
* MPOL_REBIND_ONCE - do rebind work at once
|
|
|
|
* MPOL_REBIND_STEP1 - set all the newly nodes
|
|
|
|
* MPOL_REBIND_STEP2 - clean all the disallowed nodes
|
|
|
|
*/
|
|
|
|
static void mpol_rebind_nodemask(struct mempolicy *pol, const nodemask_t *nodes,
|
|
|
|
enum mpol_rebind_step step)
|
2008-04-28 03:12:33 -06:00
|
|
|
{
|
|
|
|
nodemask_t tmp;
|
|
|
|
|
|
|
|
if (pol->flags & MPOL_F_STATIC_NODES)
|
|
|
|
nodes_and(tmp, pol->w.user_nodemask, *nodes);
|
|
|
|
else if (pol->flags & MPOL_F_RELATIVE_NODES)
|
|
|
|
mpol_relative_nodemask(&tmp, &pol->w.user_nodemask, nodes);
|
|
|
|
else {
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
/*
|
|
|
|
* if step == 1, we use ->w.cpuset_mems_allowed to cache the
|
|
|
|
* result
|
|
|
|
*/
|
|
|
|
if (step == MPOL_REBIND_ONCE || step == MPOL_REBIND_STEP1) {
|
|
|
|
nodes_remap(tmp, pol->v.nodes,
|
|
|
|
pol->w.cpuset_mems_allowed, *nodes);
|
|
|
|
pol->w.cpuset_mems_allowed = step ? tmp : *nodes;
|
|
|
|
} else if (step == MPOL_REBIND_STEP2) {
|
|
|
|
tmp = pol->w.cpuset_mems_allowed;
|
|
|
|
pol->w.cpuset_mems_allowed = *nodes;
|
|
|
|
} else
|
|
|
|
BUG();
|
2008-04-28 03:12:33 -06:00
|
|
|
}
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
if (nodes_empty(tmp))
|
|
|
|
tmp = *nodes;
|
|
|
|
|
|
|
|
if (step == MPOL_REBIND_STEP1)
|
|
|
|
nodes_or(pol->v.nodes, pol->v.nodes, tmp);
|
|
|
|
else if (step == MPOL_REBIND_ONCE || step == MPOL_REBIND_STEP2)
|
|
|
|
pol->v.nodes = tmp;
|
|
|
|
else
|
|
|
|
BUG();
|
|
|
|
|
2008-04-28 03:12:33 -06:00
|
|
|
if (!node_isset(current->il_next, tmp)) {
|
2016-05-19 18:10:58 -06:00
|
|
|
current->il_next = next_node_in(current->il_next, tmp);
|
2008-04-28 03:12:33 -06:00
|
|
|
if (current->il_next >= MAX_NUMNODES)
|
|
|
|
current->il_next = numa_node_id();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void mpol_rebind_preferred(struct mempolicy *pol,
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
const nodemask_t *nodes,
|
|
|
|
enum mpol_rebind_step step)
|
2008-04-28 03:12:33 -06:00
|
|
|
{
|
|
|
|
nodemask_t tmp;
|
|
|
|
|
|
|
|
if (pol->flags & MPOL_F_STATIC_NODES) {
|
|
|
|
int node = first_node(pol->w.user_nodemask);
|
|
|
|
|
2008-04-28 03:13:21 -06:00
|
|
|
if (node_isset(node, *nodes)) {
|
2008-04-28 03:12:33 -06:00
|
|
|
pol->v.preferred_node = node;
|
2008-04-28 03:13:21 -06:00
|
|
|
pol->flags &= ~MPOL_F_LOCAL;
|
|
|
|
} else
|
|
|
|
pol->flags |= MPOL_F_LOCAL;
|
2008-04-28 03:12:33 -06:00
|
|
|
} else if (pol->flags & MPOL_F_RELATIVE_NODES) {
|
|
|
|
mpol_relative_nodemask(&tmp, &pol->w.user_nodemask, nodes);
|
|
|
|
pol->v.preferred_node = first_node(tmp);
|
2008-04-28 03:13:21 -06:00
|
|
|
} else if (!(pol->flags & MPOL_F_LOCAL)) {
|
2008-04-28 03:12:33 -06:00
|
|
|
pol->v.preferred_node = node_remap(pol->v.preferred_node,
|
|
|
|
pol->w.cpuset_mems_allowed,
|
|
|
|
*nodes);
|
|
|
|
pol->w.cpuset_mems_allowed = *nodes;
|
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
/*
|
|
|
|
* mpol_rebind_policy - Migrate a policy to a different set of nodes
|
|
|
|
*
|
|
|
|
* If read-side task has no lock to protect task->mempolicy, write-side
|
|
|
|
* task will rebind the task->mempolicy by two step. The first step is
|
|
|
|
* setting all the newly nodes, and the second step is cleaning all the
|
|
|
|
* disallowed nodes. In this way, we can avoid finding no node to alloc
|
|
|
|
* page.
|
|
|
|
* If we have a lock to protect task->mempolicy in read-side, we do
|
|
|
|
* rebind directly.
|
|
|
|
*
|
|
|
|
* step:
|
|
|
|
* MPOL_REBIND_ONCE - do rebind work at once
|
|
|
|
* MPOL_REBIND_STEP1 - set all the newly nodes
|
|
|
|
* MPOL_REBIND_STEP2 - clean all the disallowed nodes
|
|
|
|
*/
|
|
|
|
static void mpol_rebind_policy(struct mempolicy *pol, const nodemask_t *newmask,
|
|
|
|
enum mpol_rebind_step step)
|
2008-04-28 03:12:32 -06:00
|
|
|
{
|
|
|
|
if (!pol)
|
|
|
|
return;
|
2012-05-29 16:06:16 -06:00
|
|
|
if (!mpol_store_user_nodemask(pol) && step == MPOL_REBIND_ONCE &&
|
2008-04-28 03:12:32 -06:00
|
|
|
nodes_equal(pol->w.cpuset_mems_allowed, *newmask))
|
|
|
|
return;
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
|
|
|
|
if (step == MPOL_REBIND_STEP1 && (pol->flags & MPOL_F_REBINDING))
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (step == MPOL_REBIND_STEP2 && !(pol->flags & MPOL_F_REBINDING))
|
|
|
|
BUG();
|
|
|
|
|
|
|
|
if (step == MPOL_REBIND_STEP1)
|
|
|
|
pol->flags |= MPOL_F_REBINDING;
|
|
|
|
else if (step == MPOL_REBIND_STEP2)
|
|
|
|
pol->flags &= ~MPOL_F_REBINDING;
|
|
|
|
else if (step >= MPOL_REBIND_NSTEP)
|
|
|
|
BUG();
|
|
|
|
|
|
|
|
mpol_ops[pol->mode].rebind(pol, newmask, step);
|
2008-04-28 03:12:32 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Wrapper for mpol_rebind_policy() that just requires task
|
|
|
|
* pointer, and updates task mempolicy.
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
*
|
|
|
|
* Called with task's alloc_lock held.
|
2008-04-28 03:12:32 -06:00
|
|
|
*/
|
|
|
|
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
void mpol_rebind_task(struct task_struct *tsk, const nodemask_t *new,
|
|
|
|
enum mpol_rebind_step step)
|
2008-04-28 03:12:32 -06:00
|
|
|
{
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
mpol_rebind_policy(tsk->mempolicy, new, step);
|
2008-04-28 03:12:32 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Rebind each vma in mm to new nodemask.
|
|
|
|
*
|
|
|
|
* Call holding a reference to mm. Takes mm->mmap_sem during call.
|
|
|
|
*/
|
|
|
|
|
|
|
|
void mpol_rebind_mm(struct mm_struct *mm, nodemask_t *new)
|
|
|
|
{
|
|
|
|
struct vm_area_struct *vma;
|
|
|
|
|
|
|
|
down_write(&mm->mmap_sem);
|
|
|
|
for (vma = mm->mmap; vma; vma = vma->vm_next)
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
mpol_rebind_policy(vma->vm_policy, new, MPOL_REBIND_ONCE);
|
2008-04-28 03:12:32 -06:00
|
|
|
up_write(&mm->mmap_sem);
|
|
|
|
}
|
|
|
|
|
2008-04-28 03:12:33 -06:00
|
|
|
static const struct mempolicy_operations mpol_ops[MPOL_MAX] = {
|
|
|
|
[MPOL_DEFAULT] = {
|
|
|
|
.rebind = mpol_rebind_default,
|
|
|
|
},
|
|
|
|
[MPOL_INTERLEAVE] = {
|
|
|
|
.create = mpol_new_interleave,
|
|
|
|
.rebind = mpol_rebind_nodemask,
|
|
|
|
},
|
|
|
|
[MPOL_PREFERRED] = {
|
|
|
|
.create = mpol_new_preferred,
|
|
|
|
.rebind = mpol_rebind_preferred,
|
|
|
|
},
|
|
|
|
[MPOL_BIND] = {
|
|
|
|
.create = mpol_new_bind,
|
|
|
|
.rebind = mpol_rebind_nodemask,
|
|
|
|
},
|
|
|
|
};
|
|
|
|
|
2006-01-18 18:42:29 -07:00
|
|
|
static void migrate_page_add(struct page *page, struct list_head *pagelist,
|
|
|
|
unsigned long flags);
|
2006-01-08 02:01:02 -07:00
|
|
|
|
2015-02-11 16:28:03 -07:00
|
|
|
struct queue_pages {
|
|
|
|
struct list_head *pagelist;
|
|
|
|
unsigned long flags;
|
|
|
|
nodemask_t *nmask;
|
|
|
|
struct vm_area_struct *prev;
|
|
|
|
};
|
|
|
|
|
2013-09-11 15:22:14 -06:00
|
|
|
/*
|
|
|
|
* Scan through pages checking if pages follow certain conditions,
|
|
|
|
* and move them to the pagelist if they do.
|
|
|
|
*/
|
2015-02-11 16:28:03 -07:00
|
|
|
static int queue_pages_pte_range(pmd_t *pmd, unsigned long addr,
|
|
|
|
unsigned long end, struct mm_walk *walk)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2015-02-11 16:28:03 -07:00
|
|
|
struct vm_area_struct *vma = walk->vma;
|
|
|
|
struct page *page;
|
|
|
|
struct queue_pages *qp = walk->private;
|
|
|
|
unsigned long flags = qp->flags;
|
2016-01-15 17:54:14 -07:00
|
|
|
int nid, ret;
|
2005-06-21 18:15:07 -06:00
|
|
|
pte_t *pte;
|
2005-10-29 19:16:27 -06:00
|
|
|
spinlock_t *ptl;
|
2005-06-21 18:15:06 -06:00
|
|
|
|
2016-01-15 17:54:14 -07:00
|
|
|
if (pmd_trans_huge(*pmd)) {
|
|
|
|
ptl = pmd_lock(walk->mm, pmd);
|
|
|
|
if (pmd_trans_huge(*pmd)) {
|
|
|
|
page = pmd_page(*pmd);
|
|
|
|
if (is_huge_zero_page(page)) {
|
|
|
|
spin_unlock(ptl);
|
|
|
|
split_huge_pmd(vma, pmd, addr);
|
|
|
|
} else {
|
|
|
|
get_page(page);
|
|
|
|
spin_unlock(ptl);
|
|
|
|
lock_page(page);
|
|
|
|
ret = split_huge_page(page);
|
|
|
|
unlock_page(page);
|
|
|
|
put_page(page);
|
|
|
|
if (ret)
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
spin_unlock(ptl);
|
|
|
|
}
|
|
|
|
}
|
2005-06-21 18:15:07 -06:00
|
|
|
|
2016-07-26 16:24:03 -06:00
|
|
|
if (pmd_trans_unstable(pmd))
|
|
|
|
return 0;
|
2016-01-15 17:54:14 -07:00
|
|
|
retry:
|
2015-02-11 16:28:03 -07:00
|
|
|
pte = pte_offset_map_lock(walk->mm, pmd, addr, &ptl);
|
|
|
|
for (; addr != end; pte++, addr += PAGE_SIZE) {
|
2005-06-21 18:15:07 -06:00
|
|
|
if (!pte_present(*pte))
|
2005-04-16 16:20:36 -06:00
|
|
|
continue;
|
2005-11-28 15:34:23 -07:00
|
|
|
page = vm_normal_page(vma, addr, *pte);
|
|
|
|
if (!page)
|
2005-04-16 16:20:36 -06:00
|
|
|
continue;
|
2006-01-18 18:42:27 -07:00
|
|
|
/*
|
ksm: memory hotremove migration only
The previous patch enables page migration of ksm pages, but that soon gets
into trouble: not surprising, since we're using the ksm page lock to lock
operations on its stable_node, but page migration switches the page whose
lock is to be used for that. Another layer of locking would fix it, but
do we need that yet?
Do we actually need page migration of ksm pages? Yes, memory hotremove
needs to offline sections of memory: and since we stopped allocating ksm
pages with GFP_HIGHUSER, they will tend to be GFP_HIGHUSER_MOVABLE
candidates for migration.
But KSM is currently unconscious of NUMA issues, happily merging pages
from different NUMA nodes: at present the rule must be, not to use
MADV_MERGEABLE where you care about NUMA. So no, NUMA page migration of
ksm pages does not make sense yet.
So, to complete support for ksm swapping we need to make hotremove safe.
ksm_memory_callback() take ksm_thread_mutex when MEM_GOING_OFFLINE and
release it when MEM_OFFLINE or MEM_CANCEL_OFFLINE. But if mapped pages
are freed before migration reaches them, stable_nodes may be left still
pointing to struct pages which have been removed from the system: the
stable_node needs to identify a page by pfn rather than page pointer, then
it can safely prune them when MEM_OFFLINE.
And make NUMA migration skip PageKsm pages where it skips PageReserved.
But it's only when we reach unmap_and_move() that the page lock is taken
and we can be sure that raised pagecount has prevented a PageAnon from
being upgraded: so add offlining arg to migrate_pages(), to migrate ksm
page when offlining (has sufficient locking) but reject it otherwise.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Izik Eidus <ieidus@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Chris Wright <chrisw@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-14 18:59:33 -07:00
|
|
|
* vm_normal_page() filters out zero pages, but there might
|
|
|
|
* still be PageReserved pages to skip, perhaps in a VDSO.
|
2006-01-18 18:42:27 -07:00
|
|
|
*/
|
2013-02-22 17:35:13 -07:00
|
|
|
if (PageReserved(page))
|
2006-01-12 02:05:20 -07:00
|
|
|
continue;
|
2005-11-28 15:34:23 -07:00
|
|
|
nid = page_to_nid(page);
|
2015-02-11 16:28:03 -07:00
|
|
|
if (node_isset(nid, *qp->nmask) == !!(flags & MPOL_MF_INVERT))
|
2006-01-08 02:01:01 -07:00
|
|
|
continue;
|
2016-07-26 16:26:18 -06:00
|
|
|
if (PageTransCompound(page)) {
|
2016-01-15 17:54:14 -07:00
|
|
|
get_page(page);
|
|
|
|
pte_unmap_unlock(pte, ptl);
|
|
|
|
lock_page(page);
|
|
|
|
ret = split_huge_page(page);
|
|
|
|
unlock_page(page);
|
|
|
|
put_page(page);
|
|
|
|
/* Failed to split -- skip. */
|
|
|
|
if (ret) {
|
|
|
|
pte = pte_offset_map_lock(walk->mm, pmd,
|
|
|
|
addr, &ptl);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
goto retry;
|
|
|
|
}
|
2006-01-08 02:01:01 -07:00
|
|
|
|
mempolicy: do not try to queue pages from !vma_migratable()
Maybe I miss some point, but I don't see a reason why we try to queue
pages from non migratable VMAs.
This testcase steps on VM_BUG_ON_PAGE() in isolate_lru_page():
#include <fcntl.h>
#include <unistd.h>
#include <stdio.h>
#include <sys/mman.h>
#include <numaif.h>
#define SIZE 0x2000
int foo;
int main()
{
int fd;
char *p;
unsigned long mask = 2;
fd = open("/dev/sg0", O_RDWR);
p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
/* Faultin pages */
foo = p[0] + p[0x1000];
mbind(p, SIZE, MPOL_BIND, &mask, 4, MPOL_MF_MOVE | MPOL_MF_STRICT);
return 0;
}
The only case when we can queue pages from such VMA is MPOL_MF_STRICT
plus MPOL_MF_MOVE or MPOL_MF_MOVE_ALL for VMA which has pages on LRU,
but gfp mask is not sutable for migaration (see mapping_gfp_mask() check
in vma_migratable()). That's looks like a bug to me.
Let's filter out non-migratable vma at start of queue_pages_test_walk()
and go to queue_pages_pte_range() only if MPOL_MF_MOVE or
MPOL_MF_MOVE_ALL flag is set.
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-02-05 16:36:33 -07:00
|
|
|
migrate_page_add(page, qp->pagelist, flags);
|
2015-02-11 16:28:03 -07:00
|
|
|
}
|
|
|
|
pte_unmap_unlock(pte - 1, ptl);
|
|
|
|
cond_resched();
|
|
|
|
return 0;
|
2005-06-21 18:15:07 -06:00
|
|
|
}
|
|
|
|
|
2015-02-11 16:28:03 -07:00
|
|
|
static int queue_pages_hugetlb(pte_t *pte, unsigned long hmask,
|
|
|
|
unsigned long addr, unsigned long end,
|
|
|
|
struct mm_walk *walk)
|
2013-09-11 15:22:03 -06:00
|
|
|
{
|
|
|
|
#ifdef CONFIG_HUGETLB_PAGE
|
2015-02-11 16:28:03 -07:00
|
|
|
struct queue_pages *qp = walk->private;
|
|
|
|
unsigned long flags = qp->flags;
|
2013-09-11 15:22:03 -06:00
|
|
|
int nid;
|
|
|
|
struct page *page;
|
2013-11-14 15:31:02 -07:00
|
|
|
spinlock_t *ptl;
|
2014-06-06 08:00:01 -06:00
|
|
|
pte_t entry;
|
2013-09-11 15:22:03 -06:00
|
|
|
|
2015-02-11 16:28:03 -07:00
|
|
|
ptl = huge_pte_lock(hstate_vma(walk->vma), walk->mm, pte);
|
|
|
|
entry = huge_ptep_get(pte);
|
2014-06-06 08:00:01 -06:00
|
|
|
if (!pte_present(entry))
|
|
|
|
goto unlock;
|
|
|
|
page = pte_page(entry);
|
2013-09-11 15:22:03 -06:00
|
|
|
nid = page_to_nid(page);
|
2015-02-11 16:28:03 -07:00
|
|
|
if (node_isset(nid, *qp->nmask) == !!(flags & MPOL_MF_INVERT))
|
2013-09-11 15:22:03 -06:00
|
|
|
goto unlock;
|
|
|
|
/* With MPOL_MF_MOVE, we migrate only unshared hugepage. */
|
|
|
|
if (flags & (MPOL_MF_MOVE_ALL) ||
|
|
|
|
(flags & MPOL_MF_MOVE && page_mapcount(page) == 1))
|
2015-02-11 16:28:03 -07:00
|
|
|
isolate_huge_page(page, qp->pagelist);
|
2013-09-11 15:22:03 -06:00
|
|
|
unlock:
|
2013-11-14 15:31:02 -07:00
|
|
|
spin_unlock(ptl);
|
2013-09-11 15:22:03 -06:00
|
|
|
#else
|
|
|
|
BUG();
|
|
|
|
#endif
|
2005-06-21 18:15:07 -06:00
|
|
|
return 0;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2013-12-05 11:38:22 -07:00
|
|
|
#ifdef CONFIG_NUMA_BALANCING
|
2012-10-25 06:16:32 -06:00
|
|
|
/*
|
2012-10-25 06:16:32 -06:00
|
|
|
* This is used to mark a range of virtual addresses to be inaccessible.
|
|
|
|
* These are later cleared by a NUMA hinting fault. Depending on these
|
|
|
|
* faults, pages may be migrated for better NUMA placement.
|
|
|
|
*
|
|
|
|
* This is assuming that NUMA faults are handled using PROT_NONE. If
|
|
|
|
* an architecture makes a different choice, it will need further
|
|
|
|
* changes to the core.
|
2012-10-25 06:16:32 -06:00
|
|
|
*/
|
2012-10-25 06:16:32 -06:00
|
|
|
unsigned long change_prot_numa(struct vm_area_struct *vma,
|
|
|
|
unsigned long addr, unsigned long end)
|
2012-10-25 06:16:32 -06:00
|
|
|
{
|
2012-10-25 06:16:32 -06:00
|
|
|
int nr_updated;
|
2012-10-25 06:16:32 -06:00
|
|
|
|
2015-02-12 15:58:28 -07:00
|
|
|
nr_updated = change_protection(vma, addr, end, PAGE_NONE, 0, 1);
|
2012-11-02 08:52:48 -06:00
|
|
|
if (nr_updated)
|
|
|
|
count_vm_numa_events(NUMA_PTE_UPDATES, nr_updated);
|
2012-10-25 06:16:32 -06:00
|
|
|
|
2012-10-25 06:16:32 -06:00
|
|
|
return nr_updated;
|
2012-10-25 06:16:32 -06:00
|
|
|
}
|
|
|
|
#else
|
|
|
|
static unsigned long change_prot_numa(struct vm_area_struct *vma,
|
|
|
|
unsigned long addr, unsigned long end)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
2013-12-05 11:38:22 -07:00
|
|
|
#endif /* CONFIG_NUMA_BALANCING */
|
2012-10-25 06:16:32 -06:00
|
|
|
|
2015-02-11 16:28:03 -07:00
|
|
|
static int queue_pages_test_walk(unsigned long start, unsigned long end,
|
|
|
|
struct mm_walk *walk)
|
|
|
|
{
|
|
|
|
struct vm_area_struct *vma = walk->vma;
|
|
|
|
struct queue_pages *qp = walk->private;
|
|
|
|
unsigned long endvma = vma->vm_end;
|
|
|
|
unsigned long flags = qp->flags;
|
|
|
|
|
mempolicy: do not try to queue pages from !vma_migratable()
Maybe I miss some point, but I don't see a reason why we try to queue
pages from non migratable VMAs.
This testcase steps on VM_BUG_ON_PAGE() in isolate_lru_page():
#include <fcntl.h>
#include <unistd.h>
#include <stdio.h>
#include <sys/mman.h>
#include <numaif.h>
#define SIZE 0x2000
int foo;
int main()
{
int fd;
char *p;
unsigned long mask = 2;
fd = open("/dev/sg0", O_RDWR);
p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
/* Faultin pages */
foo = p[0] + p[0x1000];
mbind(p, SIZE, MPOL_BIND, &mask, 4, MPOL_MF_MOVE | MPOL_MF_STRICT);
return 0;
}
The only case when we can queue pages from such VMA is MPOL_MF_STRICT
plus MPOL_MF_MOVE or MPOL_MF_MOVE_ALL for VMA which has pages on LRU,
but gfp mask is not sutable for migaration (see mapping_gfp_mask() check
in vma_migratable()). That's looks like a bug to me.
Let's filter out non-migratable vma at start of queue_pages_test_walk()
and go to queue_pages_pte_range() only if MPOL_MF_MOVE or
MPOL_MF_MOVE_ALL flag is set.
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-02-05 16:36:33 -07:00
|
|
|
if (!vma_migratable(vma))
|
2015-02-11 16:28:06 -07:00
|
|
|
return 1;
|
|
|
|
|
2015-02-11 16:28:03 -07:00
|
|
|
if (endvma > end)
|
|
|
|
endvma = end;
|
|
|
|
if (vma->vm_start > start)
|
|
|
|
start = vma->vm_start;
|
|
|
|
|
|
|
|
if (!(flags & MPOL_MF_DISCONTIG_OK)) {
|
|
|
|
if (!vma->vm_next && vma->vm_end < end)
|
|
|
|
return -EFAULT;
|
|
|
|
if (qp->prev && qp->prev->vm_end < vma->vm_start)
|
|
|
|
return -EFAULT;
|
|
|
|
}
|
|
|
|
|
|
|
|
qp->prev = vma;
|
|
|
|
|
|
|
|
if (flags & MPOL_MF_LAZY) {
|
|
|
|
/* Similar to task_numa_work, skip inaccessible VMAs */
|
2016-03-15 15:56:42 -06:00
|
|
|
if (!is_vm_hugetlb_page(vma) &&
|
|
|
|
(vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE)) &&
|
|
|
|
!(vma->vm_flags & VM_MIXEDMAP))
|
2015-02-11 16:28:03 -07:00
|
|
|
change_prot_numa(vma, start, endvma);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
mempolicy: do not try to queue pages from !vma_migratable()
Maybe I miss some point, but I don't see a reason why we try to queue
pages from non migratable VMAs.
This testcase steps on VM_BUG_ON_PAGE() in isolate_lru_page():
#include <fcntl.h>
#include <unistd.h>
#include <stdio.h>
#include <sys/mman.h>
#include <numaif.h>
#define SIZE 0x2000
int foo;
int main()
{
int fd;
char *p;
unsigned long mask = 2;
fd = open("/dev/sg0", O_RDWR);
p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
/* Faultin pages */
foo = p[0] + p[0x1000];
mbind(p, SIZE, MPOL_BIND, &mask, 4, MPOL_MF_MOVE | MPOL_MF_STRICT);
return 0;
}
The only case when we can queue pages from such VMA is MPOL_MF_STRICT
plus MPOL_MF_MOVE or MPOL_MF_MOVE_ALL for VMA which has pages on LRU,
but gfp mask is not sutable for migaration (see mapping_gfp_mask() check
in vma_migratable()). That's looks like a bug to me.
Let's filter out non-migratable vma at start of queue_pages_test_walk()
and go to queue_pages_pte_range() only if MPOL_MF_MOVE or
MPOL_MF_MOVE_ALL flag is set.
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-02-05 16:36:33 -07:00
|
|
|
/* queue pages from current vma */
|
|
|
|
if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
|
2015-02-11 16:28:03 -07:00
|
|
|
return 0;
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2006-01-08 02:00:50 -07:00
|
|
|
/*
|
2013-09-11 15:22:14 -06:00
|
|
|
* Walk through page tables and collect pages to be migrated.
|
|
|
|
*
|
|
|
|
* If pages found in a given range are on a set of nodes (determined by
|
|
|
|
* @nodes and @flags,) it's isolated and queued to the pagelist which is
|
|
|
|
* passed via @private.)
|
2006-01-08 02:00:50 -07:00
|
|
|
*/
|
2014-06-23 14:22:07 -06:00
|
|
|
static int
|
2013-09-11 15:22:14 -06:00
|
|
|
queue_pages_range(struct mm_struct *mm, unsigned long start, unsigned long end,
|
2015-02-11 16:28:03 -07:00
|
|
|
nodemask_t *nodes, unsigned long flags,
|
|
|
|
struct list_head *pagelist)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2015-02-11 16:28:03 -07:00
|
|
|
struct queue_pages qp = {
|
|
|
|
.pagelist = pagelist,
|
|
|
|
.flags = flags,
|
|
|
|
.nmask = nodes,
|
|
|
|
.prev = NULL,
|
|
|
|
};
|
|
|
|
struct mm_walk queue_pages_walk = {
|
|
|
|
.hugetlb_entry = queue_pages_hugetlb,
|
|
|
|
.pmd_entry = queue_pages_pte_range,
|
|
|
|
.test_walk = queue_pages_test_walk,
|
|
|
|
.mm = mm,
|
|
|
|
.private = &qp,
|
|
|
|
};
|
|
|
|
|
|
|
|
return walk_page_range(start, end, &queue_pages_walk);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
/*
|
|
|
|
* Apply policy to a single VMA
|
|
|
|
* This must be called with the mmap_sem held for writing.
|
|
|
|
*/
|
|
|
|
static int vma_replace_policy(struct vm_area_struct *vma,
|
|
|
|
struct mempolicy *pol)
|
2012-10-08 17:29:14 -06:00
|
|
|
{
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
int err;
|
|
|
|
struct mempolicy *old;
|
|
|
|
struct mempolicy *new;
|
2012-10-08 17:29:14 -06:00
|
|
|
|
|
|
|
pr_debug("vma %lx-%lx/%lx vm_ops %p vm_file %p set_policy %p\n",
|
|
|
|
vma->vm_start, vma->vm_end, vma->vm_pgoff,
|
|
|
|
vma->vm_ops, vma->vm_file,
|
|
|
|
vma->vm_ops ? vma->vm_ops->set_policy : NULL);
|
|
|
|
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
new = mpol_dup(pol);
|
|
|
|
if (IS_ERR(new))
|
|
|
|
return PTR_ERR(new);
|
|
|
|
|
|
|
|
if (vma->vm_ops && vma->vm_ops->set_policy) {
|
2012-10-08 17:29:14 -06:00
|
|
|
err = vma->vm_ops->set_policy(vma, new);
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
if (err)
|
|
|
|
goto err_out;
|
2012-10-08 17:29:14 -06:00
|
|
|
}
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
|
|
|
|
old = vma->vm_policy;
|
|
|
|
vma->vm_policy = new; /* protected by mmap_sem */
|
|
|
|
mpol_put(old);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
err_out:
|
|
|
|
mpol_put(new);
|
2012-10-08 17:29:14 -06:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/* Step 2: apply policy to a range and do splits. */
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
static int mbind_range(struct mm_struct *mm, unsigned long start,
|
|
|
|
unsigned long end, struct mempolicy *new_pol)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct vm_area_struct *next;
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
struct vm_area_struct *prev;
|
|
|
|
struct vm_area_struct *vma;
|
|
|
|
int err = 0;
|
mm/mempolicy.c: refix mbind_range() vma issue
commit 8aacc9f550 ("mm/mempolicy.c: fix pgoff in mbind vma merge") is the
slightly incorrect fix.
Why? Think following case.
1. map 4 pages of a file at offset 0
[0123]
2. map 2 pages just after the first mapping of the same file but with
page offset 2
[0123][23]
3. mbind() 2 pages from the first mapping at offset 2.
mbind_range() should treat new vma is,
[0123][23]
|23|
mbind vma
but it does
[0123][23]
|01|
mbind vma
Oops. then, it makes wrong vma merge and splitting ([01][0123] or similar).
This patch fixes it.
[testcase]
test result - before the patch
case4: 126: test failed. expect '2,4', actual '2,2,2'
case5: passed
case6: passed
case7: passed
case8: passed
case_n: 246: test failed. expect '4,2', actual '1,4'
------------[ cut here ]------------
kernel BUG at mm/filemap.c:135!
invalid opcode: 0000 [#4] SMP DEBUG_PAGEALLOC
(snip long bug on messages)
test result - after the patch
case4: passed
case5: passed
case6: passed
case7: passed
case8: passed
case_n: passed
source: mbind_vma_test.c
============================================================
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
void* mmap_addr;
struct bitmask *nmask;
char buf[1024];
FILE *file;
char retbuf[10240] = "";
int mapped_fd;
char *rubysrc = "ruby -e '\
pid = %d; \
vstart = 0x%llx; \
vend = 0x%llx; \
s = `pmap -q #{pid}`; \
rary = []; \
s.each_line {|line|; \
ary=line.split(\" \"); \
addr = ary[0].to_i(16); \
if(vstart <= addr && addr < vend) then \
rary.push(ary[1].to_i()/4); \
end; \
}; \
print rary.join(\",\"); \
'";
void init(void)
{
void* addr;
char buf[128];
nmask = numa_allocate_nodemask();
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
sprintf(buf, "%s", "mbind_vma_XXXXXX");
mapped_fd = mkstemp(buf);
if (mapped_fd == -1)
perror("mkstemp "), exit(1);
unlink(buf);
if (lseek(mapped_fd, pagesize*8, SEEK_SET) < 0)
perror("lseek "), exit(1);
if (write(mapped_fd, "\0", 1) < 0)
perror("write "), exit(1);
addr = mmap(NULL, pagesize*8, PROT_NONE,
MAP_SHARED, mapped_fd, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
if (mprotect(addr+pagesize, pagesize*6, PROT_READ|PROT_WRITE) < 0)
perror("mprotect "), exit(1);
mmap_addr = addr + pagesize;
/* make page populate */
memset(mmap_addr, 0, pagesize*6);
}
void fin(void)
{
void* addr = mmap_addr - pagesize;
munmap(addr, pagesize*8);
memset(buf, 0, sizeof(buf));
memset(retbuf, 0, sizeof(retbuf));
}
void mem_bind(int index, int len)
{
int err;
err = mbind(mmap_addr+pagesize*index, pagesize*len,
MPOL_BIND, nmask->maskp, nmask->size, 0);
if (err)
perror("mbind "), exit(err);
}
void mem_interleave(int index, int len)
{
int err;
err = mbind(mmap_addr+pagesize*index, pagesize*len,
MPOL_INTERLEAVE, nmask->maskp, nmask->size, 0);
if (err)
perror("mbind "), exit(err);
}
void mem_unbind(int index, int len)
{
int err;
err = mbind(mmap_addr+pagesize*index, pagesize*len,
MPOL_DEFAULT, NULL, 0, 0);
if (err)
perror("mbind "), exit(err);
}
void Assert(char *expected, char *value, char *name, int line)
{
if (strcmp(expected, value) == 0) {
fprintf(stderr, "%s: passed\n", name);
return;
}
else {
fprintf(stderr, "%s: %d: test failed. expect '%s', actual '%s'\n",
name, line,
expected, value);
// exit(1);
}
}
/*
AAAA
PPPPPPNNNNNN
might become
PPNNNNNNNNNN
case 4 below
*/
void case4(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 4);
mem_unbind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("2,4", retbuf, "case4", __LINE__);
fin();
}
/*
AAAA
PPPPPPNNNNNN
might become
PPPPPPPPPPNN
case 5 below
*/
void case5(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_bind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case5", __LINE__);
fin();
}
/*
AAAA
PPPPNNNNXXXX
might become
PPPPPPPPPPPP 6
*/
void case6(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_bind(4, 2);
mem_bind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("6", retbuf, "case6", __LINE__);
fin();
}
/*
AAAA
PPPPNNNNXXXX
might become
PPPPPPPPXXXX 7
*/
void case7(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_interleave(4, 2);
mem_bind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case7", __LINE__);
fin();
}
/*
AAAA
PPPPNNNNXXXX
might become
PPPPNNNNNNNN 8
*/
void case8(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_interleave(4, 2);
mem_interleave(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("2,4", retbuf, "case8", __LINE__);
fin();
}
void case_n(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
/* make redundunt mappings [0][1234][34][7] */
mmap(mmap_addr + pagesize*4, pagesize*2, PROT_READ|PROT_WRITE,
MAP_FIXED|MAP_SHARED, mapped_fd, pagesize*3);
/* Expect to do nothing. */
mem_unbind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case_n", __LINE__);
fin();
}
int main(int argc, char** argv)
{
case4();
case5();
case6();
case7();
case8();
case_n();
return 0;
}
=============================================================
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Caspar Zhang <caspar@casparzhang.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: <stable@vger.kernel.org> [3.1.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-12-28 16:57:11 -07:00
|
|
|
pgoff_t pgoff;
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
unsigned long vmstart;
|
|
|
|
unsigned long vmend;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2012-03-06 19:23:36 -07:00
|
|
|
vma = find_vma(mm, start);
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
if (!vma || vma->vm_start > start)
|
|
|
|
return -EFAULT;
|
|
|
|
|
2012-03-06 19:23:36 -07:00
|
|
|
prev = vma->vm_prev;
|
mm/mempolicy.c: refix mbind_range() vma issue
commit 8aacc9f550 ("mm/mempolicy.c: fix pgoff in mbind vma merge") is the
slightly incorrect fix.
Why? Think following case.
1. map 4 pages of a file at offset 0
[0123]
2. map 2 pages just after the first mapping of the same file but with
page offset 2
[0123][23]
3. mbind() 2 pages from the first mapping at offset 2.
mbind_range() should treat new vma is,
[0123][23]
|23|
mbind vma
but it does
[0123][23]
|01|
mbind vma
Oops. then, it makes wrong vma merge and splitting ([01][0123] or similar).
This patch fixes it.
[testcase]
test result - before the patch
case4: 126: test failed. expect '2,4', actual '2,2,2'
case5: passed
case6: passed
case7: passed
case8: passed
case_n: 246: test failed. expect '4,2', actual '1,4'
------------[ cut here ]------------
kernel BUG at mm/filemap.c:135!
invalid opcode: 0000 [#4] SMP DEBUG_PAGEALLOC
(snip long bug on messages)
test result - after the patch
case4: passed
case5: passed
case6: passed
case7: passed
case8: passed
case_n: passed
source: mbind_vma_test.c
============================================================
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
void* mmap_addr;
struct bitmask *nmask;
char buf[1024];
FILE *file;
char retbuf[10240] = "";
int mapped_fd;
char *rubysrc = "ruby -e '\
pid = %d; \
vstart = 0x%llx; \
vend = 0x%llx; \
s = `pmap -q #{pid}`; \
rary = []; \
s.each_line {|line|; \
ary=line.split(\" \"); \
addr = ary[0].to_i(16); \
if(vstart <= addr && addr < vend) then \
rary.push(ary[1].to_i()/4); \
end; \
}; \
print rary.join(\",\"); \
'";
void init(void)
{
void* addr;
char buf[128];
nmask = numa_allocate_nodemask();
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
sprintf(buf, "%s", "mbind_vma_XXXXXX");
mapped_fd = mkstemp(buf);
if (mapped_fd == -1)
perror("mkstemp "), exit(1);
unlink(buf);
if (lseek(mapped_fd, pagesize*8, SEEK_SET) < 0)
perror("lseek "), exit(1);
if (write(mapped_fd, "\0", 1) < 0)
perror("write "), exit(1);
addr = mmap(NULL, pagesize*8, PROT_NONE,
MAP_SHARED, mapped_fd, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
if (mprotect(addr+pagesize, pagesize*6, PROT_READ|PROT_WRITE) < 0)
perror("mprotect "), exit(1);
mmap_addr = addr + pagesize;
/* make page populate */
memset(mmap_addr, 0, pagesize*6);
}
void fin(void)
{
void* addr = mmap_addr - pagesize;
munmap(addr, pagesize*8);
memset(buf, 0, sizeof(buf));
memset(retbuf, 0, sizeof(retbuf));
}
void mem_bind(int index, int len)
{
int err;
err = mbind(mmap_addr+pagesize*index, pagesize*len,
MPOL_BIND, nmask->maskp, nmask->size, 0);
if (err)
perror("mbind "), exit(err);
}
void mem_interleave(int index, int len)
{
int err;
err = mbind(mmap_addr+pagesize*index, pagesize*len,
MPOL_INTERLEAVE, nmask->maskp, nmask->size, 0);
if (err)
perror("mbind "), exit(err);
}
void mem_unbind(int index, int len)
{
int err;
err = mbind(mmap_addr+pagesize*index, pagesize*len,
MPOL_DEFAULT, NULL, 0, 0);
if (err)
perror("mbind "), exit(err);
}
void Assert(char *expected, char *value, char *name, int line)
{
if (strcmp(expected, value) == 0) {
fprintf(stderr, "%s: passed\n", name);
return;
}
else {
fprintf(stderr, "%s: %d: test failed. expect '%s', actual '%s'\n",
name, line,
expected, value);
// exit(1);
}
}
/*
AAAA
PPPPPPNNNNNN
might become
PPNNNNNNNNNN
case 4 below
*/
void case4(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 4);
mem_unbind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("2,4", retbuf, "case4", __LINE__);
fin();
}
/*
AAAA
PPPPPPNNNNNN
might become
PPPPPPPPPPNN
case 5 below
*/
void case5(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_bind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case5", __LINE__);
fin();
}
/*
AAAA
PPPPNNNNXXXX
might become
PPPPPPPPPPPP 6
*/
void case6(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_bind(4, 2);
mem_bind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("6", retbuf, "case6", __LINE__);
fin();
}
/*
AAAA
PPPPNNNNXXXX
might become
PPPPPPPPXXXX 7
*/
void case7(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_interleave(4, 2);
mem_bind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case7", __LINE__);
fin();
}
/*
AAAA
PPPPNNNNXXXX
might become
PPPPNNNNNNNN 8
*/
void case8(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_interleave(4, 2);
mem_interleave(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("2,4", retbuf, "case8", __LINE__);
fin();
}
void case_n(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
/* make redundunt mappings [0][1234][34][7] */
mmap(mmap_addr + pagesize*4, pagesize*2, PROT_READ|PROT_WRITE,
MAP_FIXED|MAP_SHARED, mapped_fd, pagesize*3);
/* Expect to do nothing. */
mem_unbind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case_n", __LINE__);
fin();
}
int main(int argc, char** argv)
{
case4();
case5();
case6();
case7();
case8();
case_n();
return 0;
}
=============================================================
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Caspar Zhang <caspar@casparzhang.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: <stable@vger.kernel.org> [3.1.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-12-28 16:57:11 -07:00
|
|
|
if (start > vma->vm_start)
|
|
|
|
prev = vma;
|
|
|
|
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
for (; vma && vma->vm_start < end; prev = vma, vma = next) {
|
2005-04-16 16:20:36 -06:00
|
|
|
next = vma->vm_next;
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
vmstart = max(start, vma->vm_start);
|
|
|
|
vmend = min(end, vma->vm_end);
|
|
|
|
|
mm/mempolicy.c: refix mbind_range() vma issue
commit 8aacc9f550 ("mm/mempolicy.c: fix pgoff in mbind vma merge") is the
slightly incorrect fix.
Why? Think following case.
1. map 4 pages of a file at offset 0
[0123]
2. map 2 pages just after the first mapping of the same file but with
page offset 2
[0123][23]
3. mbind() 2 pages from the first mapping at offset 2.
mbind_range() should treat new vma is,
[0123][23]
|23|
mbind vma
but it does
[0123][23]
|01|
mbind vma
Oops. then, it makes wrong vma merge and splitting ([01][0123] or similar).
This patch fixes it.
[testcase]
test result - before the patch
case4: 126: test failed. expect '2,4', actual '2,2,2'
case5: passed
case6: passed
case7: passed
case8: passed
case_n: 246: test failed. expect '4,2', actual '1,4'
------------[ cut here ]------------
kernel BUG at mm/filemap.c:135!
invalid opcode: 0000 [#4] SMP DEBUG_PAGEALLOC
(snip long bug on messages)
test result - after the patch
case4: passed
case5: passed
case6: passed
case7: passed
case8: passed
case_n: passed
source: mbind_vma_test.c
============================================================
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
void* mmap_addr;
struct bitmask *nmask;
char buf[1024];
FILE *file;
char retbuf[10240] = "";
int mapped_fd;
char *rubysrc = "ruby -e '\
pid = %d; \
vstart = 0x%llx; \
vend = 0x%llx; \
s = `pmap -q #{pid}`; \
rary = []; \
s.each_line {|line|; \
ary=line.split(\" \"); \
addr = ary[0].to_i(16); \
if(vstart <= addr && addr < vend) then \
rary.push(ary[1].to_i()/4); \
end; \
}; \
print rary.join(\",\"); \
'";
void init(void)
{
void* addr;
char buf[128];
nmask = numa_allocate_nodemask();
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
sprintf(buf, "%s", "mbind_vma_XXXXXX");
mapped_fd = mkstemp(buf);
if (mapped_fd == -1)
perror("mkstemp "), exit(1);
unlink(buf);
if (lseek(mapped_fd, pagesize*8, SEEK_SET) < 0)
perror("lseek "), exit(1);
if (write(mapped_fd, "\0", 1) < 0)
perror("write "), exit(1);
addr = mmap(NULL, pagesize*8, PROT_NONE,
MAP_SHARED, mapped_fd, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
if (mprotect(addr+pagesize, pagesize*6, PROT_READ|PROT_WRITE) < 0)
perror("mprotect "), exit(1);
mmap_addr = addr + pagesize;
/* make page populate */
memset(mmap_addr, 0, pagesize*6);
}
void fin(void)
{
void* addr = mmap_addr - pagesize;
munmap(addr, pagesize*8);
memset(buf, 0, sizeof(buf));
memset(retbuf, 0, sizeof(retbuf));
}
void mem_bind(int index, int len)
{
int err;
err = mbind(mmap_addr+pagesize*index, pagesize*len,
MPOL_BIND, nmask->maskp, nmask->size, 0);
if (err)
perror("mbind "), exit(err);
}
void mem_interleave(int index, int len)
{
int err;
err = mbind(mmap_addr+pagesize*index, pagesize*len,
MPOL_INTERLEAVE, nmask->maskp, nmask->size, 0);
if (err)
perror("mbind "), exit(err);
}
void mem_unbind(int index, int len)
{
int err;
err = mbind(mmap_addr+pagesize*index, pagesize*len,
MPOL_DEFAULT, NULL, 0, 0);
if (err)
perror("mbind "), exit(err);
}
void Assert(char *expected, char *value, char *name, int line)
{
if (strcmp(expected, value) == 0) {
fprintf(stderr, "%s: passed\n", name);
return;
}
else {
fprintf(stderr, "%s: %d: test failed. expect '%s', actual '%s'\n",
name, line,
expected, value);
// exit(1);
}
}
/*
AAAA
PPPPPPNNNNNN
might become
PPNNNNNNNNNN
case 4 below
*/
void case4(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 4);
mem_unbind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("2,4", retbuf, "case4", __LINE__);
fin();
}
/*
AAAA
PPPPPPNNNNNN
might become
PPPPPPPPPPNN
case 5 below
*/
void case5(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_bind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case5", __LINE__);
fin();
}
/*
AAAA
PPPPNNNNXXXX
might become
PPPPPPPPPPPP 6
*/
void case6(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_bind(4, 2);
mem_bind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("6", retbuf, "case6", __LINE__);
fin();
}
/*
AAAA
PPPPNNNNXXXX
might become
PPPPPPPPXXXX 7
*/
void case7(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_interleave(4, 2);
mem_bind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case7", __LINE__);
fin();
}
/*
AAAA
PPPPNNNNXXXX
might become
PPPPNNNNNNNN 8
*/
void case8(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
mem_bind(0, 2);
mem_interleave(4, 2);
mem_interleave(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("2,4", retbuf, "case8", __LINE__);
fin();
}
void case_n(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);
/* make redundunt mappings [0][1234][34][7] */
mmap(mmap_addr + pagesize*4, pagesize*2, PROT_READ|PROT_WRITE,
MAP_FIXED|MAP_SHARED, mapped_fd, pagesize*3);
/* Expect to do nothing. */
mem_unbind(2, 2);
file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case_n", __LINE__);
fin();
}
int main(int argc, char** argv)
{
case4();
case5();
case6();
case7();
case8();
case_n();
return 0;
}
=============================================================
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Caspar Zhang <caspar@casparzhang.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: <stable@vger.kernel.org> [3.1.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-12-28 16:57:11 -07:00
|
|
|
if (mpol_equal(vma_policy(vma), new_pol))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
pgoff = vma->vm_pgoff +
|
|
|
|
((vmstart - vma->vm_start) >> PAGE_SHIFT);
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
prev = vma_merge(mm, prev, vmstart, vmend, vma->vm_flags,
|
2015-09-04 16:46:24 -06:00
|
|
|
vma->anon_vma, vma->vm_file, pgoff,
|
|
|
|
new_pol, vma->vm_userfaultfd_ctx);
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
if (prev) {
|
|
|
|
vma = prev;
|
|
|
|
next = vma->vm_next;
|
2013-07-31 14:53:28 -06:00
|
|
|
if (mpol_equal(vma_policy(vma), new_pol))
|
|
|
|
continue;
|
|
|
|
/* vma_merge() joined vma && vma->next, case 8 */
|
|
|
|
goto replace;
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
}
|
|
|
|
if (vma->vm_start != vmstart) {
|
|
|
|
err = split_vma(vma->vm_mm, vma, vmstart, 1);
|
|
|
|
if (err)
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
if (vma->vm_end != vmend) {
|
|
|
|
err = split_vma(vma->vm_mm, vma, vmend, 0);
|
|
|
|
if (err)
|
|
|
|
goto out;
|
|
|
|
}
|
2013-07-31 14:53:28 -06:00
|
|
|
replace:
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
err = vma_replace_policy(vma, new_pol);
|
2012-10-08 17:29:14 -06:00
|
|
|
if (err)
|
|
|
|
goto out;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
|
|
|
|
out:
|
2005-04-16 16:20:36 -06:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Set the process memory policy */
|
2008-04-28 03:12:25 -06:00
|
|
|
static long do_set_mempolicy(unsigned short mode, unsigned short flags,
|
|
|
|
nodemask_t *nodes)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
struct mempolicy *new, *old;
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
NODEMASK_SCRATCH(scratch);
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
int ret;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
if (!scratch)
|
|
|
|
return -ENOMEM;
|
2008-04-28 03:13:10 -06:00
|
|
|
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
new = mpol_new(mode, flags, nodes);
|
|
|
|
if (IS_ERR(new)) {
|
|
|
|
ret = PTR_ERR(new);
|
|
|
|
goto out;
|
|
|
|
}
|
2014-10-09 16:27:55 -06:00
|
|
|
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
task_lock(current);
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
ret = mpol_set_nodemask(new, nodes, scratch);
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
if (ret) {
|
|
|
|
task_unlock(current);
|
|
|
|
mpol_put(new);
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
goto out;
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
}
|
|
|
|
old = current->mempolicy;
|
2005-04-16 16:20:36 -06:00
|
|
|
current->mempolicy = new;
|
2008-04-28 03:13:12 -06:00
|
|
|
if (new && new->mode == MPOL_INTERLEAVE &&
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
nodes_weight(new->v.nodes))
|
2005-10-29 19:15:48 -06:00
|
|
|
current->il_next = first_node(new->v.nodes);
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
task_unlock(current);
|
|
|
|
mpol_put(old);
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
ret = 0;
|
|
|
|
out:
|
|
|
|
NODEMASK_SCRATCH_FREE(scratch);
|
|
|
|
return ret;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
/*
|
|
|
|
* Return nodemask for policy for get_mempolicy() query
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
*
|
|
|
|
* Called with task's alloc_lock held
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
*/
|
|
|
|
static void get_policy_nodemask(struct mempolicy *p, nodemask_t *nodes)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2005-10-29 19:15:48 -06:00
|
|
|
nodes_clear(*nodes);
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
if (p == &default_policy)
|
|
|
|
return;
|
|
|
|
|
2008-04-28 03:13:12 -06:00
|
|
|
switch (p->mode) {
|
2008-04-28 03:12:18 -06:00
|
|
|
case MPOL_BIND:
|
|
|
|
/* Fall through */
|
2005-04-16 16:20:36 -06:00
|
|
|
case MPOL_INTERLEAVE:
|
2005-10-29 19:15:48 -06:00
|
|
|
*nodes = p->v.nodes;
|
2005-04-16 16:20:36 -06:00
|
|
|
break;
|
|
|
|
case MPOL_PREFERRED:
|
2008-04-28 03:13:21 -06:00
|
|
|
if (!(p->flags & MPOL_F_LOCAL))
|
2005-10-29 19:15:48 -06:00
|
|
|
node_set(p->v.preferred_node, *nodes);
|
mempolicy: mPOL_PREFERRED cleanups for "local allocation"
Here are a couple of "cleanups" for MPOL_PREFERRED behavior when
v.preferred_node < 0 -- i.e., "local allocation":
1) [do_]get_mempolicy() calls the now renamed get_policy_nodemask()
to fetch the nodemask associated with a policy. Currently,
get_policy_nodemask() returns the set of nodes with memory, when
the policy 'mode' is 'PREFERRED, and the preferred_node is < 0.
Change to return an empty nodemask, as this is what was specified
to achieve "local allocation".
2) When a task is moved into a [new] cpuset, mpol_rebind_policy() is
called to adjust any task and vma policy nodes to be valid in the
new cpuset. However, when the policy is MPOL_PREFERRED, and the
preferred_node is <0, no rebind is necessary. The "local allocation"
indication is valid in any cpuset. Existing code will "do the right
thing" because node_remap() will just return the argument node when
it is outside of the valid range of node ids. However, I think it is
clearer and cleaner to skip the remap explicitly in this case.
3) mpol_to_str() produces a printable, "human readable" string from a
struct mempolicy. For MPOL_PREFERRED with preferred_node <0, show
"local", as this indicates local allocation, as the task migrates
among nodes. Note that this matches the usage of "local allocation"
in libnuma() and numactl. Without this change, I believe that node_set()
[via set_bit()] will set bit 31, resulting in a misleading display.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:20 -06:00
|
|
|
/* else return empty node mask for local allocation */
|
2005-04-16 16:20:36 -06:00
|
|
|
break;
|
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-02-12 14:01:56 -07:00
|
|
|
static int lookup_node(unsigned long addr)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct page *p;
|
|
|
|
int err;
|
|
|
|
|
2016-02-12 14:01:56 -07:00
|
|
|
err = get_user_pages(addr & PAGE_MASK, 1, 0, 0, &p, NULL);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (err >= 0) {
|
|
|
|
err = page_to_nid(p);
|
|
|
|
put_page(p);
|
|
|
|
}
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Retrieve NUMA policy */
|
2007-10-16 02:26:26 -06:00
|
|
|
static long do_get_mempolicy(int *policy, nodemask_t *nmask,
|
|
|
|
unsigned long addr, unsigned long flags)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2005-10-29 19:16:59 -06:00
|
|
|
int err;
|
2005-04-16 16:20:36 -06:00
|
|
|
struct mm_struct *mm = current->mm;
|
|
|
|
struct vm_area_struct *vma = NULL;
|
|
|
|
struct mempolicy *pol = current->mempolicy;
|
|
|
|
|
2007-10-16 02:24:51 -06:00
|
|
|
if (flags &
|
|
|
|
~(unsigned long)(MPOL_F_NODE|MPOL_F_ADDR|MPOL_F_MEMS_ALLOWED))
|
2005-04-16 16:20:36 -06:00
|
|
|
return -EINVAL;
|
2007-10-16 02:24:51 -06:00
|
|
|
|
|
|
|
if (flags & MPOL_F_MEMS_ALLOWED) {
|
|
|
|
if (flags & (MPOL_F_NODE|MPOL_F_ADDR))
|
|
|
|
return -EINVAL;
|
|
|
|
*policy = 0; /* just so it's initialized */
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
task_lock(current);
|
2007-10-16 02:24:51 -06:00
|
|
|
*nmask = cpuset_current_mems_allowed;
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
task_unlock(current);
|
2007-10-16 02:24:51 -06:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
if (flags & MPOL_F_ADDR) {
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
/*
|
|
|
|
* Do NOT fall back to task policy if the
|
|
|
|
* vma/shared policy at addr is NULL. We
|
|
|
|
* want to return MPOL_DEFAULT in this case.
|
|
|
|
*/
|
2005-04-16 16:20:36 -06:00
|
|
|
down_read(&mm->mmap_sem);
|
|
|
|
vma = find_vma_intersection(mm, addr, addr+1);
|
|
|
|
if (!vma) {
|
|
|
|
up_read(&mm->mmap_sem);
|
|
|
|
return -EFAULT;
|
|
|
|
}
|
|
|
|
if (vma->vm_ops && vma->vm_ops->get_policy)
|
|
|
|
pol = vma->vm_ops->get_policy(vma, addr);
|
|
|
|
else
|
|
|
|
pol = vma->vm_policy;
|
|
|
|
} else if (addr)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
if (!pol)
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
pol = &default_policy; /* indicates default behavior */
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (flags & MPOL_F_NODE) {
|
|
|
|
if (flags & MPOL_F_ADDR) {
|
2016-02-12 14:01:56 -07:00
|
|
|
err = lookup_node(addr);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (err < 0)
|
|
|
|
goto out;
|
2005-10-29 19:16:59 -06:00
|
|
|
*policy = err;
|
2005-04-16 16:20:36 -06:00
|
|
|
} else if (pol == current->mempolicy &&
|
2008-04-28 03:13:12 -06:00
|
|
|
pol->mode == MPOL_INTERLEAVE) {
|
2005-10-29 19:16:59 -06:00
|
|
|
*policy = current->il_next;
|
2005-04-16 16:20:36 -06:00
|
|
|
} else {
|
|
|
|
err = -EINVAL;
|
|
|
|
goto out;
|
|
|
|
}
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
} else {
|
|
|
|
*policy = pol == &default_policy ? MPOL_DEFAULT :
|
|
|
|
pol->mode;
|
2008-07-04 13:24:13 -06:00
|
|
|
/*
|
|
|
|
* Internal mempolicy flags must be masked off before exposing
|
|
|
|
* the policy to userspace.
|
|
|
|
*/
|
|
|
|
*policy |= (pol->flags & MPOL_MODE_FLAGS);
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (vma) {
|
|
|
|
up_read(¤t->mm->mmap_sem);
|
|
|
|
vma = NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
err = 0;
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
if (nmask) {
|
2010-03-23 14:35:41 -06:00
|
|
|
if (mpol_store_user_nodemask(pol)) {
|
|
|
|
*nmask = pol->w.user_nodemask;
|
|
|
|
} else {
|
|
|
|
task_lock(current);
|
|
|
|
get_policy_nodemask(pol, nmask);
|
|
|
|
task_unlock(current);
|
|
|
|
}
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
out:
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
mpol_cond_put(pol);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (vma)
|
|
|
|
up_read(¤t->mm->mmap_sem);
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2006-03-22 01:09:12 -07:00
|
|
|
#ifdef CONFIG_MIGRATION
|
2006-01-08 02:01:04 -07:00
|
|
|
/*
|
|
|
|
* page migration
|
|
|
|
*/
|
2006-01-18 18:42:29 -07:00
|
|
|
static void migrate_page_add(struct page *page, struct list_head *pagelist,
|
|
|
|
unsigned long flags)
|
2006-01-08 02:01:04 -07:00
|
|
|
{
|
|
|
|
/*
|
2006-01-18 18:42:29 -07:00
|
|
|
* Avoid migrating a page that is shared with others.
|
2006-01-08 02:01:04 -07:00
|
|
|
*/
|
vmscan: move isolate_lru_page() to vmscan.c
On large memory systems, the VM can spend way too much time scanning
through pages that it cannot (or should not) evict from memory. Not only
does it use up CPU time, but it also provokes lock contention and can
leave large systems under memory presure in a catatonic state.
This patch series improves VM scalability by:
1) putting filesystem backed, swap backed and unevictable pages
onto their own LRUs, so the system only scans the pages that it
can/should evict from memory
2) switching to two handed clock replacement for the anonymous LRUs,
so the number of pages that need to be scanned when the system
starts swapping is bound to a reasonable number
3) keeping unevictable pages off the LRU completely, so the
VM does not waste CPU time scanning them. ramfs, ramdisk,
SHM_LOCKED shared memory segments and mlock()ed VMA pages
are keept on the unevictable list.
This patch:
isolate_lru_page logically belongs to be in vmscan.c than migrate.c.
It is tough, because we don't need that function without memory migration
so there is a valid argument to have it in migrate.c. However a
subsequent patch needs to make use of it in the core mm, so we can happily
move it to vmscan.c.
Also, make the function a little more generic by not requiring that it
adds an isolated page to a given list. Callers can do that.
Note that we now have '__isolate_lru_page()', that does
something quite different, visible outside of vmscan.c
for use with memory controller. Methinks we need to
rationalize these names/purposes. --lts
[akpm@linux-foundation.org: fix mm/memory_hotplug.c build]
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-10-18 21:26:09 -06:00
|
|
|
if ((flags & MPOL_MF_MOVE_ALL) || page_mapcount(page) == 1) {
|
|
|
|
if (!isolate_lru_page(page)) {
|
|
|
|
list_add_tail(&page->lru, pagelist);
|
2016-07-28 16:45:31 -06:00
|
|
|
inc_node_page_state(page, NR_ISOLATED_ANON +
|
2009-12-14 18:58:11 -07:00
|
|
|
page_is_file_cache(page));
|
vmscan: move isolate_lru_page() to vmscan.c
On large memory systems, the VM can spend way too much time scanning
through pages that it cannot (or should not) evict from memory. Not only
does it use up CPU time, but it also provokes lock contention and can
leave large systems under memory presure in a catatonic state.
This patch series improves VM scalability by:
1) putting filesystem backed, swap backed and unevictable pages
onto their own LRUs, so the system only scans the pages that it
can/should evict from memory
2) switching to two handed clock replacement for the anonymous LRUs,
so the number of pages that need to be scanned when the system
starts swapping is bound to a reasonable number
3) keeping unevictable pages off the LRU completely, so the
VM does not waste CPU time scanning them. ramfs, ramdisk,
SHM_LOCKED shared memory segments and mlock()ed VMA pages
are keept on the unevictable list.
This patch:
isolate_lru_page logically belongs to be in vmscan.c than migrate.c.
It is tough, because we don't need that function without memory migration
so there is a valid argument to have it in migrate.c. However a
subsequent patch needs to make use of it in the core mm, so we can happily
move it to vmscan.c.
Also, make the function a little more generic by not requiring that it
adds an isolated page to a given list. Callers can do that.
Note that we now have '__isolate_lru_page()', that does
something quite different, visible outside of vmscan.c
for use with memory controller. Methinks we need to
rationalize these names/purposes. --lts
[akpm@linux-foundation.org: fix mm/memory_hotplug.c build]
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-10-18 21:26:09 -06:00
|
|
|
}
|
|
|
|
}
|
2006-02-01 04:05:40 -07:00
|
|
|
}
|
2006-01-08 02:01:04 -07:00
|
|
|
|
2006-06-23 03:03:55 -06:00
|
|
|
static struct page *new_node_page(struct page *page, unsigned long node, int **x)
|
2006-06-23 03:03:53 -06:00
|
|
|
{
|
2013-09-11 15:22:03 -06:00
|
|
|
if (PageHuge(page))
|
|
|
|
return alloc_huge_page_node(page_hstate(compound_head(page)),
|
|
|
|
node);
|
|
|
|
else
|
mm: rename alloc_pages_exact_node() to __alloc_pages_node()
alloc_pages_exact_node() was introduced in commit 6484eb3e2a81 ("page
allocator: do not check NUMA node ID when the caller knows the node is
valid") as an optimized variant of alloc_pages_node(), that doesn't
fallback to current node for nid == NUMA_NO_NODE. Unfortunately the
name of the function can easily suggest that the allocation is
restricted to the given node and fails otherwise. In truth, the node is
only preferred, unless __GFP_THISNODE is passed among the gfp flags.
The misleading name has lead to mistakes in the past, see for example
commits 5265047ac301 ("mm, thp: really limit transparent hugepage
allocation to local node") and b360edb43f8e ("mm, mempolicy:
migrate_to_node should only migrate to node").
Another issue with the name is that there's a family of
alloc_pages_exact*() functions where 'exact' means exact size (instead
of page order), which leads to more confusion.
To prevent further mistakes, this patch effectively renames
alloc_pages_exact_node() to __alloc_pages_node() to better convey that
it's an optimized variant of alloc_pages_node() not intended for general
usage. Both functions get described in comments.
It has been also considered to really provide a convenience function for
allocations restricted to a node, but the major opinion seems to be that
__GFP_THISNODE already provides that functionality and we shouldn't
duplicate the API needlessly. The number of users would be small
anyway.
Existing callers of alloc_pages_exact_node() are simply converted to
call __alloc_pages_node(), with the exception of sba_alloc_coherent()
which open-codes the check for NUMA_NO_NODE, so it is converted to use
alloc_pages_node() instead. This means it no longer performs some
VM_BUG_ON checks, and since the current check for nid in
alloc_pages_node() uses a 'nid < 0' comparison (which includes
NUMA_NO_NODE), it may hide wrong values which would be previously
exposed.
Both differences will be rectified by the next patch.
To sum up, this patch makes no functional changes, except temporarily
hiding potentially buggy callers. Restricting the checks in
alloc_pages_node() is left for the next patch which can in turn expose
more existing buggy callers.
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Robin Holt <robinmholt@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Christoph Lameter <cl@linux.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Cc: Mel Gorman <mgorman@suse.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Cliff Whickman <cpw@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 16:03:50 -06:00
|
|
|
return __alloc_pages_node(node, GFP_HIGHUSER_MOVABLE |
|
2015-04-14 16:46:52 -06:00
|
|
|
__GFP_THISNODE, 0);
|
2006-06-23 03:03:53 -06:00
|
|
|
}
|
|
|
|
|
2006-02-01 04:05:40 -07:00
|
|
|
/*
|
|
|
|
* Migrate pages from one node to a target node.
|
|
|
|
* Returns error or the number of pages not migrated.
|
|
|
|
*/
|
2007-10-16 02:26:26 -06:00
|
|
|
static int migrate_to_node(struct mm_struct *mm, int source, int dest,
|
|
|
|
int flags)
|
2006-02-01 04:05:40 -07:00
|
|
|
{
|
|
|
|
nodemask_t nmask;
|
|
|
|
LIST_HEAD(pagelist);
|
|
|
|
int err = 0;
|
|
|
|
|
|
|
|
nodes_clear(nmask);
|
|
|
|
node_set(source, nmask);
|
2006-01-08 02:01:04 -07:00
|
|
|
|
2012-10-08 17:33:38 -06:00
|
|
|
/*
|
|
|
|
* This does not "check" the range but isolates all pages that
|
|
|
|
* need migration. Between passing in the full user address
|
|
|
|
* space range and MPOL_MF_DISCONTIG_OK, this call can not fail.
|
|
|
|
*/
|
|
|
|
VM_BUG_ON(!(flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)));
|
2013-09-11 15:22:14 -06:00
|
|
|
queue_pages_range(mm, mm->mmap->vm_start, mm->task_size, &nmask,
|
2006-02-01 04:05:40 -07:00
|
|
|
flags | MPOL_MF_DISCONTIG_OK, &pagelist);
|
|
|
|
|
2010-10-26 15:21:29 -06:00
|
|
|
if (!list_empty(&pagelist)) {
|
2014-06-04 17:08:25 -06:00
|
|
|
err = migrate_pages(&pagelist, new_node_page, NULL, dest,
|
2013-02-22 17:35:14 -07:00
|
|
|
MIGRATE_SYNC, MR_SYSCALL);
|
2010-10-26 15:21:29 -06:00
|
|
|
if (err)
|
2013-09-11 15:22:03 -06:00
|
|
|
putback_movable_pages(&pagelist);
|
2010-10-26 15:21:29 -06:00
|
|
|
}
|
2006-06-23 03:03:53 -06:00
|
|
|
|
2006-02-01 04:05:40 -07:00
|
|
|
return err;
|
2006-01-08 02:01:04 -07:00
|
|
|
}
|
|
|
|
|
2006-01-08 02:00:51 -07:00
|
|
|
/*
|
2006-02-01 04:05:40 -07:00
|
|
|
* Move pages between the two nodesets so as to preserve the physical
|
|
|
|
* layout as much as possible.
|
2006-01-08 02:00:51 -07:00
|
|
|
*
|
|
|
|
* Returns the number of page that could not be moved.
|
|
|
|
*/
|
2012-05-29 16:06:24 -06:00
|
|
|
int do_migrate_pages(struct mm_struct *mm, const nodemask_t *from,
|
|
|
|
const nodemask_t *to, int flags)
|
2006-01-08 02:00:51 -07:00
|
|
|
{
|
2006-02-01 04:05:40 -07:00
|
|
|
int busy = 0;
|
2008-11-06 13:53:30 -07:00
|
|
|
int err;
|
2006-02-01 04:05:40 -07:00
|
|
|
nodemask_t tmp;
|
2006-01-08 02:00:51 -07:00
|
|
|
|
2008-11-06 13:53:30 -07:00
|
|
|
err = migrate_prep();
|
|
|
|
if (err)
|
|
|
|
return err;
|
|
|
|
|
mempolicy: mPOL_PREFERRED cleanups for "local allocation"
Here are a couple of "cleanups" for MPOL_PREFERRED behavior when
v.preferred_node < 0 -- i.e., "local allocation":
1) [do_]get_mempolicy() calls the now renamed get_policy_nodemask()
to fetch the nodemask associated with a policy. Currently,
get_policy_nodemask() returns the set of nodes with memory, when
the policy 'mode' is 'PREFERRED, and the preferred_node is < 0.
Change to return an empty nodemask, as this is what was specified
to achieve "local allocation".
2) When a task is moved into a [new] cpuset, mpol_rebind_policy() is
called to adjust any task and vma policy nodes to be valid in the
new cpuset. However, when the policy is MPOL_PREFERRED, and the
preferred_node is <0, no rebind is necessary. The "local allocation"
indication is valid in any cpuset. Existing code will "do the right
thing" because node_remap() will just return the argument node when
it is outside of the valid range of node ids. However, I think it is
clearer and cleaner to skip the remap explicitly in this case.
3) mpol_to_str() produces a printable, "human readable" string from a
struct mempolicy. For MPOL_PREFERRED with preferred_node <0, show
"local", as this indicates local allocation, as the task migrates
among nodes. Note that this matches the usage of "local allocation"
in libnuma() and numactl. Without this change, I believe that node_set()
[via set_bit()] will set bit 31, resulting in a misleading display.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:20 -06:00
|
|
|
down_read(&mm->mmap_sem);
|
2006-01-08 02:00:51 -07:00
|
|
|
|
2010-03-05 14:41:59 -07:00
|
|
|
/*
|
|
|
|
* Find a 'source' bit set in 'tmp' whose corresponding 'dest'
|
|
|
|
* bit in 'to' is not also set in 'tmp'. Clear the found 'source'
|
|
|
|
* bit in 'tmp', and return that <source, dest> pair for migration.
|
|
|
|
* The pair of nodemasks 'to' and 'from' define the map.
|
|
|
|
*
|
|
|
|
* If no pair of bits is found that way, fallback to picking some
|
|
|
|
* pair of 'source' and 'dest' bits that are not the same. If the
|
|
|
|
* 'source' and 'dest' bits are the same, this represents a node
|
|
|
|
* that will be migrating to itself, so no pages need move.
|
|
|
|
*
|
|
|
|
* If no bits are left in 'tmp', or if all remaining bits left
|
|
|
|
* in 'tmp' correspond to the same bit in 'to', return false
|
|
|
|
* (nothing left to migrate).
|
|
|
|
*
|
|
|
|
* This lets us pick a pair of nodes to migrate between, such that
|
|
|
|
* if possible the dest node is not already occupied by some other
|
|
|
|
* source node, minimizing the risk of overloading the memory on a
|
|
|
|
* node that would happen if we migrated incoming memory to a node
|
|
|
|
* before migrating outgoing memory source that same node.
|
|
|
|
*
|
|
|
|
* A single scan of tmp is sufficient. As we go, we remember the
|
|
|
|
* most recent <s, d> pair that moved (s != d). If we find a pair
|
|
|
|
* that not only moved, but what's better, moved to an empty slot
|
|
|
|
* (d is not set in tmp), then we break out then, with that pair.
|
2011-03-01 07:06:02 -07:00
|
|
|
* Otherwise when we finish scanning from_tmp, we at least have the
|
2010-03-05 14:41:59 -07:00
|
|
|
* most recent <s, d> pair that moved. If we get all the way through
|
|
|
|
* the scan of tmp without finding any node that moved, much less
|
|
|
|
* moved to an empty node, then there is nothing left worth migrating.
|
|
|
|
*/
|
2006-01-08 02:00:55 -07:00
|
|
|
|
2012-05-29 16:06:24 -06:00
|
|
|
tmp = *from;
|
2006-02-01 04:05:40 -07:00
|
|
|
while (!nodes_empty(tmp)) {
|
|
|
|
int s,d;
|
2013-11-12 16:07:39 -07:00
|
|
|
int source = NUMA_NO_NODE;
|
2006-02-01 04:05:40 -07:00
|
|
|
int dest = 0;
|
|
|
|
|
|
|
|
for_each_node_mask(s, tmp) {
|
mm: do_migrate_pages() calls migrate_to_node() even if task is already on a correct node
While running an application that moves tasks from one cpuset to another
I noticed that it takes much longer and moves many more pages than
expected.
The reason for this is do_migrate_pages() does its best to preserve the
relative node differential from the first node of the cpuset because the
application may have been written with that in mind. If memory was
interleaved on the nodes of the source cpuset by an application
do_migrate_pages() will try its best to maintain that interleaving on
the nodes of the destination cpuset. This means copying the memory from
all source nodes to the destination nodes even if the source and
destination nodes overlap.
This is a problem for userspace NUMA placement tools. The amount of
time spent doing extra memory moves cancels out some of the NUMA
performance improvements. Furthermore, if the number of source and
destination nodes are to maintain the previous interleaving layout
anyway.
This patch changes do_migrate_pages() to only preserve the relative
layout inside the program if the number of NUMA nodes in the source and
destination mask are the same. If the number is different, we do a much
more efficient migration by not touching memory that is in an allowed
node.
This preserves the old behaviour for programs that want it, while
allowing a userspace NUMA placement tool to use the new, faster
migration. This improves performance in our tests by up to a factor of
7.
Without this change migrating tasks from a cpuset containing nodes 0-7
to a cpuset containing nodes 3-4, we migrate from ALL the nodes even if
they are in the both the source and destination nodesets:
Migrating 7 to 4
Migrating 6 to 3
Migrating 5 to 4
Migrating 4 to 3
Migrating 1 to 4
Migrating 3 to 4
Migrating 0 to 3
Migrating 2 to 3
With this change we only migrate from nodes that are not in the
destination nodesets:
Migrating 7 to 4
Migrating 6 to 3
Migrating 5 to 4
Migrating 2 to 3
Migrating 1 to 4
Migrating 0 to 3
Yet if we move from a cpuset containing nodes 2,3,4 to a cpuset
containing 3,4,5 we still do move everything so that we preserve the
desired NUMA offsets:
Migrating 4 to 5
Migrating 3 to 4
Migrating 2 to 3
As far as performance is concerned this simple patch improves the time
it takes to move 14, 20 and 26 large tasks from a cpuset containing
nodes 0-7 to a cpuset containing nodes 1 & 3 by up to a factor of 7.
Here are the timings with and without the patch:
BEFORE PATCH -- Move times: 59, 140, 651 seconds
============
Moving 14 tasks from nodes (0-7) to nodes (1,3)
numad(8780) do_migrate_pages (mm=0xffff88081d414400
from_nodes=0xffff880818c81d28 to_nodes=0xffff880818c81ce8 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x7 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x6 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x5 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x4 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x2 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x1 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x0 dest=0x1 flags=0x4)
(Above moves repeated for each of the 14 tasks...)
PID 8890 moved to node(s) 1,3 in 59.2 seconds
Moving 20 tasks from nodes (0-7) to nodes (1,4-5)
numad(8780) do_migrate_pages (mm=0xffff88081d88c700
from_nodes=0xffff880818c81d28 to_nodes=0xffff880818c81ce8 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x7 dest=0x4 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x6 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x3 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x2 dest=0x5 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x1 dest=0x4 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x0 dest=0x1 flags=0x4)
(Above moves repeated for each of the 20 tasks...)
PID 8962 moved to node(s) 1,4-5 in 139.88 seconds
Moving 26 tasks from nodes (0-7) to nodes (1-3,5)
numad(8780) do_migrate_pages (mm=0xffff88081d5bc740
from_nodes=0xffff880818c81d28 to_nodes=0xffff880818c81ce8 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x7 dest=0x5 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x6 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x5 dest=0x2 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x3 dest=0x5 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x2 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x1 dest=0x2 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x0 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x4 dest=0x1 flags=0x4)
(Above moves repeated for each of the 26 tasks...)
PID 9058 moved to node(s) 1-3,5 in 651.45 seconds
AFTER PATCH -- Move times: 42, 56, 93 seconds
===========
Moving 14 tasks from nodes (0-7) to nodes (5,7)
numad(33209) do_migrate_pages (mm=0xffff88101d5ff140
from_nodes=0xffff88101e7b5d28 to_nodes=0xffff88101e7b5ce8 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x6 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x4 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x3 dest=0x7 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x2 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x1 dest=0x7 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x0 dest=0x5 flags=0x4)
(Above moves repeated for each of the 14 tasks...)
PID 33221 moved to node(s) 5,7 in 41.67 seconds
Moving 20 tasks from nodes (0-7) to nodes (1,3,5)
numad(33209) do_migrate_pages (mm=0xffff88101d6c37c0
from_nodes=0xffff88101e7b5d28 to_nodes=0xffff88101e7b5ce8 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x7 dest=0x3 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x6 dest=0x1 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x4 dest=0x3 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x2 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x0 dest=0x1 flags=0x4)
(Above moves repeated for each of the 20 tasks...)
PID 33289 moved to node(s) 1,3,5 in 56.3 seconds
Moving 26 tasks from nodes (0-7) to nodes (1,3,5,7)
numad(33209) do_migrate_pages (mm=0xffff88101d924400
from_nodes=0xffff88101e7b5d28 to_nodes=0xffff88101e7b5ce8 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d924400 source=0x6 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d924400 source=0x4 dest=0x1 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d924400 source=0x2 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d924400 source=0x0 dest=0x1 flags=0x4)
(Above moves repeated for each of the 26 tasks...)
PID 33372 moved to node(s) 1,3,5,7 in 92.67 seconds
[akpm@linux-foundation.org: clean up comment layout]
Signed-off-by: Larry Woodman <lwoodman@redhat.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-05-29 16:06:24 -06:00
|
|
|
|
|
|
|
/*
|
|
|
|
* do_migrate_pages() tries to maintain the relative
|
|
|
|
* node relationship of the pages established between
|
|
|
|
* threads and memory areas.
|
|
|
|
*
|
|
|
|
* However if the number of source nodes is not equal to
|
|
|
|
* the number of destination nodes we can not preserve
|
|
|
|
* this node relative relationship. In that case, skip
|
|
|
|
* copying memory from a node that is in the destination
|
|
|
|
* mask.
|
|
|
|
*
|
|
|
|
* Example: [2,3,4] -> [3,4,5] moves everything.
|
|
|
|
* [0-7] - > [3,4,5] moves only 0,1,2,6,7.
|
|
|
|
*/
|
|
|
|
|
2012-05-29 16:06:24 -06:00
|
|
|
if ((nodes_weight(*from) != nodes_weight(*to)) &&
|
|
|
|
(node_isset(s, *to)))
|
mm: do_migrate_pages() calls migrate_to_node() even if task is already on a correct node
While running an application that moves tasks from one cpuset to another
I noticed that it takes much longer and moves many more pages than
expected.
The reason for this is do_migrate_pages() does its best to preserve the
relative node differential from the first node of the cpuset because the
application may have been written with that in mind. If memory was
interleaved on the nodes of the source cpuset by an application
do_migrate_pages() will try its best to maintain that interleaving on
the nodes of the destination cpuset. This means copying the memory from
all source nodes to the destination nodes even if the source and
destination nodes overlap.
This is a problem for userspace NUMA placement tools. The amount of
time spent doing extra memory moves cancels out some of the NUMA
performance improvements. Furthermore, if the number of source and
destination nodes are to maintain the previous interleaving layout
anyway.
This patch changes do_migrate_pages() to only preserve the relative
layout inside the program if the number of NUMA nodes in the source and
destination mask are the same. If the number is different, we do a much
more efficient migration by not touching memory that is in an allowed
node.
This preserves the old behaviour for programs that want it, while
allowing a userspace NUMA placement tool to use the new, faster
migration. This improves performance in our tests by up to a factor of
7.
Without this change migrating tasks from a cpuset containing nodes 0-7
to a cpuset containing nodes 3-4, we migrate from ALL the nodes even if
they are in the both the source and destination nodesets:
Migrating 7 to 4
Migrating 6 to 3
Migrating 5 to 4
Migrating 4 to 3
Migrating 1 to 4
Migrating 3 to 4
Migrating 0 to 3
Migrating 2 to 3
With this change we only migrate from nodes that are not in the
destination nodesets:
Migrating 7 to 4
Migrating 6 to 3
Migrating 5 to 4
Migrating 2 to 3
Migrating 1 to 4
Migrating 0 to 3
Yet if we move from a cpuset containing nodes 2,3,4 to a cpuset
containing 3,4,5 we still do move everything so that we preserve the
desired NUMA offsets:
Migrating 4 to 5
Migrating 3 to 4
Migrating 2 to 3
As far as performance is concerned this simple patch improves the time
it takes to move 14, 20 and 26 large tasks from a cpuset containing
nodes 0-7 to a cpuset containing nodes 1 & 3 by up to a factor of 7.
Here are the timings with and without the patch:
BEFORE PATCH -- Move times: 59, 140, 651 seconds
============
Moving 14 tasks from nodes (0-7) to nodes (1,3)
numad(8780) do_migrate_pages (mm=0xffff88081d414400
from_nodes=0xffff880818c81d28 to_nodes=0xffff880818c81ce8 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x7 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x6 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x5 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x4 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x2 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x1 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d414400 source=0x0 dest=0x1 flags=0x4)
(Above moves repeated for each of the 14 tasks...)
PID 8890 moved to node(s) 1,3 in 59.2 seconds
Moving 20 tasks from nodes (0-7) to nodes (1,4-5)
numad(8780) do_migrate_pages (mm=0xffff88081d88c700
from_nodes=0xffff880818c81d28 to_nodes=0xffff880818c81ce8 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x7 dest=0x4 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x6 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x3 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x2 dest=0x5 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x1 dest=0x4 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d88c700 source=0x0 dest=0x1 flags=0x4)
(Above moves repeated for each of the 20 tasks...)
PID 8962 moved to node(s) 1,4-5 in 139.88 seconds
Moving 26 tasks from nodes (0-7) to nodes (1-3,5)
numad(8780) do_migrate_pages (mm=0xffff88081d5bc740
from_nodes=0xffff880818c81d28 to_nodes=0xffff880818c81ce8 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x7 dest=0x5 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x6 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x5 dest=0x2 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x3 dest=0x5 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x2 dest=0x3 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x1 dest=0x2 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x0 dest=0x1 flags=0x4)
numad(8780) migrate_to_node (mm=0xffff88081d5bc740 source=0x4 dest=0x1 flags=0x4)
(Above moves repeated for each of the 26 tasks...)
PID 9058 moved to node(s) 1-3,5 in 651.45 seconds
AFTER PATCH -- Move times: 42, 56, 93 seconds
===========
Moving 14 tasks from nodes (0-7) to nodes (5,7)
numad(33209) do_migrate_pages (mm=0xffff88101d5ff140
from_nodes=0xffff88101e7b5d28 to_nodes=0xffff88101e7b5ce8 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x6 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x4 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x3 dest=0x7 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x2 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x1 dest=0x7 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d5ff140 source=0x0 dest=0x5 flags=0x4)
(Above moves repeated for each of the 14 tasks...)
PID 33221 moved to node(s) 5,7 in 41.67 seconds
Moving 20 tasks from nodes (0-7) to nodes (1,3,5)
numad(33209) do_migrate_pages (mm=0xffff88101d6c37c0
from_nodes=0xffff88101e7b5d28 to_nodes=0xffff88101e7b5ce8 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x7 dest=0x3 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x6 dest=0x1 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x4 dest=0x3 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x2 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d6c37c0 source=0x0 dest=0x1 flags=0x4)
(Above moves repeated for each of the 20 tasks...)
PID 33289 moved to node(s) 1,3,5 in 56.3 seconds
Moving 26 tasks from nodes (0-7) to nodes (1,3,5,7)
numad(33209) do_migrate_pages (mm=0xffff88101d924400
from_nodes=0xffff88101e7b5d28 to_nodes=0xffff88101e7b5ce8 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d924400 source=0x6 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d924400 source=0x4 dest=0x1 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d924400 source=0x2 dest=0x5 flags=0x4)
numad(33209) migrate_to_node (mm=0xffff88101d924400 source=0x0 dest=0x1 flags=0x4)
(Above moves repeated for each of the 26 tasks...)
PID 33372 moved to node(s) 1,3,5,7 in 92.67 seconds
[akpm@linux-foundation.org: clean up comment layout]
Signed-off-by: Larry Woodman <lwoodman@redhat.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-05-29 16:06:24 -06:00
|
|
|
continue;
|
|
|
|
|
2012-05-29 16:06:24 -06:00
|
|
|
d = node_remap(s, *from, *to);
|
2006-02-01 04:05:40 -07:00
|
|
|
if (s == d)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
source = s; /* Node moved. Memorize */
|
|
|
|
dest = d;
|
|
|
|
|
|
|
|
/* dest not in remaining from nodes? */
|
|
|
|
if (!node_isset(dest, tmp))
|
|
|
|
break;
|
|
|
|
}
|
2013-11-12 16:07:39 -07:00
|
|
|
if (source == NUMA_NO_NODE)
|
2006-02-01 04:05:40 -07:00
|
|
|
break;
|
|
|
|
|
|
|
|
node_clear(source, tmp);
|
|
|
|
err = migrate_to_node(mm, source, dest, flags);
|
|
|
|
if (err > 0)
|
|
|
|
busy += err;
|
|
|
|
if (err < 0)
|
|
|
|
break;
|
2006-01-08 02:00:51 -07:00
|
|
|
}
|
|
|
|
up_read(&mm->mmap_sem);
|
2006-02-01 04:05:40 -07:00
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
return busy;
|
2006-03-22 01:09:12 -07:00
|
|
|
|
|
|
|
}
|
|
|
|
|
Migration: find correct vma in new_vma_page()
We hit the BUG_ON() in mm/rmap.c:vma_address() when trying to migrate via
mbind(MPOL_MF_MOVE) a non-anon region that spans multiple vmas. For
anon-regions, we just fail to migrate any pages beyond the 1st vma in the
range.
This occurs because do_mbind() collects a list of pages to migrate by
calling check_range(). check_range() walks the task's mm, spanning vmas as
necessary, to collect the migratable pages into a list. Then, do_mbind()
calls migrate_pages() passing the list of pages, a function to allocate new
pages based on vma policy [new_vma_page()], and a pointer to the first vma
of the range.
For each page in the list, new_vma_page() calls page_address_in_vma()
passing the page and the vma [first in range] to obtain the address to get
for alloc_page_vma(). The page address is needed to get interleaving
policy correct. If the pages in the list come from multiple vmas,
eventually, new_page_address() will pass that page to page_address_in_vma()
with the incorrect vma. For !PageAnon pages, this will result in a bug
check in rmap.c:vma_address(). For anon pages, vma_address() will just
return EFAULT and fail the migration.
This patch modifies new_vma_page() to check the return value from
page_address_in_vma(). If the return value is EFAULT, new_vma_page()
searchs forward via vm_next for the vma that maps the page--i.e., that does
not return EFAULT. This assumes that the pages in the list handed to
migrate_pages() is in address order. This is currently case. The patch
documents this assumption in a new comment block for new_vma_page().
If new_vma_page() cannot locate the vma mapping the page in a forward
search in the mm, it will pass a NULL vma to alloc_page_vma(). This will
result in the allocation using the task policy, if any, else system default
policy. This situation is unlikely, but the patch documents this behavior
with a comment.
Note, this patch results in restarting from the first vma in a multi-vma
range each time new_vma_page() is called. If this is not acceptable, we
can make the vma argument a pointer, both in new_vma_page() and it's caller
unmap_and_move() so that the value held by the loop in migrate_pages()
always passes down the last vma in which a page was found. This will
require changes to all new_page_t functions passed to migrate_pages(). Is
this necessary?
For this patch to work, we can't bug check in vma_address() for pages
outside the argument vma. This patch removes the BUG_ON(). All other
callers [besides new_vma_page()] already check the return status.
Tested on x86_64, 4 node NUMA platform.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-11-14 17:59:10 -07:00
|
|
|
/*
|
|
|
|
* Allocate a new page for page migration based on vma policy.
|
2014-06-23 14:22:07 -06:00
|
|
|
* Start by assuming the page is mapped by the same vma as contains @start.
|
Migration: find correct vma in new_vma_page()
We hit the BUG_ON() in mm/rmap.c:vma_address() when trying to migrate via
mbind(MPOL_MF_MOVE) a non-anon region that spans multiple vmas. For
anon-regions, we just fail to migrate any pages beyond the 1st vma in the
range.
This occurs because do_mbind() collects a list of pages to migrate by
calling check_range(). check_range() walks the task's mm, spanning vmas as
necessary, to collect the migratable pages into a list. Then, do_mbind()
calls migrate_pages() passing the list of pages, a function to allocate new
pages based on vma policy [new_vma_page()], and a pointer to the first vma
of the range.
For each page in the list, new_vma_page() calls page_address_in_vma()
passing the page and the vma [first in range] to obtain the address to get
for alloc_page_vma(). The page address is needed to get interleaving
policy correct. If the pages in the list come from multiple vmas,
eventually, new_page_address() will pass that page to page_address_in_vma()
with the incorrect vma. For !PageAnon pages, this will result in a bug
check in rmap.c:vma_address(). For anon pages, vma_address() will just
return EFAULT and fail the migration.
This patch modifies new_vma_page() to check the return value from
page_address_in_vma(). If the return value is EFAULT, new_vma_page()
searchs forward via vm_next for the vma that maps the page--i.e., that does
not return EFAULT. This assumes that the pages in the list handed to
migrate_pages() is in address order. This is currently case. The patch
documents this assumption in a new comment block for new_vma_page().
If new_vma_page() cannot locate the vma mapping the page in a forward
search in the mm, it will pass a NULL vma to alloc_page_vma(). This will
result in the allocation using the task policy, if any, else system default
policy. This situation is unlikely, but the patch documents this behavior
with a comment.
Note, this patch results in restarting from the first vma in a multi-vma
range each time new_vma_page() is called. If this is not acceptable, we
can make the vma argument a pointer, both in new_vma_page() and it's caller
unmap_and_move() so that the value held by the loop in migrate_pages()
always passes down the last vma in which a page was found. This will
require changes to all new_page_t functions passed to migrate_pages(). Is
this necessary?
For this patch to work, we can't bug check in vma_address() for pages
outside the argument vma. This patch removes the BUG_ON(). All other
callers [besides new_vma_page()] already check the return status.
Tested on x86_64, 4 node NUMA platform.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-11-14 17:59:10 -07:00
|
|
|
* Search forward from there, if not. N.B., this assumes that the
|
|
|
|
* list of pages handed to migrate_pages()--which is how we get here--
|
|
|
|
* is in virtual address order.
|
|
|
|
*/
|
2014-06-23 14:22:07 -06:00
|
|
|
static struct page *new_page(struct page *page, unsigned long start, int **x)
|
2006-06-23 03:03:53 -06:00
|
|
|
{
|
2014-06-23 14:22:07 -06:00
|
|
|
struct vm_area_struct *vma;
|
Migration: find correct vma in new_vma_page()
We hit the BUG_ON() in mm/rmap.c:vma_address() when trying to migrate via
mbind(MPOL_MF_MOVE) a non-anon region that spans multiple vmas. For
anon-regions, we just fail to migrate any pages beyond the 1st vma in the
range.
This occurs because do_mbind() collects a list of pages to migrate by
calling check_range(). check_range() walks the task's mm, spanning vmas as
necessary, to collect the migratable pages into a list. Then, do_mbind()
calls migrate_pages() passing the list of pages, a function to allocate new
pages based on vma policy [new_vma_page()], and a pointer to the first vma
of the range.
For each page in the list, new_vma_page() calls page_address_in_vma()
passing the page and the vma [first in range] to obtain the address to get
for alloc_page_vma(). The page address is needed to get interleaving
policy correct. If the pages in the list come from multiple vmas,
eventually, new_page_address() will pass that page to page_address_in_vma()
with the incorrect vma. For !PageAnon pages, this will result in a bug
check in rmap.c:vma_address(). For anon pages, vma_address() will just
return EFAULT and fail the migration.
This patch modifies new_vma_page() to check the return value from
page_address_in_vma(). If the return value is EFAULT, new_vma_page()
searchs forward via vm_next for the vma that maps the page--i.e., that does
not return EFAULT. This assumes that the pages in the list handed to
migrate_pages() is in address order. This is currently case. The patch
documents this assumption in a new comment block for new_vma_page().
If new_vma_page() cannot locate the vma mapping the page in a forward
search in the mm, it will pass a NULL vma to alloc_page_vma(). This will
result in the allocation using the task policy, if any, else system default
policy. This situation is unlikely, but the patch documents this behavior
with a comment.
Note, this patch results in restarting from the first vma in a multi-vma
range each time new_vma_page() is called. If this is not acceptable, we
can make the vma argument a pointer, both in new_vma_page() and it's caller
unmap_and_move() so that the value held by the loop in migrate_pages()
always passes down the last vma in which a page was found. This will
require changes to all new_page_t functions passed to migrate_pages(). Is
this necessary?
For this patch to work, we can't bug check in vma_address() for pages
outside the argument vma. This patch removes the BUG_ON(). All other
callers [besides new_vma_page()] already check the return status.
Tested on x86_64, 4 node NUMA platform.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-11-14 17:59:10 -07:00
|
|
|
unsigned long uninitialized_var(address);
|
2006-06-23 03:03:53 -06:00
|
|
|
|
2014-06-23 14:22:07 -06:00
|
|
|
vma = find_vma(current->mm, start);
|
Migration: find correct vma in new_vma_page()
We hit the BUG_ON() in mm/rmap.c:vma_address() when trying to migrate via
mbind(MPOL_MF_MOVE) a non-anon region that spans multiple vmas. For
anon-regions, we just fail to migrate any pages beyond the 1st vma in the
range.
This occurs because do_mbind() collects a list of pages to migrate by
calling check_range(). check_range() walks the task's mm, spanning vmas as
necessary, to collect the migratable pages into a list. Then, do_mbind()
calls migrate_pages() passing the list of pages, a function to allocate new
pages based on vma policy [new_vma_page()], and a pointer to the first vma
of the range.
For each page in the list, new_vma_page() calls page_address_in_vma()
passing the page and the vma [first in range] to obtain the address to get
for alloc_page_vma(). The page address is needed to get interleaving
policy correct. If the pages in the list come from multiple vmas,
eventually, new_page_address() will pass that page to page_address_in_vma()
with the incorrect vma. For !PageAnon pages, this will result in a bug
check in rmap.c:vma_address(). For anon pages, vma_address() will just
return EFAULT and fail the migration.
This patch modifies new_vma_page() to check the return value from
page_address_in_vma(). If the return value is EFAULT, new_vma_page()
searchs forward via vm_next for the vma that maps the page--i.e., that does
not return EFAULT. This assumes that the pages in the list handed to
migrate_pages() is in address order. This is currently case. The patch
documents this assumption in a new comment block for new_vma_page().
If new_vma_page() cannot locate the vma mapping the page in a forward
search in the mm, it will pass a NULL vma to alloc_page_vma(). This will
result in the allocation using the task policy, if any, else system default
policy. This situation is unlikely, but the patch documents this behavior
with a comment.
Note, this patch results in restarting from the first vma in a multi-vma
range each time new_vma_page() is called. If this is not acceptable, we
can make the vma argument a pointer, both in new_vma_page() and it's caller
unmap_and_move() so that the value held by the loop in migrate_pages()
always passes down the last vma in which a page was found. This will
require changes to all new_page_t functions passed to migrate_pages(). Is
this necessary?
For this patch to work, we can't bug check in vma_address() for pages
outside the argument vma. This patch removes the BUG_ON(). All other
callers [besides new_vma_page()] already check the return status.
Tested on x86_64, 4 node NUMA platform.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-11-14 17:59:10 -07:00
|
|
|
while (vma) {
|
|
|
|
address = page_address_in_vma(page, vma);
|
|
|
|
if (address != -EFAULT)
|
|
|
|
break;
|
|
|
|
vma = vma->vm_next;
|
|
|
|
}
|
2013-12-18 18:08:56 -07:00
|
|
|
|
|
|
|
if (PageHuge(page)) {
|
2014-01-23 16:53:15 -07:00
|
|
|
BUG_ON(!vma);
|
|
|
|
return alloc_huge_page_noerr(vma, address, 1);
|
2013-12-18 18:08:56 -07:00
|
|
|
}
|
2013-09-11 15:22:16 -06:00
|
|
|
/*
|
2013-12-18 18:08:56 -07:00
|
|
|
* if !vma, alloc_page_vma() will use task or system default policy
|
2013-09-11 15:22:16 -06:00
|
|
|
*/
|
Migration: find correct vma in new_vma_page()
We hit the BUG_ON() in mm/rmap.c:vma_address() when trying to migrate via
mbind(MPOL_MF_MOVE) a non-anon region that spans multiple vmas. For
anon-regions, we just fail to migrate any pages beyond the 1st vma in the
range.
This occurs because do_mbind() collects a list of pages to migrate by
calling check_range(). check_range() walks the task's mm, spanning vmas as
necessary, to collect the migratable pages into a list. Then, do_mbind()
calls migrate_pages() passing the list of pages, a function to allocate new
pages based on vma policy [new_vma_page()], and a pointer to the first vma
of the range.
For each page in the list, new_vma_page() calls page_address_in_vma()
passing the page and the vma [first in range] to obtain the address to get
for alloc_page_vma(). The page address is needed to get interleaving
policy correct. If the pages in the list come from multiple vmas,
eventually, new_page_address() will pass that page to page_address_in_vma()
with the incorrect vma. For !PageAnon pages, this will result in a bug
check in rmap.c:vma_address(). For anon pages, vma_address() will just
return EFAULT and fail the migration.
This patch modifies new_vma_page() to check the return value from
page_address_in_vma(). If the return value is EFAULT, new_vma_page()
searchs forward via vm_next for the vma that maps the page--i.e., that does
not return EFAULT. This assumes that the pages in the list handed to
migrate_pages() is in address order. This is currently case. The patch
documents this assumption in a new comment block for new_vma_page().
If new_vma_page() cannot locate the vma mapping the page in a forward
search in the mm, it will pass a NULL vma to alloc_page_vma(). This will
result in the allocation using the task policy, if any, else system default
policy. This situation is unlikely, but the patch documents this behavior
with a comment.
Note, this patch results in restarting from the first vma in a multi-vma
range each time new_vma_page() is called. If this is not acceptable, we
can make the vma argument a pointer, both in new_vma_page() and it's caller
unmap_and_move() so that the value held by the loop in migrate_pages()
always passes down the last vma in which a page was found. This will
require changes to all new_page_t functions passed to migrate_pages(). Is
this necessary?
For this patch to work, we can't bug check in vma_address() for pages
outside the argument vma. This patch removes the BUG_ON(). All other
callers [besides new_vma_page()] already check the return status.
Tested on x86_64, 4 node NUMA platform.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-11-14 17:59:10 -07:00
|
|
|
return alloc_page_vma(GFP_HIGHUSER_MOVABLE, vma, address);
|
2006-06-23 03:03:53 -06:00
|
|
|
}
|
2006-03-22 01:09:12 -07:00
|
|
|
#else
|
|
|
|
|
|
|
|
static void migrate_page_add(struct page *page, struct list_head *pagelist,
|
|
|
|
unsigned long flags)
|
|
|
|
{
|
2006-01-08 02:00:51 -07:00
|
|
|
}
|
|
|
|
|
2012-05-29 16:06:24 -06:00
|
|
|
int do_migrate_pages(struct mm_struct *mm, const nodemask_t *from,
|
|
|
|
const nodemask_t *to, int flags)
|
2006-03-22 01:09:12 -07:00
|
|
|
{
|
|
|
|
return -ENOSYS;
|
|
|
|
}
|
2006-06-23 03:03:53 -06:00
|
|
|
|
2014-06-23 14:22:07 -06:00
|
|
|
static struct page *new_page(struct page *page, unsigned long start, int **x)
|
2006-06-23 03:03:53 -06:00
|
|
|
{
|
|
|
|
return NULL;
|
|
|
|
}
|
2006-03-22 01:09:12 -07:00
|
|
|
#endif
|
|
|
|
|
2007-10-16 02:26:26 -06:00
|
|
|
static long do_mbind(unsigned long start, unsigned long len,
|
2008-04-28 03:12:25 -06:00
|
|
|
unsigned short mode, unsigned short mode_flags,
|
|
|
|
nodemask_t *nmask, unsigned long flags)
|
2006-01-08 02:01:04 -07:00
|
|
|
{
|
|
|
|
struct mm_struct *mm = current->mm;
|
|
|
|
struct mempolicy *new;
|
|
|
|
unsigned long end;
|
|
|
|
int err;
|
|
|
|
LIST_HEAD(pagelist);
|
|
|
|
|
2012-10-25 06:16:32 -06:00
|
|
|
if (flags & ~(unsigned long)MPOL_MF_VALID)
|
2006-01-08 02:01:04 -07:00
|
|
|
return -EINVAL;
|
2006-03-14 20:50:21 -07:00
|
|
|
if ((flags & MPOL_MF_MOVE_ALL) && !capable(CAP_SYS_NICE))
|
2006-01-08 02:01:04 -07:00
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
if (start & ~PAGE_MASK)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
if (mode == MPOL_DEFAULT)
|
|
|
|
flags &= ~MPOL_MF_STRICT;
|
|
|
|
|
|
|
|
len = (len + PAGE_SIZE - 1) & PAGE_MASK;
|
|
|
|
end = start + len;
|
|
|
|
|
|
|
|
if (end < start)
|
|
|
|
return -EINVAL;
|
|
|
|
if (end == start)
|
|
|
|
return 0;
|
|
|
|
|
2008-04-28 03:12:25 -06:00
|
|
|
new = mpol_new(mode, mode_flags, nmask);
|
2006-01-08 02:01:04 -07:00
|
|
|
if (IS_ERR(new))
|
|
|
|
return PTR_ERR(new);
|
|
|
|
|
2012-10-25 06:16:32 -06:00
|
|
|
if (flags & MPOL_MF_LAZY)
|
|
|
|
new->flags |= MPOL_F_MOF;
|
|
|
|
|
2006-01-08 02:01:04 -07:00
|
|
|
/*
|
|
|
|
* If we are using the default policy then operation
|
|
|
|
* on discontinuous address spaces is okay after all
|
|
|
|
*/
|
|
|
|
if (!new)
|
|
|
|
flags |= MPOL_MF_DISCONTIG_OK;
|
|
|
|
|
2008-04-28 03:12:25 -06:00
|
|
|
pr_debug("mbind %lx-%lx mode:%d flags:%d nodes:%lx\n",
|
|
|
|
start, start + len, mode, mode_flags,
|
2013-02-22 17:35:36 -07:00
|
|
|
nmask ? nodes_addr(*nmask)[0] : NUMA_NO_NODE);
|
2006-01-08 02:01:04 -07:00
|
|
|
|
2008-11-06 13:53:30 -07:00
|
|
|
if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) {
|
|
|
|
|
|
|
|
err = migrate_prep();
|
|
|
|
if (err)
|
2009-10-26 17:49:59 -06:00
|
|
|
goto mpol_out;
|
2008-11-06 13:53:30 -07:00
|
|
|
}
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
{
|
|
|
|
NODEMASK_SCRATCH(scratch);
|
|
|
|
if (scratch) {
|
|
|
|
down_write(&mm->mmap_sem);
|
|
|
|
task_lock(current);
|
|
|
|
err = mpol_set_nodemask(new, nmask, scratch);
|
|
|
|
task_unlock(current);
|
|
|
|
if (err)
|
|
|
|
up_write(&mm->mmap_sem);
|
|
|
|
} else
|
|
|
|
err = -ENOMEM;
|
|
|
|
NODEMASK_SCRATCH_FREE(scratch);
|
|
|
|
}
|
2009-10-26 17:49:59 -06:00
|
|
|
if (err)
|
|
|
|
goto mpol_out;
|
|
|
|
|
2014-06-23 14:22:07 -06:00
|
|
|
err = queue_pages_range(mm, start, end, nmask,
|
2006-01-08 02:01:04 -07:00
|
|
|
flags | MPOL_MF_INVERT, &pagelist);
|
2014-06-23 14:22:07 -06:00
|
|
|
if (!err)
|
mm: fix mbind vma merge problem
Strangely, current mbind() doesn't merge vma with neighbor vma although it's possible.
Unfortunately, many vma can reduce performance...
This patch fixes it.
reproduced program
----------------------------------------------------------------
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
char buf[128];
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);
fprintf(stderr, "pid = %d \n" "addr = %p\n", getpid(), addr);
/* make page populate */
memset(addr, 0, pagesize*3);
/* first mbind */
err = mbind(addr+pagesize, pagesize, MPOL_BIND, nmask->maskp,
nmask->size, MPOL_MF_MOVE_ALL);
if (err)
error("mbind1 ");
/* second mbind */
err = mbind(addr, pagesize*3, MPOL_DEFAULT, NULL, 0, 0);
if (err)
error("mbind2 ");
sprintf(buf, "cat /proc/%d/maps", getpid());
system(buf);
return 0;
}
----------------------------------------------------------------
result without this patch
addr = 0x7fe26ef09000
[snip]
7fe26ef09000-7fe26ef0a000 rw-p 00000000 00:00 0
7fe26ef0a000-7fe26ef0b000 rw-p 00000000 00:00 0
7fe26ef0b000-7fe26ef0c000 rw-p 00000000 00:00 0
7fe26ef0c000-7fe26ef0d000 rw-p 00000000 00:00 0
=> 0x7fe26ef09000-0x7fe26ef0c000 have three vmas.
result with this patch
addr = 0x7fc9ebc76000
[snip]
7fc9ebc76000-7fc9ebc7a000 rw-p 00000000 00:00 0
7fffbe690000-7fffbe6a5000 rw-p 00000000 00:00 0 [stack]
=> 0x7fc9ebc76000-0x7fc9ebc7a000 have only one vma.
[minchan.kim@gmail.com: fix file offset passed to vma_merge()]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-03-05 14:41:57 -07:00
|
|
|
err = mbind_range(mm, start, end, new);
|
2006-02-01 04:05:40 -07:00
|
|
|
|
2012-10-25 06:16:32 -06:00
|
|
|
if (!err) {
|
|
|
|
int nr_failed = 0;
|
|
|
|
|
2010-10-26 15:21:29 -06:00
|
|
|
if (!list_empty(&pagelist)) {
|
2012-10-25 06:16:32 -06:00
|
|
|
WARN_ON_ONCE(flags & MPOL_MF_LAZY);
|
2014-06-23 14:22:07 -06:00
|
|
|
nr_failed = migrate_pages(&pagelist, new_page, NULL,
|
|
|
|
start, MIGRATE_SYNC, MR_MEMPOLICY_MBIND);
|
2010-10-26 15:21:29 -06:00
|
|
|
if (nr_failed)
|
2013-09-11 15:22:06 -06:00
|
|
|
putback_movable_pages(&pagelist);
|
2010-10-26 15:21:29 -06:00
|
|
|
}
|
2006-01-08 02:01:04 -07:00
|
|
|
|
2012-10-25 06:16:32 -06:00
|
|
|
if (nr_failed && (flags & MPOL_MF_STRICT))
|
2006-01-08 02:01:04 -07:00
|
|
|
err = -EIO;
|
mbind(): fix leak of never putback pages
If mbind() receives an invalid address, do_mbind leaks a page. The
following test program detects this leak.
This patch fixes it.
migrate_efault.c
=======================================
#include <numaif.h>
#include <numa.h>
#include <sys/mman.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static unsigned long pagesize;
static void* make_hole_mapping(void)
{
void* addr;
addr = mmap(NULL, pagesize*3, PROT_READ|PROT_WRITE,
MAP_ANON|MAP_PRIVATE, 0, 0);
if (addr == MAP_FAILED)
return NULL;
/* make page populate */
memset(addr, 0, pagesize*3);
/* make memory hole */
munmap(addr+pagesize, pagesize);
return addr;
}
int main(int argc, char** argv)
{
void* addr;
int ch;
int node;
struct bitmask *nmask = numa_allocate_nodemask();
int err;
int node_set = 0;
while ((ch = getopt(argc, argv, "n:")) != -1){
switch (ch){
case 'n':
node = strtol(optarg, NULL, 0);
numa_bitmask_setbit(nmask, node);
node_set = 1;
break;
default:
;
}
}
argc -= optind;
argv += optind;
if (!node_set)
numa_bitmask_setbit(nmask, 0);
pagesize = getpagesize();
addr = make_hole_mapping();
err = mbind(addr, pagesize*3, MPOL_BIND, nmask->maskp, nmask->size, MPOL_MF_MOVE_ALL);
if (err)
perror("mbind ");
return 0;
}
=======================================
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Christoph Lameter <cl@linux-foundation.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-10-26 17:49:58 -06:00
|
|
|
} else
|
2013-12-18 18:08:51 -07:00
|
|
|
putback_movable_pages(&pagelist);
|
2006-03-22 01:09:12 -07:00
|
|
|
|
2006-01-08 02:01:04 -07:00
|
|
|
up_write(&mm->mmap_sem);
|
2009-10-26 17:49:59 -06:00
|
|
|
mpol_out:
|
2008-04-28 03:13:08 -06:00
|
|
|
mpol_put(new);
|
2006-01-08 02:01:04 -07:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2005-10-29 19:16:59 -06:00
|
|
|
/*
|
|
|
|
* User space interface with variable sized bitmaps for nodelists.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* Copy a node mask from user space. */
|
2006-01-08 02:00:51 -07:00
|
|
|
static int get_nodes(nodemask_t *nodes, const unsigned long __user *nmask,
|
2005-10-29 19:16:59 -06:00
|
|
|
unsigned long maxnode)
|
|
|
|
{
|
|
|
|
unsigned long k;
|
|
|
|
unsigned long nlongs;
|
|
|
|
unsigned long endmask;
|
|
|
|
|
|
|
|
--maxnode;
|
|
|
|
nodes_clear(*nodes);
|
|
|
|
if (maxnode == 0 || !nmask)
|
|
|
|
return 0;
|
2006-02-20 19:27:59 -07:00
|
|
|
if (maxnode > PAGE_SIZE*BITS_PER_BYTE)
|
2006-02-17 14:59:36 -07:00
|
|
|
return -EINVAL;
|
2005-10-29 19:16:59 -06:00
|
|
|
|
|
|
|
nlongs = BITS_TO_LONGS(maxnode);
|
|
|
|
if ((maxnode % BITS_PER_LONG) == 0)
|
|
|
|
endmask = ~0UL;
|
|
|
|
else
|
|
|
|
endmask = (1UL << (maxnode % BITS_PER_LONG)) - 1;
|
|
|
|
|
|
|
|
/* When the user specified more nodes than supported just check
|
|
|
|
if the non supported part is all zero. */
|
|
|
|
if (nlongs > BITS_TO_LONGS(MAX_NUMNODES)) {
|
|
|
|
if (nlongs > PAGE_SIZE/sizeof(long))
|
|
|
|
return -EINVAL;
|
|
|
|
for (k = BITS_TO_LONGS(MAX_NUMNODES); k < nlongs; k++) {
|
|
|
|
unsigned long t;
|
|
|
|
if (get_user(t, nmask + k))
|
|
|
|
return -EFAULT;
|
|
|
|
if (k == nlongs - 1) {
|
|
|
|
if (t & endmask)
|
|
|
|
return -EINVAL;
|
|
|
|
} else if (t)
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
nlongs = BITS_TO_LONGS(MAX_NUMNODES);
|
|
|
|
endmask = ~0UL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (copy_from_user(nodes_addr(*nodes), nmask, nlongs*sizeof(unsigned long)))
|
|
|
|
return -EFAULT;
|
|
|
|
nodes_addr(*nodes)[nlongs-1] &= endmask;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Copy a kernel node mask to user space */
|
|
|
|
static int copy_nodes_to_user(unsigned long __user *mask, unsigned long maxnode,
|
|
|
|
nodemask_t *nodes)
|
|
|
|
{
|
|
|
|
unsigned long copy = ALIGN(maxnode-1, 64) / 8;
|
|
|
|
const int nbytes = BITS_TO_LONGS(MAX_NUMNODES) * sizeof(long);
|
|
|
|
|
|
|
|
if (copy > nbytes) {
|
|
|
|
if (copy > PAGE_SIZE)
|
|
|
|
return -EINVAL;
|
|
|
|
if (clear_user((char __user *)mask + nbytes, copy - nbytes))
|
|
|
|
return -EFAULT;
|
|
|
|
copy = nbytes;
|
|
|
|
}
|
|
|
|
return copy_to_user(mask, nodes_addr(*nodes), copy) ? -EFAULT : 0;
|
|
|
|
}
|
|
|
|
|
2009-01-14 06:14:30 -07:00
|
|
|
SYSCALL_DEFINE6(mbind, unsigned long, start, unsigned long, len,
|
2014-06-04 17:07:57 -06:00
|
|
|
unsigned long, mode, const unsigned long __user *, nmask,
|
2009-01-14 06:14:30 -07:00
|
|
|
unsigned long, maxnode, unsigned, flags)
|
2005-10-29 19:16:59 -06:00
|
|
|
{
|
|
|
|
nodemask_t nodes;
|
|
|
|
int err;
|
2008-04-28 03:12:25 -06:00
|
|
|
unsigned short mode_flags;
|
2005-10-29 19:16:59 -06:00
|
|
|
|
2008-04-28 03:12:25 -06:00
|
|
|
mode_flags = mode & MPOL_MODE_FLAGS;
|
|
|
|
mode &= ~MPOL_MODE_FLAGS;
|
mempolicy: convert MPOL constants to enum
The mempolicy mode constants, MPOL_DEFAULT, MPOL_PREFERRED, MPOL_BIND, and
MPOL_INTERLEAVE, are better declared as part of an enum since they are
sequentially numbered and cannot be combined.
The policy member of struct mempolicy is also converted from type short to
type unsigned short. A negative policy does not have any legitimate meaning,
so it is possible to change its type in preparation for adding optional mode
flags later.
The equivalent member of struct shmem_sb_info is also changed from int to
unsigned short.
For compatibility, the policy formal to get_mempolicy() remains as a pointer
to an int:
int get_mempolicy(int *policy, unsigned long *nmask,
unsigned long maxnode, unsigned long addr,
unsigned long flags);
although the only possible values is the range of type unsigned short.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:23 -06:00
|
|
|
if (mode >= MPOL_MAX)
|
|
|
|
return -EINVAL;
|
mempolicy: add MPOL_F_RELATIVE_NODES flag
Adds another optional mode flag, MPOL_F_RELATIVE_NODES, that specifies
nodemasks passed via set_mempolicy() or mbind() should be considered relative
to the current task's mems_allowed.
When the mempolicy is created, the passed nodemask is folded and mapped onto
the current task's mems_allowed. For example, consider a task using
set_mempolicy() to pass MPOL_INTERLEAVE | MPOL_F_RELATIVE_NODES with a
nodemask of 1-3. If current's mems_allowed is 4-7, the effected nodemask is
5-7 (the second, third, and fourth node of mems_allowed).
If the same task is attached to a cpuset, the mempolicy nodemask is rebound
each time the mems are changed. Some possible rebinds and results are:
mems result
1-3 1-3
1-7 2-4
1,5-6 1,5-6
1,5-7 5-7
Likewise, the zonelist built for MPOL_BIND acts on the set of zones assigned
to the resultant nodemask from the relative remap.
In the MPOL_PREFERRED case, the preferred node is remapped from the currently
effected nodemask to the relative nodemask.
This mempolicy mode flag was conceived of by Paul Jackson <pj@sgi.com>.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:30 -06:00
|
|
|
if ((mode_flags & MPOL_F_STATIC_NODES) &&
|
|
|
|
(mode_flags & MPOL_F_RELATIVE_NODES))
|
|
|
|
return -EINVAL;
|
2005-10-29 19:16:59 -06:00
|
|
|
err = get_nodes(&nodes, nmask, maxnode);
|
|
|
|
if (err)
|
|
|
|
return err;
|
2008-04-28 03:12:25 -06:00
|
|
|
return do_mbind(start, len, mode, mode_flags, &nodes, flags);
|
2005-10-29 19:16:59 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Set the process memory policy */
|
2014-06-04 17:07:58 -06:00
|
|
|
SYSCALL_DEFINE3(set_mempolicy, int, mode, const unsigned long __user *, nmask,
|
2009-01-14 06:14:30 -07:00
|
|
|
unsigned long, maxnode)
|
2005-10-29 19:16:59 -06:00
|
|
|
{
|
|
|
|
int err;
|
|
|
|
nodemask_t nodes;
|
2008-04-28 03:12:25 -06:00
|
|
|
unsigned short flags;
|
2005-10-29 19:16:59 -06:00
|
|
|
|
2008-04-28 03:12:25 -06:00
|
|
|
flags = mode & MPOL_MODE_FLAGS;
|
|
|
|
mode &= ~MPOL_MODE_FLAGS;
|
|
|
|
if ((unsigned int)mode >= MPOL_MAX)
|
2005-10-29 19:16:59 -06:00
|
|
|
return -EINVAL;
|
mempolicy: add MPOL_F_RELATIVE_NODES flag
Adds another optional mode flag, MPOL_F_RELATIVE_NODES, that specifies
nodemasks passed via set_mempolicy() or mbind() should be considered relative
to the current task's mems_allowed.
When the mempolicy is created, the passed nodemask is folded and mapped onto
the current task's mems_allowed. For example, consider a task using
set_mempolicy() to pass MPOL_INTERLEAVE | MPOL_F_RELATIVE_NODES with a
nodemask of 1-3. If current's mems_allowed is 4-7, the effected nodemask is
5-7 (the second, third, and fourth node of mems_allowed).
If the same task is attached to a cpuset, the mempolicy nodemask is rebound
each time the mems are changed. Some possible rebinds and results are:
mems result
1-3 1-3
1-7 2-4
1,5-6 1,5-6
1,5-7 5-7
Likewise, the zonelist built for MPOL_BIND acts on the set of zones assigned
to the resultant nodemask from the relative remap.
In the MPOL_PREFERRED case, the preferred node is remapped from the currently
effected nodemask to the relative nodemask.
This mempolicy mode flag was conceived of by Paul Jackson <pj@sgi.com>.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:30 -06:00
|
|
|
if ((flags & MPOL_F_STATIC_NODES) && (flags & MPOL_F_RELATIVE_NODES))
|
|
|
|
return -EINVAL;
|
2005-10-29 19:16:59 -06:00
|
|
|
err = get_nodes(&nodes, nmask, maxnode);
|
|
|
|
if (err)
|
|
|
|
return err;
|
2008-04-28 03:12:25 -06:00
|
|
|
return do_set_mempolicy(mode, flags, &nodes);
|
2005-10-29 19:16:59 -06:00
|
|
|
}
|
|
|
|
|
2009-01-14 06:14:30 -07:00
|
|
|
SYSCALL_DEFINE4(migrate_pages, pid_t, pid, unsigned long, maxnode,
|
|
|
|
const unsigned long __user *, old_nodes,
|
|
|
|
const unsigned long __user *, new_nodes)
|
2006-01-08 02:00:51 -07:00
|
|
|
{
|
2008-11-13 16:39:19 -07:00
|
|
|
const struct cred *cred = current_cred(), *tcred;
|
2010-08-09 18:19:01 -06:00
|
|
|
struct mm_struct *mm = NULL;
|
2006-01-08 02:00:51 -07:00
|
|
|
struct task_struct *task;
|
|
|
|
nodemask_t task_nodes;
|
|
|
|
int err;
|
2010-08-09 18:19:01 -06:00
|
|
|
nodemask_t *old;
|
|
|
|
nodemask_t *new;
|
|
|
|
NODEMASK_SCRATCH(scratch);
|
|
|
|
|
|
|
|
if (!scratch)
|
|
|
|
return -ENOMEM;
|
2006-01-08 02:00:51 -07:00
|
|
|
|
2010-08-09 18:19:01 -06:00
|
|
|
old = &scratch->mask1;
|
|
|
|
new = &scratch->mask2;
|
|
|
|
|
|
|
|
err = get_nodes(old, old_nodes, maxnode);
|
2006-01-08 02:00:51 -07:00
|
|
|
if (err)
|
2010-08-09 18:19:01 -06:00
|
|
|
goto out;
|
2006-01-08 02:00:51 -07:00
|
|
|
|
2010-08-09 18:19:01 -06:00
|
|
|
err = get_nodes(new, new_nodes, maxnode);
|
2006-01-08 02:00:51 -07:00
|
|
|
if (err)
|
2010-08-09 18:19:01 -06:00
|
|
|
goto out;
|
2006-01-08 02:00:51 -07:00
|
|
|
|
|
|
|
/* Find the mm_struct */
|
2010-12-02 15:31:13 -07:00
|
|
|
rcu_read_lock();
|
2007-10-19 00:40:16 -06:00
|
|
|
task = pid ? find_task_by_vpid(pid) : current;
|
2006-01-08 02:00:51 -07:00
|
|
|
if (!task) {
|
2010-12-02 15:31:13 -07:00
|
|
|
rcu_read_unlock();
|
2010-08-09 18:19:01 -06:00
|
|
|
err = -ESRCH;
|
|
|
|
goto out;
|
2006-01-08 02:00:51 -07:00
|
|
|
}
|
2012-03-21 17:34:06 -06:00
|
|
|
get_task_struct(task);
|
2006-01-08 02:00:51 -07:00
|
|
|
|
2010-08-09 18:19:01 -06:00
|
|
|
err = -EINVAL;
|
2006-01-08 02:00:51 -07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Check if this process has the right to modify the specified
|
|
|
|
* process. The right exists if the process has administrative
|
2006-03-28 02:56:53 -07:00
|
|
|
* capabilities, superuser privileges or the same
|
2006-01-08 02:00:51 -07:00
|
|
|
* userid as the target process.
|
|
|
|
*/
|
2008-11-13 16:39:19 -07:00
|
|
|
tcred = __task_cred(task);
|
2012-03-12 16:48:24 -06:00
|
|
|
if (!uid_eq(cred->euid, tcred->suid) && !uid_eq(cred->euid, tcred->uid) &&
|
|
|
|
!uid_eq(cred->uid, tcred->suid) && !uid_eq(cred->uid, tcred->uid) &&
|
2006-03-14 20:50:21 -07:00
|
|
|
!capable(CAP_SYS_NICE)) {
|
2008-11-13 16:39:19 -07:00
|
|
|
rcu_read_unlock();
|
2006-01-08 02:00:51 -07:00
|
|
|
err = -EPERM;
|
2012-03-21 17:34:06 -06:00
|
|
|
goto out_put;
|
2006-01-08 02:00:51 -07:00
|
|
|
}
|
2008-11-13 16:39:19 -07:00
|
|
|
rcu_read_unlock();
|
2006-01-08 02:00:51 -07:00
|
|
|
|
|
|
|
task_nodes = cpuset_mems_allowed(task);
|
|
|
|
/* Is the user allowed to access the target nodes? */
|
2010-08-09 18:19:01 -06:00
|
|
|
if (!nodes_subset(*new, task_nodes) && !capable(CAP_SYS_NICE)) {
|
2006-01-08 02:00:51 -07:00
|
|
|
err = -EPERM;
|
2012-03-21 17:34:06 -06:00
|
|
|
goto out_put;
|
2006-01-08 02:00:51 -07:00
|
|
|
}
|
|
|
|
|
2012-12-12 14:51:33 -07:00
|
|
|
if (!nodes_subset(*new, node_states[N_MEMORY])) {
|
2007-08-31 01:12:08 -06:00
|
|
|
err = -EINVAL;
|
2012-03-21 17:34:06 -06:00
|
|
|
goto out_put;
|
2007-08-31 01:12:08 -06:00
|
|
|
}
|
|
|
|
|
2006-06-23 03:04:02 -06:00
|
|
|
err = security_task_movememory(task);
|
|
|
|
if (err)
|
2012-03-21 17:34:06 -06:00
|
|
|
goto out_put;
|
2006-06-23 03:04:02 -06:00
|
|
|
|
2012-03-21 17:34:06 -06:00
|
|
|
mm = get_task_mm(task);
|
|
|
|
put_task_struct(task);
|
2012-04-25 17:01:52 -06:00
|
|
|
|
|
|
|
if (!mm) {
|
2012-03-21 17:34:06 -06:00
|
|
|
err = -EINVAL;
|
2012-04-25 17:01:52 -06:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
err = do_migrate_pages(mm, old, new,
|
|
|
|
capable(CAP_SYS_NICE) ? MPOL_MF_MOVE_ALL : MPOL_MF_MOVE);
|
2012-03-21 17:34:06 -06:00
|
|
|
|
|
|
|
mmput(mm);
|
|
|
|
out:
|
2010-08-09 18:19:01 -06:00
|
|
|
NODEMASK_SCRATCH_FREE(scratch);
|
|
|
|
|
2006-01-08 02:00:51 -07:00
|
|
|
return err;
|
2012-03-21 17:34:06 -06:00
|
|
|
|
|
|
|
out_put:
|
|
|
|
put_task_struct(task);
|
|
|
|
goto out;
|
|
|
|
|
2006-01-08 02:00:51 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2005-10-29 19:16:59 -06:00
|
|
|
/* Retrieve NUMA policy */
|
2009-01-14 06:14:30 -07:00
|
|
|
SYSCALL_DEFINE5(get_mempolicy, int __user *, policy,
|
|
|
|
unsigned long __user *, nmask, unsigned long, maxnode,
|
|
|
|
unsigned long, addr, unsigned long, flags)
|
2005-10-29 19:16:59 -06:00
|
|
|
{
|
2007-10-16 02:26:26 -06:00
|
|
|
int err;
|
|
|
|
int uninitialized_var(pval);
|
2005-10-29 19:16:59 -06:00
|
|
|
nodemask_t nodes;
|
|
|
|
|
|
|
|
if (nmask != NULL && maxnode < MAX_NUMNODES)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
err = do_get_mempolicy(&pval, &nodes, addr, flags);
|
|
|
|
|
|
|
|
if (err)
|
|
|
|
return err;
|
|
|
|
|
|
|
|
if (policy && put_user(pval, policy))
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
if (nmask)
|
|
|
|
err = copy_nodes_to_user(nmask, maxnode, &nodes);
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
#ifdef CONFIG_COMPAT
|
|
|
|
|
2014-03-03 08:32:26 -07:00
|
|
|
COMPAT_SYSCALL_DEFINE5(get_mempolicy, int __user *, policy,
|
|
|
|
compat_ulong_t __user *, nmask,
|
|
|
|
compat_ulong_t, maxnode,
|
|
|
|
compat_ulong_t, addr, compat_ulong_t, flags)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
long err;
|
|
|
|
unsigned long __user *nm = NULL;
|
|
|
|
unsigned long nr_bits, alloc_size;
|
|
|
|
DECLARE_BITMAP(bm, MAX_NUMNODES);
|
|
|
|
|
|
|
|
nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
|
|
|
|
alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
|
|
|
|
|
|
|
|
if (nmask)
|
|
|
|
nm = compat_alloc_user_space(alloc_size);
|
|
|
|
|
|
|
|
err = sys_get_mempolicy(policy, nm, nr_bits+1, addr, flags);
|
|
|
|
|
|
|
|
if (!err && nmask) {
|
2011-09-14 17:21:02 -06:00
|
|
|
unsigned long copy_size;
|
|
|
|
copy_size = min_t(unsigned long, sizeof(bm), alloc_size);
|
|
|
|
err = copy_from_user(bm, nm, copy_size);
|
2005-04-16 16:20:36 -06:00
|
|
|
/* ensure entire bitmap is zeroed */
|
|
|
|
err |= clear_user(nmask, ALIGN(maxnode-1, 8) / 8);
|
|
|
|
err |= compat_put_bitmap(nmask, bm, nr_bits);
|
|
|
|
}
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2014-03-03 08:32:26 -07:00
|
|
|
COMPAT_SYSCALL_DEFINE3(set_mempolicy, int, mode, compat_ulong_t __user *, nmask,
|
|
|
|
compat_ulong_t, maxnode)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
long err = 0;
|
|
|
|
unsigned long __user *nm = NULL;
|
|
|
|
unsigned long nr_bits, alloc_size;
|
|
|
|
DECLARE_BITMAP(bm, MAX_NUMNODES);
|
|
|
|
|
|
|
|
nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
|
|
|
|
alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
|
|
|
|
|
|
|
|
if (nmask) {
|
|
|
|
err = compat_get_bitmap(bm, nmask, nr_bits);
|
|
|
|
nm = compat_alloc_user_space(alloc_size);
|
|
|
|
err |= copy_to_user(nm, bm, alloc_size);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (err)
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
return sys_set_mempolicy(mode, nm, nr_bits+1);
|
|
|
|
}
|
|
|
|
|
2014-03-03 08:32:26 -07:00
|
|
|
COMPAT_SYSCALL_DEFINE6(mbind, compat_ulong_t, start, compat_ulong_t, len,
|
|
|
|
compat_ulong_t, mode, compat_ulong_t __user *, nmask,
|
|
|
|
compat_ulong_t, maxnode, compat_ulong_t, flags)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
long err = 0;
|
|
|
|
unsigned long __user *nm = NULL;
|
|
|
|
unsigned long nr_bits, alloc_size;
|
2005-10-29 19:15:48 -06:00
|
|
|
nodemask_t bm;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
nr_bits = min_t(unsigned long, maxnode-1, MAX_NUMNODES);
|
|
|
|
alloc_size = ALIGN(nr_bits, BITS_PER_LONG) / 8;
|
|
|
|
|
|
|
|
if (nmask) {
|
2005-10-29 19:15:48 -06:00
|
|
|
err = compat_get_bitmap(nodes_addr(bm), nmask, nr_bits);
|
2005-04-16 16:20:36 -06:00
|
|
|
nm = compat_alloc_user_space(alloc_size);
|
2005-10-29 19:15:48 -06:00
|
|
|
err |= copy_to_user(nm, nodes_addr(bm), alloc_size);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
if (err)
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
return sys_mbind(start, len, mode, nm, nr_bits+1, flags);
|
|
|
|
}
|
|
|
|
|
|
|
|
#endif
|
|
|
|
|
2014-10-09 16:27:50 -06:00
|
|
|
struct mempolicy *__get_vma_policy(struct vm_area_struct *vma,
|
|
|
|
unsigned long addr)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2014-10-09 16:27:45 -06:00
|
|
|
struct mempolicy *pol = NULL;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (vma) {
|
Fix NUMA Memory Policy Reference Counting
This patch proposes fixes to the reference counting of memory policy in the
page allocation paths and in show_numa_map(). Extracted from my "Memory
Policy Cleanups and Enhancements" series as stand-alone.
Shared policy lookup [shmem] has always added a reference to the policy,
but this was never unrefed after page allocation or after formatting the
numa map data.
Default system policy should not require additional ref counting, nor
should the current task's task policy. However, show_numa_map() calls
get_vma_policy() to examine what may be [likely is] another task's policy.
The latter case needs protection against freeing of the policy.
This patch adds a reference count to a mempolicy returned by
get_vma_policy() when the policy is a vma policy or another task's
mempolicy. Again, shared policy is already reference counted on lookup. A
matching "unref" [__mpol_free()] is performed in alloc_page_vma() for
shared and vma policies, and in show_numa_map() for shared and another
task's mempolicy. We can call __mpol_free() directly, saving an admittedly
inexpensive inline NULL test, because we know we have a non-NULL policy.
Handling policy ref counts for hugepages is a bit trickier.
huge_zonelist() returns a zone list that might come from a shared or vma
'BIND policy. In this case, we should hold the reference until after the
huge page allocation in dequeue_hugepage(). The patch modifies
huge_zonelist() to return a pointer to the mempolicy if it needs to be
unref'd after allocation.
Kernel Build [16cpu, 32GB, ia64] - average of 10 runs:
w/o patch w/ refcount patch
Avg Std Devn Avg Std Devn
Real: 100.59 0.38 100.63 0.43
User: 1209.60 0.37 1209.91 0.31
System: 81.52 0.42 81.64 0.34
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Andi Kleen <ak@suse.de>
Cc: Christoph Lameter <clameter@sgi.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-09-18 23:46:47 -06:00
|
|
|
if (vma->vm_ops && vma->vm_ops->get_policy) {
|
2014-10-09 16:27:45 -06:00
|
|
|
pol = vma->vm_ops->get_policy(vma, addr);
|
2012-10-08 17:29:20 -06:00
|
|
|
} else if (vma->vm_policy) {
|
2005-04-16 16:20:36 -06:00
|
|
|
pol = vma->vm_policy;
|
2012-10-08 17:29:20 -06:00
|
|
|
|
|
|
|
/*
|
|
|
|
* shmem_alloc_page() passes MPOL_F_SHARED policy with
|
|
|
|
* a pseudo vma whose vma->vm_ops=NULL. Take a reference
|
|
|
|
* count on these policies which will be dropped by
|
|
|
|
* mpol_cond_put() later
|
|
|
|
*/
|
|
|
|
if (mpol_needs_cond_ref(pol))
|
|
|
|
mpol_get(pol);
|
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
2014-10-09 16:27:43 -06:00
|
|
|
|
2014-10-09 16:27:50 -06:00
|
|
|
return pol;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2014-10-09 16:27:57 -06:00
|
|
|
* get_vma_policy(@vma, @addr)
|
2014-10-09 16:27:50 -06:00
|
|
|
* @vma: virtual memory area whose policy is sought
|
|
|
|
* @addr: address in @vma for shared policy lookup
|
|
|
|
*
|
|
|
|
* Returns effective policy for a VMA at specified address.
|
2014-10-09 16:27:57 -06:00
|
|
|
* Falls back to current->mempolicy or system default policy, as necessary.
|
2014-10-09 16:27:50 -06:00
|
|
|
* Shared policies [those marked as MPOL_F_SHARED] require an extra reference
|
|
|
|
* count--added by the get_policy() vm_op, as appropriate--to protect against
|
|
|
|
* freeing by another task. It is the caller's responsibility to free the
|
|
|
|
* extra reference for shared policies.
|
|
|
|
*/
|
2014-10-09 16:27:57 -06:00
|
|
|
static struct mempolicy *get_vma_policy(struct vm_area_struct *vma,
|
|
|
|
unsigned long addr)
|
2014-10-09 16:27:50 -06:00
|
|
|
{
|
|
|
|
struct mempolicy *pol = __get_vma_policy(vma, addr);
|
|
|
|
|
2014-10-09 16:27:45 -06:00
|
|
|
if (!pol)
|
2014-10-09 16:27:57 -06:00
|
|
|
pol = get_task_policy(current);
|
2014-10-09 16:27:45 -06:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
return pol;
|
|
|
|
}
|
|
|
|
|
2014-10-09 16:27:48 -06:00
|
|
|
bool vma_policy_mof(struct vm_area_struct *vma)
|
2013-10-07 04:29:09 -06:00
|
|
|
{
|
2014-10-09 16:27:48 -06:00
|
|
|
struct mempolicy *pol;
|
2013-10-07 04:29:09 -06:00
|
|
|
|
2014-10-09 16:27:48 -06:00
|
|
|
if (vma->vm_ops && vma->vm_ops->get_policy) {
|
|
|
|
bool ret = false;
|
2013-10-07 04:29:09 -06:00
|
|
|
|
2014-10-09 16:27:48 -06:00
|
|
|
pol = vma->vm_ops->get_policy(vma, vma->vm_start);
|
|
|
|
if (pol && (pol->flags & MPOL_F_MOF))
|
|
|
|
ret = true;
|
|
|
|
mpol_cond_put(pol);
|
2014-10-09 16:27:45 -06:00
|
|
|
|
2014-10-09 16:27:48 -06:00
|
|
|
return ret;
|
2013-10-07 04:29:09 -06:00
|
|
|
}
|
|
|
|
|
2014-10-09 16:27:48 -06:00
|
|
|
pol = vma->vm_policy;
|
2014-10-09 16:27:45 -06:00
|
|
|
if (!pol)
|
2014-10-09 16:27:48 -06:00
|
|
|
pol = get_task_policy(current);
|
2014-10-09 16:27:45 -06:00
|
|
|
|
2013-10-07 04:29:09 -06:00
|
|
|
return pol->flags & MPOL_F_MOF;
|
|
|
|
}
|
|
|
|
|
2013-02-22 17:33:22 -07:00
|
|
|
static int apply_policy_zone(struct mempolicy *policy, enum zone_type zone)
|
|
|
|
{
|
|
|
|
enum zone_type dynamic_policy_zone = policy_zone;
|
|
|
|
|
|
|
|
BUG_ON(dynamic_policy_zone == ZONE_MOVABLE);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if policy->v.nodes has movable memory only,
|
|
|
|
* we apply policy when gfp_zone(gfp) = ZONE_MOVABLE only.
|
|
|
|
*
|
|
|
|
* policy->v.nodes is intersect with node_states[N_MEMORY].
|
|
|
|
* so if the following test faile, it implies
|
|
|
|
* policy->v.nodes has movable memory only.
|
|
|
|
*/
|
|
|
|
if (!nodes_intersects(policy->v.nodes, node_states[N_HIGH_MEMORY]))
|
|
|
|
dynamic_policy_zone = ZONE_MOVABLE;
|
|
|
|
|
|
|
|
return zone >= dynamic_policy_zone;
|
|
|
|
}
|
|
|
|
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
/*
|
|
|
|
* Return a nodemask representing a mempolicy for filtering nodes for
|
|
|
|
* page allocation
|
|
|
|
*/
|
|
|
|
static nodemask_t *policy_nodemask(gfp_t gfp, struct mempolicy *policy)
|
2008-04-28 03:12:18 -06:00
|
|
|
{
|
|
|
|
/* Lower zones don't get a nodemask applied for MPOL_BIND */
|
2008-04-28 03:13:12 -06:00
|
|
|
if (unlikely(policy->mode == MPOL_BIND) &&
|
2013-02-22 17:33:22 -07:00
|
|
|
apply_policy_zone(policy, gfp_zone(gfp)) &&
|
2008-04-28 03:12:18 -06:00
|
|
|
cpuset_nodemask_valid_mems_allowed(&policy->v.nodes))
|
|
|
|
return &policy->v.nodes;
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
/* Return a zonelist indicated by gfp for node representing a mempolicy */
|
2011-03-04 18:36:29 -07:00
|
|
|
static struct zonelist *policy_zonelist(gfp_t gfp, struct mempolicy *policy,
|
|
|
|
int nd)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2008-04-28 03:13:12 -06:00
|
|
|
switch (policy->mode) {
|
2005-04-16 16:20:36 -06:00
|
|
|
case MPOL_PREFERRED:
|
2008-04-28 03:13:21 -06:00
|
|
|
if (!(policy->flags & MPOL_F_LOCAL))
|
|
|
|
nd = policy->v.preferred_node;
|
2005-04-16 16:20:36 -06:00
|
|
|
break;
|
|
|
|
case MPOL_BIND:
|
2008-04-28 03:12:18 -06:00
|
|
|
/*
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
* Normally, MPOL_BIND allocations are node-local within the
|
|
|
|
* allowed nodemask. However, if __GFP_THISNODE is set and the
|
2010-05-24 15:32:00 -06:00
|
|
|
* current node isn't part of the mask, we use the zonelist for
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
* the first node in the mask instead.
|
2008-04-28 03:12:18 -06:00
|
|
|
*/
|
|
|
|
if (unlikely(gfp & __GFP_THISNODE) &&
|
|
|
|
unlikely(!node_isset(nd, policy->v.nodes)))
|
|
|
|
nd = first_node(policy->v.nodes);
|
|
|
|
break;
|
2005-04-16 16:20:36 -06:00
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
2008-04-28 03:12:14 -06:00
|
|
|
return node_zonelist(nd, gfp);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Do dynamic interleaving for a process */
|
|
|
|
static unsigned interleave_nodes(struct mempolicy *policy)
|
|
|
|
{
|
|
|
|
unsigned nid, next;
|
|
|
|
struct task_struct *me = current;
|
|
|
|
|
|
|
|
nid = me->il_next;
|
2016-05-19 18:10:58 -06:00
|
|
|
next = next_node_in(nid, policy->v.nodes);
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
if (next < MAX_NUMNODES)
|
|
|
|
me->il_next = next;
|
2005-04-16 16:20:36 -06:00
|
|
|
return nid;
|
|
|
|
}
|
|
|
|
|
2006-01-18 18:42:36 -07:00
|
|
|
/*
|
|
|
|
* Depending on the memory policy provide a node from which to allocate the
|
|
|
|
* next slab entry.
|
|
|
|
*/
|
2014-04-07 16:37:29 -06:00
|
|
|
unsigned int mempolicy_slab_node(void)
|
2006-01-18 18:42:36 -07:00
|
|
|
{
|
2012-06-09 03:40:03 -06:00
|
|
|
struct mempolicy *policy;
|
2014-04-07 16:37:29 -06:00
|
|
|
int node = numa_mem_id();
|
2012-06-09 03:40:03 -06:00
|
|
|
|
|
|
|
if (in_interrupt())
|
2014-04-07 16:37:29 -06:00
|
|
|
return node;
|
2012-06-09 03:40:03 -06:00
|
|
|
|
|
|
|
policy = current->mempolicy;
|
2008-04-28 03:13:21 -06:00
|
|
|
if (!policy || policy->flags & MPOL_F_LOCAL)
|
2014-04-07 16:37:29 -06:00
|
|
|
return node;
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
|
|
|
|
switch (policy->mode) {
|
|
|
|
case MPOL_PREFERRED:
|
2008-04-28 03:13:21 -06:00
|
|
|
/*
|
|
|
|
* handled MPOL_F_LOCAL above
|
|
|
|
*/
|
|
|
|
return policy->v.preferred_node;
|
2006-09-27 02:50:08 -06:00
|
|
|
|
2006-01-18 18:42:36 -07:00
|
|
|
case MPOL_INTERLEAVE:
|
|
|
|
return interleave_nodes(policy);
|
|
|
|
|
2008-04-28 03:12:17 -06:00
|
|
|
case MPOL_BIND: {
|
2016-05-19 18:14:10 -06:00
|
|
|
struct zoneref *z;
|
|
|
|
|
2006-01-18 18:42:36 -07:00
|
|
|
/*
|
|
|
|
* Follow bind policy behavior and start allocation at the
|
|
|
|
* first node.
|
|
|
|
*/
|
2008-04-28 03:12:18 -06:00
|
|
|
struct zonelist *zonelist;
|
|
|
|
enum zone_type highest_zoneidx = gfp_zone(GFP_KERNEL);
|
2014-04-07 16:37:29 -06:00
|
|
|
zonelist = &NODE_DATA(node)->node_zonelists[0];
|
2016-05-19 18:14:10 -06:00
|
|
|
z = first_zones_zonelist(zonelist, highest_zoneidx,
|
|
|
|
&policy->v.nodes);
|
|
|
|
return z->zone ? z->zone->node : node;
|
2008-04-28 03:12:17 -06:00
|
|
|
}
|
2006-01-18 18:42:36 -07:00
|
|
|
|
|
|
|
default:
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
BUG();
|
2006-01-18 18:42:36 -07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-05-19 18:11:43 -06:00
|
|
|
/*
|
|
|
|
* Do static interleaving for a VMA with known offset @n. Returns the n'th
|
|
|
|
* node in pol->v.nodes (starting from n=0), wrapping around if n exceeds the
|
|
|
|
* number of present nodes.
|
|
|
|
*/
|
2005-04-16 16:20:36 -06:00
|
|
|
static unsigned offset_il_node(struct mempolicy *pol,
|
2016-05-19 18:11:43 -06:00
|
|
|
struct vm_area_struct *vma, unsigned long n)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2005-10-29 19:15:48 -06:00
|
|
|
unsigned nnodes = nodes_weight(pol->v.nodes);
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
unsigned target;
|
2016-05-19 18:11:43 -06:00
|
|
|
int i;
|
|
|
|
int nid;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
if (!nnodes)
|
|
|
|
return numa_node_id();
|
2016-05-19 18:11:43 -06:00
|
|
|
target = (unsigned int)n % nnodes;
|
|
|
|
nid = first_node(pol->v.nodes);
|
|
|
|
for (i = 0; i < target; i++)
|
2005-10-29 19:15:48 -06:00
|
|
|
nid = next_node(nid, pol->v.nodes);
|
2005-04-16 16:20:36 -06:00
|
|
|
return nid;
|
|
|
|
}
|
|
|
|
|
2006-01-06 01:10:46 -07:00
|
|
|
/* Determine a node number for interleave */
|
|
|
|
static inline unsigned interleave_nid(struct mempolicy *pol,
|
|
|
|
struct vm_area_struct *vma, unsigned long addr, int shift)
|
|
|
|
{
|
|
|
|
if (vma) {
|
|
|
|
unsigned long off;
|
|
|
|
|
2006-08-31 22:27:53 -06:00
|
|
|
/*
|
|
|
|
* for small pages, there is no difference between
|
|
|
|
* shift and PAGE_SHIFT, so the bit-shift is safe.
|
|
|
|
* for huge pages, since vm_pgoff is in units of small
|
|
|
|
* pages, we need to shift off the always 0 bits to get
|
|
|
|
* a useful offset.
|
|
|
|
*/
|
|
|
|
BUG_ON(shift < PAGE_SHIFT);
|
|
|
|
off = vma->vm_pgoff >> (shift - PAGE_SHIFT);
|
2006-01-06 01:10:46 -07:00
|
|
|
off += (addr - vma->vm_start) >> shift;
|
|
|
|
return offset_il_node(pol, vma, off);
|
|
|
|
} else
|
|
|
|
return interleave_nodes(pol);
|
|
|
|
}
|
|
|
|
|
2006-02-03 13:51:14 -07:00
|
|
|
#ifdef CONFIG_HUGETLBFS
|
Fix NUMA Memory Policy Reference Counting
This patch proposes fixes to the reference counting of memory policy in the
page allocation paths and in show_numa_map(). Extracted from my "Memory
Policy Cleanups and Enhancements" series as stand-alone.
Shared policy lookup [shmem] has always added a reference to the policy,
but this was never unrefed after page allocation or after formatting the
numa map data.
Default system policy should not require additional ref counting, nor
should the current task's task policy. However, show_numa_map() calls
get_vma_policy() to examine what may be [likely is] another task's policy.
The latter case needs protection against freeing of the policy.
This patch adds a reference count to a mempolicy returned by
get_vma_policy() when the policy is a vma policy or another task's
mempolicy. Again, shared policy is already reference counted on lookup. A
matching "unref" [__mpol_free()] is performed in alloc_page_vma() for
shared and vma policies, and in show_numa_map() for shared and another
task's mempolicy. We can call __mpol_free() directly, saving an admittedly
inexpensive inline NULL test, because we know we have a non-NULL policy.
Handling policy ref counts for hugepages is a bit trickier.
huge_zonelist() returns a zone list that might come from a shared or vma
'BIND policy. In this case, we should hold the reference until after the
huge page allocation in dequeue_hugepage(). The patch modifies
huge_zonelist() to return a pointer to the mempolicy if it needs to be
unref'd after allocation.
Kernel Build [16cpu, 32GB, ia64] - average of 10 runs:
w/o patch w/ refcount patch
Avg Std Devn Avg Std Devn
Real: 100.59 0.38 100.63 0.43
User: 1209.60 0.37 1209.91 0.31
System: 81.52 0.42 81.64 0.34
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Andi Kleen <ak@suse.de>
Cc: Christoph Lameter <clameter@sgi.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-09-18 23:46:47 -06:00
|
|
|
/*
|
|
|
|
* huge_zonelist(@vma, @addr, @gfp_flags, @mpol)
|
2014-06-04 17:08:18 -06:00
|
|
|
* @vma: virtual memory area whose policy is sought
|
|
|
|
* @addr: address in @vma for shared policy lookup and interleave policy
|
|
|
|
* @gfp_flags: for requested zone
|
|
|
|
* @mpol: pointer to mempolicy pointer for reference counted mempolicy
|
|
|
|
* @nodemask: pointer to nodemask pointer for MPOL_BIND nodemask
|
Fix NUMA Memory Policy Reference Counting
This patch proposes fixes to the reference counting of memory policy in the
page allocation paths and in show_numa_map(). Extracted from my "Memory
Policy Cleanups and Enhancements" series as stand-alone.
Shared policy lookup [shmem] has always added a reference to the policy,
but this was never unrefed after page allocation or after formatting the
numa map data.
Default system policy should not require additional ref counting, nor
should the current task's task policy. However, show_numa_map() calls
get_vma_policy() to examine what may be [likely is] another task's policy.
The latter case needs protection against freeing of the policy.
This patch adds a reference count to a mempolicy returned by
get_vma_policy() when the policy is a vma policy or another task's
mempolicy. Again, shared policy is already reference counted on lookup. A
matching "unref" [__mpol_free()] is performed in alloc_page_vma() for
shared and vma policies, and in show_numa_map() for shared and another
task's mempolicy. We can call __mpol_free() directly, saving an admittedly
inexpensive inline NULL test, because we know we have a non-NULL policy.
Handling policy ref counts for hugepages is a bit trickier.
huge_zonelist() returns a zone list that might come from a shared or vma
'BIND policy. In this case, we should hold the reference until after the
huge page allocation in dequeue_hugepage(). The patch modifies
huge_zonelist() to return a pointer to the mempolicy if it needs to be
unref'd after allocation.
Kernel Build [16cpu, 32GB, ia64] - average of 10 runs:
w/o patch w/ refcount patch
Avg Std Devn Avg Std Devn
Real: 100.59 0.38 100.63 0.43
User: 1209.60 0.37 1209.91 0.31
System: 81.52 0.42 81.64 0.34
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Andi Kleen <ak@suse.de>
Cc: Christoph Lameter <clameter@sgi.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-09-18 23:46:47 -06:00
|
|
|
*
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
* Returns a zonelist suitable for a huge page allocation and a pointer
|
|
|
|
* to the struct mempolicy for conditional unref after allocation.
|
|
|
|
* If the effective policy is 'BIND, returns a pointer to the mempolicy's
|
|
|
|
* @nodemask for filtering the zonelist.
|
2010-05-24 15:32:08 -06:00
|
|
|
*
|
2014-04-03 15:47:24 -06:00
|
|
|
* Must be protected by read_mems_allowed_begin()
|
Fix NUMA Memory Policy Reference Counting
This patch proposes fixes to the reference counting of memory policy in the
page allocation paths and in show_numa_map(). Extracted from my "Memory
Policy Cleanups and Enhancements" series as stand-alone.
Shared policy lookup [shmem] has always added a reference to the policy,
but this was never unrefed after page allocation or after formatting the
numa map data.
Default system policy should not require additional ref counting, nor
should the current task's task policy. However, show_numa_map() calls
get_vma_policy() to examine what may be [likely is] another task's policy.
The latter case needs protection against freeing of the policy.
This patch adds a reference count to a mempolicy returned by
get_vma_policy() when the policy is a vma policy or another task's
mempolicy. Again, shared policy is already reference counted on lookup. A
matching "unref" [__mpol_free()] is performed in alloc_page_vma() for
shared and vma policies, and in show_numa_map() for shared and another
task's mempolicy. We can call __mpol_free() directly, saving an admittedly
inexpensive inline NULL test, because we know we have a non-NULL policy.
Handling policy ref counts for hugepages is a bit trickier.
huge_zonelist() returns a zone list that might come from a shared or vma
'BIND policy. In this case, we should hold the reference until after the
huge page allocation in dequeue_hugepage(). The patch modifies
huge_zonelist() to return a pointer to the mempolicy if it needs to be
unref'd after allocation.
Kernel Build [16cpu, 32GB, ia64] - average of 10 runs:
w/o patch w/ refcount patch
Avg Std Devn Avg Std Devn
Real: 100.59 0.38 100.63 0.43
User: 1209.60 0.37 1209.91 0.31
System: 81.52 0.42 81.64 0.34
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Andi Kleen <ak@suse.de>
Cc: Christoph Lameter <clameter@sgi.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-09-18 23:46:47 -06:00
|
|
|
*/
|
2007-07-17 05:03:13 -06:00
|
|
|
struct zonelist *huge_zonelist(struct vm_area_struct *vma, unsigned long addr,
|
2008-04-28 03:12:18 -06:00
|
|
|
gfp_t gfp_flags, struct mempolicy **mpol,
|
|
|
|
nodemask_t **nodemask)
|
2006-01-06 01:10:46 -07:00
|
|
|
{
|
Fix NUMA Memory Policy Reference Counting
This patch proposes fixes to the reference counting of memory policy in the
page allocation paths and in show_numa_map(). Extracted from my "Memory
Policy Cleanups and Enhancements" series as stand-alone.
Shared policy lookup [shmem] has always added a reference to the policy,
but this was never unrefed after page allocation or after formatting the
numa map data.
Default system policy should not require additional ref counting, nor
should the current task's task policy. However, show_numa_map() calls
get_vma_policy() to examine what may be [likely is] another task's policy.
The latter case needs protection against freeing of the policy.
This patch adds a reference count to a mempolicy returned by
get_vma_policy() when the policy is a vma policy or another task's
mempolicy. Again, shared policy is already reference counted on lookup. A
matching "unref" [__mpol_free()] is performed in alloc_page_vma() for
shared and vma policies, and in show_numa_map() for shared and another
task's mempolicy. We can call __mpol_free() directly, saving an admittedly
inexpensive inline NULL test, because we know we have a non-NULL policy.
Handling policy ref counts for hugepages is a bit trickier.
huge_zonelist() returns a zone list that might come from a shared or vma
'BIND policy. In this case, we should hold the reference until after the
huge page allocation in dequeue_hugepage(). The patch modifies
huge_zonelist() to return a pointer to the mempolicy if it needs to be
unref'd after allocation.
Kernel Build [16cpu, 32GB, ia64] - average of 10 runs:
w/o patch w/ refcount patch
Avg Std Devn Avg Std Devn
Real: 100.59 0.38 100.63 0.43
User: 1209.60 0.37 1209.91 0.31
System: 81.52 0.42 81.64 0.34
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Andi Kleen <ak@suse.de>
Cc: Christoph Lameter <clameter@sgi.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-09-18 23:46:47 -06:00
|
|
|
struct zonelist *zl;
|
2006-01-06 01:10:46 -07:00
|
|
|
|
2014-10-09 16:27:57 -06:00
|
|
|
*mpol = get_vma_policy(vma, addr);
|
2008-04-28 03:12:18 -06:00
|
|
|
*nodemask = NULL; /* assume !MPOL_BIND */
|
2006-01-06 01:10:46 -07:00
|
|
|
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
if (unlikely((*mpol)->mode == MPOL_INTERLEAVE)) {
|
|
|
|
zl = node_zonelist(interleave_nid(*mpol, vma, addr,
|
2008-07-23 22:27:41 -06:00
|
|
|
huge_page_shift(hstate_vma(vma))), gfp_flags);
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
} else {
|
2011-03-04 18:36:29 -07:00
|
|
|
zl = policy_zonelist(gfp_flags, *mpol, numa_node_id());
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
if ((*mpol)->mode == MPOL_BIND)
|
|
|
|
*nodemask = &(*mpol)->v.nodes;
|
Fix NUMA Memory Policy Reference Counting
This patch proposes fixes to the reference counting of memory policy in the
page allocation paths and in show_numa_map(). Extracted from my "Memory
Policy Cleanups and Enhancements" series as stand-alone.
Shared policy lookup [shmem] has always added a reference to the policy,
but this was never unrefed after page allocation or after formatting the
numa map data.
Default system policy should not require additional ref counting, nor
should the current task's task policy. However, show_numa_map() calls
get_vma_policy() to examine what may be [likely is] another task's policy.
The latter case needs protection against freeing of the policy.
This patch adds a reference count to a mempolicy returned by
get_vma_policy() when the policy is a vma policy or another task's
mempolicy. Again, shared policy is already reference counted on lookup. A
matching "unref" [__mpol_free()] is performed in alloc_page_vma() for
shared and vma policies, and in show_numa_map() for shared and another
task's mempolicy. We can call __mpol_free() directly, saving an admittedly
inexpensive inline NULL test, because we know we have a non-NULL policy.
Handling policy ref counts for hugepages is a bit trickier.
huge_zonelist() returns a zone list that might come from a shared or vma
'BIND policy. In this case, we should hold the reference until after the
huge page allocation in dequeue_hugepage(). The patch modifies
huge_zonelist() to return a pointer to the mempolicy if it needs to be
unref'd after allocation.
Kernel Build [16cpu, 32GB, ia64] - average of 10 runs:
w/o patch w/ refcount patch
Avg Std Devn Avg Std Devn
Real: 100.59 0.38 100.63 0.43
User: 1209.60 0.37 1209.91 0.31
System: 81.52 0.42 81.64 0.34
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Andi Kleen <ak@suse.de>
Cc: Christoph Lameter <clameter@sgi.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-09-18 23:46:47 -06:00
|
|
|
}
|
|
|
|
return zl;
|
2006-01-06 01:10:46 -07:00
|
|
|
}
|
hugetlb: derive huge pages nodes allowed from task mempolicy
This patch derives a "nodes_allowed" node mask from the numa mempolicy of
the task modifying the number of persistent huge pages to control the
allocation, freeing and adjusting of surplus huge pages when the pool page
count is modified via the new sysctl or sysfs attribute
"nr_hugepages_mempolicy". The nodes_allowed mask is derived as follows:
* For "default" [NULL] task mempolicy, a NULL nodemask_t pointer
is produced. This will cause the hugetlb subsystem to use
node_online_map as the "nodes_allowed". This preserves the
behavior before this patch.
* For "preferred" mempolicy, including explicit local allocation,
a nodemask with the single preferred node will be produced.
"local" policy will NOT track any internode migrations of the
task adjusting nr_hugepages.
* For "bind" and "interleave" policy, the mempolicy's nodemask
will be used.
* Other than to inform the construction of the nodes_allowed node
mask, the actual mempolicy mode is ignored. That is, all modes
behave like interleave over the resulting nodes_allowed mask
with no "fallback".
See the updated documentation [next patch] for more information
about the implications of this patch.
Examples:
Starting with:
Node 0 HugePages_Total: 0
Node 1 HugePages_Total: 0
Node 2 HugePages_Total: 0
Node 3 HugePages_Total: 0
Default behavior [with or without this patch] balances persistent
hugepage allocation across nodes [with sufficient contiguous memory]:
sysctl vm.nr_hugepages[_mempolicy]=32
yields:
Node 0 HugePages_Total: 8
Node 1 HugePages_Total: 8
Node 2 HugePages_Total: 8
Node 3 HugePages_Total: 8
Of course, we only have nr_hugepages_mempolicy with the patch,
but with default mempolicy, nr_hugepages_mempolicy behaves the
same as nr_hugepages.
Applying mempolicy--e.g., with numactl [using '-m' a.k.a.
'--membind' because it allows multiple nodes to be specified
and it's easy to type]--we can allocate huge pages on
individual nodes or sets of nodes. So, starting from the
condition above, with 8 huge pages per node, add 8 more to
node 2 using:
numactl -m 2 sysctl vm.nr_hugepages_mempolicy=40
This yields:
Node 0 HugePages_Total: 8
Node 1 HugePages_Total: 8
Node 2 HugePages_Total: 16
Node 3 HugePages_Total: 8
The incremental 8 huge pages were restricted to node 2 by the
specified mempolicy.
Similarly, we can use mempolicy to free persistent huge pages
from specified nodes:
numactl -m 0,1 sysctl vm.nr_hugepages_mempolicy=32
yields:
Node 0 HugePages_Total: 4
Node 1 HugePages_Total: 4
Node 2 HugePages_Total: 16
Node 3 HugePages_Total: 8
The 8 huge pages freed were balanced over nodes 0 and 1.
[rientjes@google.com: accomodate reworked NODEMASK_ALLOC]
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Andi Kleen <andi@firstfloor.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Nishanth Aravamudan <nacc@us.ibm.com>
Cc: Adam Litke <agl@us.ibm.com>
Cc: Andy Whitcroft <apw@canonical.com>
Cc: Eric Whitney <eric.whitney@hp.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-14 18:58:21 -07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* init_nodemask_of_mempolicy
|
|
|
|
*
|
|
|
|
* If the current task's mempolicy is "default" [NULL], return 'false'
|
|
|
|
* to indicate default policy. Otherwise, extract the policy nodemask
|
|
|
|
* for 'bind' or 'interleave' policy into the argument nodemask, or
|
|
|
|
* initialize the argument nodemask to contain the single node for
|
|
|
|
* 'preferred' or 'local' policy and return 'true' to indicate presence
|
|
|
|
* of non-default mempolicy.
|
|
|
|
*
|
|
|
|
* We don't bother with reference counting the mempolicy [mpol_get/put]
|
|
|
|
* because the current task is examining it's own mempolicy and a task's
|
|
|
|
* mempolicy is only ever changed by the task itself.
|
|
|
|
*
|
|
|
|
* N.B., it is the caller's responsibility to free a returned nodemask.
|
|
|
|
*/
|
|
|
|
bool init_nodemask_of_mempolicy(nodemask_t *mask)
|
|
|
|
{
|
|
|
|
struct mempolicy *mempolicy;
|
|
|
|
int nid;
|
|
|
|
|
|
|
|
if (!(mask && current->mempolicy))
|
|
|
|
return false;
|
|
|
|
|
2010-05-24 15:32:08 -06:00
|
|
|
task_lock(current);
|
hugetlb: derive huge pages nodes allowed from task mempolicy
This patch derives a "nodes_allowed" node mask from the numa mempolicy of
the task modifying the number of persistent huge pages to control the
allocation, freeing and adjusting of surplus huge pages when the pool page
count is modified via the new sysctl or sysfs attribute
"nr_hugepages_mempolicy". The nodes_allowed mask is derived as follows:
* For "default" [NULL] task mempolicy, a NULL nodemask_t pointer
is produced. This will cause the hugetlb subsystem to use
node_online_map as the "nodes_allowed". This preserves the
behavior before this patch.
* For "preferred" mempolicy, including explicit local allocation,
a nodemask with the single preferred node will be produced.
"local" policy will NOT track any internode migrations of the
task adjusting nr_hugepages.
* For "bind" and "interleave" policy, the mempolicy's nodemask
will be used.
* Other than to inform the construction of the nodes_allowed node
mask, the actual mempolicy mode is ignored. That is, all modes
behave like interleave over the resulting nodes_allowed mask
with no "fallback".
See the updated documentation [next patch] for more information
about the implications of this patch.
Examples:
Starting with:
Node 0 HugePages_Total: 0
Node 1 HugePages_Total: 0
Node 2 HugePages_Total: 0
Node 3 HugePages_Total: 0
Default behavior [with or without this patch] balances persistent
hugepage allocation across nodes [with sufficient contiguous memory]:
sysctl vm.nr_hugepages[_mempolicy]=32
yields:
Node 0 HugePages_Total: 8
Node 1 HugePages_Total: 8
Node 2 HugePages_Total: 8
Node 3 HugePages_Total: 8
Of course, we only have nr_hugepages_mempolicy with the patch,
but with default mempolicy, nr_hugepages_mempolicy behaves the
same as nr_hugepages.
Applying mempolicy--e.g., with numactl [using '-m' a.k.a.
'--membind' because it allows multiple nodes to be specified
and it's easy to type]--we can allocate huge pages on
individual nodes or sets of nodes. So, starting from the
condition above, with 8 huge pages per node, add 8 more to
node 2 using:
numactl -m 2 sysctl vm.nr_hugepages_mempolicy=40
This yields:
Node 0 HugePages_Total: 8
Node 1 HugePages_Total: 8
Node 2 HugePages_Total: 16
Node 3 HugePages_Total: 8
The incremental 8 huge pages were restricted to node 2 by the
specified mempolicy.
Similarly, we can use mempolicy to free persistent huge pages
from specified nodes:
numactl -m 0,1 sysctl vm.nr_hugepages_mempolicy=32
yields:
Node 0 HugePages_Total: 4
Node 1 HugePages_Total: 4
Node 2 HugePages_Total: 16
Node 3 HugePages_Total: 8
The 8 huge pages freed were balanced over nodes 0 and 1.
[rientjes@google.com: accomodate reworked NODEMASK_ALLOC]
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Andi Kleen <andi@firstfloor.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Nishanth Aravamudan <nacc@us.ibm.com>
Cc: Adam Litke <agl@us.ibm.com>
Cc: Andy Whitcroft <apw@canonical.com>
Cc: Eric Whitney <eric.whitney@hp.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-14 18:58:21 -07:00
|
|
|
mempolicy = current->mempolicy;
|
|
|
|
switch (mempolicy->mode) {
|
|
|
|
case MPOL_PREFERRED:
|
|
|
|
if (mempolicy->flags & MPOL_F_LOCAL)
|
|
|
|
nid = numa_node_id();
|
|
|
|
else
|
|
|
|
nid = mempolicy->v.preferred_node;
|
|
|
|
init_nodemask_of_node(mask, nid);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case MPOL_BIND:
|
|
|
|
/* Fall through */
|
|
|
|
case MPOL_INTERLEAVE:
|
|
|
|
*mask = mempolicy->v.nodes;
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
2010-05-24 15:32:08 -06:00
|
|
|
task_unlock(current);
|
hugetlb: derive huge pages nodes allowed from task mempolicy
This patch derives a "nodes_allowed" node mask from the numa mempolicy of
the task modifying the number of persistent huge pages to control the
allocation, freeing and adjusting of surplus huge pages when the pool page
count is modified via the new sysctl or sysfs attribute
"nr_hugepages_mempolicy". The nodes_allowed mask is derived as follows:
* For "default" [NULL] task mempolicy, a NULL nodemask_t pointer
is produced. This will cause the hugetlb subsystem to use
node_online_map as the "nodes_allowed". This preserves the
behavior before this patch.
* For "preferred" mempolicy, including explicit local allocation,
a nodemask with the single preferred node will be produced.
"local" policy will NOT track any internode migrations of the
task adjusting nr_hugepages.
* For "bind" and "interleave" policy, the mempolicy's nodemask
will be used.
* Other than to inform the construction of the nodes_allowed node
mask, the actual mempolicy mode is ignored. That is, all modes
behave like interleave over the resulting nodes_allowed mask
with no "fallback".
See the updated documentation [next patch] for more information
about the implications of this patch.
Examples:
Starting with:
Node 0 HugePages_Total: 0
Node 1 HugePages_Total: 0
Node 2 HugePages_Total: 0
Node 3 HugePages_Total: 0
Default behavior [with or without this patch] balances persistent
hugepage allocation across nodes [with sufficient contiguous memory]:
sysctl vm.nr_hugepages[_mempolicy]=32
yields:
Node 0 HugePages_Total: 8
Node 1 HugePages_Total: 8
Node 2 HugePages_Total: 8
Node 3 HugePages_Total: 8
Of course, we only have nr_hugepages_mempolicy with the patch,
but with default mempolicy, nr_hugepages_mempolicy behaves the
same as nr_hugepages.
Applying mempolicy--e.g., with numactl [using '-m' a.k.a.
'--membind' because it allows multiple nodes to be specified
and it's easy to type]--we can allocate huge pages on
individual nodes or sets of nodes. So, starting from the
condition above, with 8 huge pages per node, add 8 more to
node 2 using:
numactl -m 2 sysctl vm.nr_hugepages_mempolicy=40
This yields:
Node 0 HugePages_Total: 8
Node 1 HugePages_Total: 8
Node 2 HugePages_Total: 16
Node 3 HugePages_Total: 8
The incremental 8 huge pages were restricted to node 2 by the
specified mempolicy.
Similarly, we can use mempolicy to free persistent huge pages
from specified nodes:
numactl -m 0,1 sysctl vm.nr_hugepages_mempolicy=32
yields:
Node 0 HugePages_Total: 4
Node 1 HugePages_Total: 4
Node 2 HugePages_Total: 16
Node 3 HugePages_Total: 8
The 8 huge pages freed were balanced over nodes 0 and 1.
[rientjes@google.com: accomodate reworked NODEMASK_ALLOC]
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Andi Kleen <andi@firstfloor.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Nishanth Aravamudan <nacc@us.ibm.com>
Cc: Adam Litke <agl@us.ibm.com>
Cc: Andy Whitcroft <apw@canonical.com>
Cc: Eric Whitney <eric.whitney@hp.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-12-14 18:58:21 -07:00
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
2006-02-03 13:51:14 -07:00
|
|
|
#endif
|
2006-01-06 01:10:46 -07:00
|
|
|
|
2010-08-09 18:18:52 -06:00
|
|
|
/*
|
|
|
|
* mempolicy_nodemask_intersects
|
|
|
|
*
|
|
|
|
* If tsk's mempolicy is "default" [NULL], return 'true' to indicate default
|
|
|
|
* policy. Otherwise, check for intersection between mask and the policy
|
|
|
|
* nodemask for 'bind' or 'interleave' policy. For 'perferred' or 'local'
|
|
|
|
* policy, always return true since it may allocate elsewhere on fallback.
|
|
|
|
*
|
|
|
|
* Takes task_lock(tsk) to prevent freeing of its mempolicy.
|
|
|
|
*/
|
|
|
|
bool mempolicy_nodemask_intersects(struct task_struct *tsk,
|
|
|
|
const nodemask_t *mask)
|
|
|
|
{
|
|
|
|
struct mempolicy *mempolicy;
|
|
|
|
bool ret = true;
|
|
|
|
|
|
|
|
if (!mask)
|
|
|
|
return ret;
|
|
|
|
task_lock(tsk);
|
|
|
|
mempolicy = tsk->mempolicy;
|
|
|
|
if (!mempolicy)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
switch (mempolicy->mode) {
|
|
|
|
case MPOL_PREFERRED:
|
|
|
|
/*
|
|
|
|
* MPOL_PREFERRED and MPOL_F_LOCAL are only preferred nodes to
|
|
|
|
* allocate from, they may fallback to other nodes when oom.
|
|
|
|
* Thus, it's possible for tsk to have allocated memory from
|
|
|
|
* nodes in mask.
|
|
|
|
*/
|
|
|
|
break;
|
|
|
|
case MPOL_BIND:
|
|
|
|
case MPOL_INTERLEAVE:
|
|
|
|
ret = nodes_intersects(mempolicy->v.nodes, *mask);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
out:
|
|
|
|
task_unlock(tsk);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/* Allocate a page in interleaved policy.
|
|
|
|
Own path because it needs to do special accounting. */
|
2005-10-29 19:15:49 -06:00
|
|
|
static struct page *alloc_page_interleave(gfp_t gfp, unsigned order,
|
|
|
|
unsigned nid)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct zonelist *zl;
|
|
|
|
struct page *page;
|
|
|
|
|
2008-04-28 03:12:14 -06:00
|
|
|
zl = node_zonelist(nid, gfp);
|
2005-04-16 16:20:36 -06:00
|
|
|
page = __alloc_pages(gfp, order, zl);
|
2008-04-28 03:12:17 -06:00
|
|
|
if (page && page_zone(page) == zonelist_zone(&zl->_zonerefs[0]))
|
2006-06-30 02:55:44 -06:00
|
|
|
inc_zone_page_state(page, NUMA_INTERLEAVE_HIT);
|
2005-04-16 16:20:36 -06:00
|
|
|
return page;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2011-01-13 16:47:05 -07:00
|
|
|
* alloc_pages_vma - Allocate a page for a VMA.
|
2005-04-16 16:20:36 -06:00
|
|
|
*
|
|
|
|
* @gfp:
|
|
|
|
* %GFP_USER user allocation.
|
|
|
|
* %GFP_KERNEL kernel allocations,
|
|
|
|
* %GFP_HIGHMEM highmem/user allocations,
|
|
|
|
* %GFP_FS allocation should not call back into a file system.
|
|
|
|
* %GFP_ATOMIC don't sleep.
|
|
|
|
*
|
2011-01-13 16:47:05 -07:00
|
|
|
* @order:Order of the GFP allocation.
|
2005-04-16 16:20:36 -06:00
|
|
|
* @vma: Pointer to VMA or NULL if not available.
|
|
|
|
* @addr: Virtual Address of the allocation. Must be inside the VMA.
|
2015-02-11 16:27:15 -07:00
|
|
|
* @node: Which node to prefer for allocation (modulo policy).
|
|
|
|
* @hugepage: for hugepages try only the preferred node if possible
|
2005-04-16 16:20:36 -06:00
|
|
|
*
|
|
|
|
* This function allocates a page from the kernel page pool and applies
|
|
|
|
* a NUMA policy associated with the VMA or the current process.
|
|
|
|
* When VMA is not NULL caller must hold down_read on the mmap_sem of the
|
|
|
|
* mm_struct of the VMA to prevent it from going away. Should be used for
|
2015-02-11 16:27:15 -07:00
|
|
|
* all allocations for pages that will be mapped into user space. Returns
|
|
|
|
* NULL when no page can be allocated.
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
|
|
|
struct page *
|
2011-01-13 16:47:05 -07:00
|
|
|
alloc_pages_vma(gfp_t gfp, int order, struct vm_area_struct *vma,
|
2015-02-11 16:27:15 -07:00
|
|
|
unsigned long addr, int node, bool hugepage)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 17:34:11 -06:00
|
|
|
struct mempolicy *pol;
|
2010-05-24 15:32:08 -06:00
|
|
|
struct page *page;
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 17:34:11 -06:00
|
|
|
unsigned int cpuset_mems_cookie;
|
2015-02-11 16:27:15 -07:00
|
|
|
struct zonelist *zl;
|
|
|
|
nodemask_t *nmask;
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 17:34:11 -06:00
|
|
|
|
|
|
|
retry_cpuset:
|
2014-10-09 16:27:57 -06:00
|
|
|
pol = get_vma_policy(vma, addr);
|
2014-04-03 15:47:24 -06:00
|
|
|
cpuset_mems_cookie = read_mems_allowed_begin();
|
2005-04-16 16:20:36 -06:00
|
|
|
|
mm, thp: respect MPOL_PREFERRED policy with non-local node
Since commit 077fcf116c8c ("mm/thp: allocate transparent hugepages on
local node"), we handle THP allocations on page fault in a special way -
for non-interleave memory policies, the allocation is only attempted on
the node local to the current CPU, if the policy's nodemask allows the
node.
This is motivated by the assumption that THP benefits cannot offset the
cost of remote accesses, so it's better to fallback to base pages on the
local node (which might still be available, while huge pages are not due
to fragmentation) than to allocate huge pages on a remote node.
The nodemask check prevents us from violating e.g. MPOL_BIND policies
where the local node is not among the allowed nodes. However, the
current implementation can still give surprising results for the
MPOL_PREFERRED policy when the preferred node is different than the
current CPU's local node.
In such case we should honor the preferred node and not use the local
node, which is what this patch does. If hugepage allocation on the
preferred node fails, we fall back to base pages and don't try other
nodes, with the same motivation as is done for the local node hugepage
allocations. The patch also moves the MPOL_INTERLEAVE check around to
simplify the hugepage specific test.
The difference can be demonstrated using in-tree transhuge-stress test
on the following 2-node machine where half memory on one node was
occupied to show the difference.
> numactl --hardware
available: 2 nodes (0-1)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 24 25 26 27 28 29 30 31 32 33 34 35
node 0 size: 7878 MB
node 0 free: 3623 MB
node 1 cpus: 12 13 14 15 16 17 18 19 20 21 22 23 36 37 38 39 40 41 42 43 44 45 46 47
node 1 size: 8045 MB
node 1 free: 7818 MB
node distances:
node 0 1
0: 10 21
1: 21 10
Before the patch:
> numactl -p0 -C0 ./transhuge-stress
transhuge-stress: 2.197 s/loop, 0.276 ms/page, 7249.168 MiB/s 7962 succeed, 0 failed, 1786 different pages
> numactl -p0 -C12 ./transhuge-stress
transhuge-stress: 2.962 s/loop, 0.372 ms/page, 5376.172 MiB/s 7962 succeed, 0 failed, 3873 different pages
Number of successful THP allocations corresponds to free memory on node 0 in
the first case and node 1 in the second case, i.e. -p parameter is ignored and
cpu binding "wins".
After the patch:
> numactl -p0 -C0 ./transhuge-stress
transhuge-stress: 2.183 s/loop, 0.274 ms/page, 7295.516 MiB/s 7962 succeed, 0 failed, 1760 different pages
> numactl -p0 -C12 ./transhuge-stress
transhuge-stress: 2.878 s/loop, 0.361 ms/page, 5533.638 MiB/s 7962 succeed, 0 failed, 1750 different pages
> numactl -p1 -C0 ./transhuge-stress
transhuge-stress: 4.628 s/loop, 0.581 ms/page, 3440.893 MiB/s 7962 succeed, 0 failed, 3918 different pages
The -p parameter is respected regardless of cpu binding.
> numactl -C0 ./transhuge-stress
transhuge-stress: 2.202 s/loop, 0.277 ms/page, 7230.003 MiB/s 7962 succeed, 0 failed, 1750 different pages
> numactl -C12 ./transhuge-stress
transhuge-stress: 3.020 s/loop, 0.379 ms/page, 5273.324 MiB/s 7962 succeed, 0 failed, 3916 different pages
Without -p parameter, hugepage restriction to CPU-local node works as before.
Fixes: 077fcf116c8c ("mm/thp: allocate transparent hugepages on local node")
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: <stable@vger.kernel.org> [4.0+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-06-24 17:58:48 -06:00
|
|
|
if (pol->mode == MPOL_INTERLEAVE) {
|
|
|
|
unsigned nid;
|
|
|
|
|
|
|
|
nid = interleave_nid(pol, vma, addr, PAGE_SHIFT + order);
|
|
|
|
mpol_cond_put(pol);
|
|
|
|
page = alloc_page_interleave(gfp, order, nid);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (unlikely(IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE) && hugepage)) {
|
|
|
|
int hpage_node = node;
|
|
|
|
|
2015-02-11 16:27:15 -07:00
|
|
|
/*
|
|
|
|
* For hugepage allocation and non-interleave policy which
|
mm, thp: respect MPOL_PREFERRED policy with non-local node
Since commit 077fcf116c8c ("mm/thp: allocate transparent hugepages on
local node"), we handle THP allocations on page fault in a special way -
for non-interleave memory policies, the allocation is only attempted on
the node local to the current CPU, if the policy's nodemask allows the
node.
This is motivated by the assumption that THP benefits cannot offset the
cost of remote accesses, so it's better to fallback to base pages on the
local node (which might still be available, while huge pages are not due
to fragmentation) than to allocate huge pages on a remote node.
The nodemask check prevents us from violating e.g. MPOL_BIND policies
where the local node is not among the allowed nodes. However, the
current implementation can still give surprising results for the
MPOL_PREFERRED policy when the preferred node is different than the
current CPU's local node.
In such case we should honor the preferred node and not use the local
node, which is what this patch does. If hugepage allocation on the
preferred node fails, we fall back to base pages and don't try other
nodes, with the same motivation as is done for the local node hugepage
allocations. The patch also moves the MPOL_INTERLEAVE check around to
simplify the hugepage specific test.
The difference can be demonstrated using in-tree transhuge-stress test
on the following 2-node machine where half memory on one node was
occupied to show the difference.
> numactl --hardware
available: 2 nodes (0-1)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 24 25 26 27 28 29 30 31 32 33 34 35
node 0 size: 7878 MB
node 0 free: 3623 MB
node 1 cpus: 12 13 14 15 16 17 18 19 20 21 22 23 36 37 38 39 40 41 42 43 44 45 46 47
node 1 size: 8045 MB
node 1 free: 7818 MB
node distances:
node 0 1
0: 10 21
1: 21 10
Before the patch:
> numactl -p0 -C0 ./transhuge-stress
transhuge-stress: 2.197 s/loop, 0.276 ms/page, 7249.168 MiB/s 7962 succeed, 0 failed, 1786 different pages
> numactl -p0 -C12 ./transhuge-stress
transhuge-stress: 2.962 s/loop, 0.372 ms/page, 5376.172 MiB/s 7962 succeed, 0 failed, 3873 different pages
Number of successful THP allocations corresponds to free memory on node 0 in
the first case and node 1 in the second case, i.e. -p parameter is ignored and
cpu binding "wins".
After the patch:
> numactl -p0 -C0 ./transhuge-stress
transhuge-stress: 2.183 s/loop, 0.274 ms/page, 7295.516 MiB/s 7962 succeed, 0 failed, 1760 different pages
> numactl -p0 -C12 ./transhuge-stress
transhuge-stress: 2.878 s/loop, 0.361 ms/page, 5533.638 MiB/s 7962 succeed, 0 failed, 1750 different pages
> numactl -p1 -C0 ./transhuge-stress
transhuge-stress: 4.628 s/loop, 0.581 ms/page, 3440.893 MiB/s 7962 succeed, 0 failed, 3918 different pages
The -p parameter is respected regardless of cpu binding.
> numactl -C0 ./transhuge-stress
transhuge-stress: 2.202 s/loop, 0.277 ms/page, 7230.003 MiB/s 7962 succeed, 0 failed, 1750 different pages
> numactl -C12 ./transhuge-stress
transhuge-stress: 3.020 s/loop, 0.379 ms/page, 5273.324 MiB/s 7962 succeed, 0 failed, 3916 different pages
Without -p parameter, hugepage restriction to CPU-local node works as before.
Fixes: 077fcf116c8c ("mm/thp: allocate transparent hugepages on local node")
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: <stable@vger.kernel.org> [4.0+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-06-24 17:58:48 -06:00
|
|
|
* allows the current node (or other explicitly preferred
|
|
|
|
* node) we only try to allocate from the current/preferred
|
|
|
|
* node and don't fall back to other nodes, as the cost of
|
|
|
|
* remote accesses would likely offset THP benefits.
|
2015-02-11 16:27:15 -07:00
|
|
|
*
|
|
|
|
* If the policy is interleave, or does not allow the current
|
|
|
|
* node in its nodemask, we allocate the standard way.
|
|
|
|
*/
|
mm, thp: respect MPOL_PREFERRED policy with non-local node
Since commit 077fcf116c8c ("mm/thp: allocate transparent hugepages on
local node"), we handle THP allocations on page fault in a special way -
for non-interleave memory policies, the allocation is only attempted on
the node local to the current CPU, if the policy's nodemask allows the
node.
This is motivated by the assumption that THP benefits cannot offset the
cost of remote accesses, so it's better to fallback to base pages on the
local node (which might still be available, while huge pages are not due
to fragmentation) than to allocate huge pages on a remote node.
The nodemask check prevents us from violating e.g. MPOL_BIND policies
where the local node is not among the allowed nodes. However, the
current implementation can still give surprising results for the
MPOL_PREFERRED policy when the preferred node is different than the
current CPU's local node.
In such case we should honor the preferred node and not use the local
node, which is what this patch does. If hugepage allocation on the
preferred node fails, we fall back to base pages and don't try other
nodes, with the same motivation as is done for the local node hugepage
allocations. The patch also moves the MPOL_INTERLEAVE check around to
simplify the hugepage specific test.
The difference can be demonstrated using in-tree transhuge-stress test
on the following 2-node machine where half memory on one node was
occupied to show the difference.
> numactl --hardware
available: 2 nodes (0-1)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 24 25 26 27 28 29 30 31 32 33 34 35
node 0 size: 7878 MB
node 0 free: 3623 MB
node 1 cpus: 12 13 14 15 16 17 18 19 20 21 22 23 36 37 38 39 40 41 42 43 44 45 46 47
node 1 size: 8045 MB
node 1 free: 7818 MB
node distances:
node 0 1
0: 10 21
1: 21 10
Before the patch:
> numactl -p0 -C0 ./transhuge-stress
transhuge-stress: 2.197 s/loop, 0.276 ms/page, 7249.168 MiB/s 7962 succeed, 0 failed, 1786 different pages
> numactl -p0 -C12 ./transhuge-stress
transhuge-stress: 2.962 s/loop, 0.372 ms/page, 5376.172 MiB/s 7962 succeed, 0 failed, 3873 different pages
Number of successful THP allocations corresponds to free memory on node 0 in
the first case and node 1 in the second case, i.e. -p parameter is ignored and
cpu binding "wins".
After the patch:
> numactl -p0 -C0 ./transhuge-stress
transhuge-stress: 2.183 s/loop, 0.274 ms/page, 7295.516 MiB/s 7962 succeed, 0 failed, 1760 different pages
> numactl -p0 -C12 ./transhuge-stress
transhuge-stress: 2.878 s/loop, 0.361 ms/page, 5533.638 MiB/s 7962 succeed, 0 failed, 1750 different pages
> numactl -p1 -C0 ./transhuge-stress
transhuge-stress: 4.628 s/loop, 0.581 ms/page, 3440.893 MiB/s 7962 succeed, 0 failed, 3918 different pages
The -p parameter is respected regardless of cpu binding.
> numactl -C0 ./transhuge-stress
transhuge-stress: 2.202 s/loop, 0.277 ms/page, 7230.003 MiB/s 7962 succeed, 0 failed, 1750 different pages
> numactl -C12 ./transhuge-stress
transhuge-stress: 3.020 s/loop, 0.379 ms/page, 5273.324 MiB/s 7962 succeed, 0 failed, 3916 different pages
Without -p parameter, hugepage restriction to CPU-local node works as before.
Fixes: 077fcf116c8c ("mm/thp: allocate transparent hugepages on local node")
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: <stable@vger.kernel.org> [4.0+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-06-24 17:58:48 -06:00
|
|
|
if (pol->mode == MPOL_PREFERRED &&
|
|
|
|
!(pol->flags & MPOL_F_LOCAL))
|
|
|
|
hpage_node = pol->v.preferred_node;
|
|
|
|
|
2015-02-11 16:27:15 -07:00
|
|
|
nmask = policy_nodemask(gfp, pol);
|
mm, thp: respect MPOL_PREFERRED policy with non-local node
Since commit 077fcf116c8c ("mm/thp: allocate transparent hugepages on
local node"), we handle THP allocations on page fault in a special way -
for non-interleave memory policies, the allocation is only attempted on
the node local to the current CPU, if the policy's nodemask allows the
node.
This is motivated by the assumption that THP benefits cannot offset the
cost of remote accesses, so it's better to fallback to base pages on the
local node (which might still be available, while huge pages are not due
to fragmentation) than to allocate huge pages on a remote node.
The nodemask check prevents us from violating e.g. MPOL_BIND policies
where the local node is not among the allowed nodes. However, the
current implementation can still give surprising results for the
MPOL_PREFERRED policy when the preferred node is different than the
current CPU's local node.
In such case we should honor the preferred node and not use the local
node, which is what this patch does. If hugepage allocation on the
preferred node fails, we fall back to base pages and don't try other
nodes, with the same motivation as is done for the local node hugepage
allocations. The patch also moves the MPOL_INTERLEAVE check around to
simplify the hugepage specific test.
The difference can be demonstrated using in-tree transhuge-stress test
on the following 2-node machine where half memory on one node was
occupied to show the difference.
> numactl --hardware
available: 2 nodes (0-1)
node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 24 25 26 27 28 29 30 31 32 33 34 35
node 0 size: 7878 MB
node 0 free: 3623 MB
node 1 cpus: 12 13 14 15 16 17 18 19 20 21 22 23 36 37 38 39 40 41 42 43 44 45 46 47
node 1 size: 8045 MB
node 1 free: 7818 MB
node distances:
node 0 1
0: 10 21
1: 21 10
Before the patch:
> numactl -p0 -C0 ./transhuge-stress
transhuge-stress: 2.197 s/loop, 0.276 ms/page, 7249.168 MiB/s 7962 succeed, 0 failed, 1786 different pages
> numactl -p0 -C12 ./transhuge-stress
transhuge-stress: 2.962 s/loop, 0.372 ms/page, 5376.172 MiB/s 7962 succeed, 0 failed, 3873 different pages
Number of successful THP allocations corresponds to free memory on node 0 in
the first case and node 1 in the second case, i.e. -p parameter is ignored and
cpu binding "wins".
After the patch:
> numactl -p0 -C0 ./transhuge-stress
transhuge-stress: 2.183 s/loop, 0.274 ms/page, 7295.516 MiB/s 7962 succeed, 0 failed, 1760 different pages
> numactl -p0 -C12 ./transhuge-stress
transhuge-stress: 2.878 s/loop, 0.361 ms/page, 5533.638 MiB/s 7962 succeed, 0 failed, 1750 different pages
> numactl -p1 -C0 ./transhuge-stress
transhuge-stress: 4.628 s/loop, 0.581 ms/page, 3440.893 MiB/s 7962 succeed, 0 failed, 3918 different pages
The -p parameter is respected regardless of cpu binding.
> numactl -C0 ./transhuge-stress
transhuge-stress: 2.202 s/loop, 0.277 ms/page, 7230.003 MiB/s 7962 succeed, 0 failed, 1750 different pages
> numactl -C12 ./transhuge-stress
transhuge-stress: 3.020 s/loop, 0.379 ms/page, 5273.324 MiB/s 7962 succeed, 0 failed, 3916 different pages
Without -p parameter, hugepage restriction to CPU-local node works as before.
Fixes: 077fcf116c8c ("mm/thp: allocate transparent hugepages on local node")
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: <stable@vger.kernel.org> [4.0+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-06-24 17:58:48 -06:00
|
|
|
if (!nmask || node_isset(hpage_node, *nmask)) {
|
2015-02-11 16:27:15 -07:00
|
|
|
mpol_cond_put(pol);
|
mm: rename alloc_pages_exact_node() to __alloc_pages_node()
alloc_pages_exact_node() was introduced in commit 6484eb3e2a81 ("page
allocator: do not check NUMA node ID when the caller knows the node is
valid") as an optimized variant of alloc_pages_node(), that doesn't
fallback to current node for nid == NUMA_NO_NODE. Unfortunately the
name of the function can easily suggest that the allocation is
restricted to the given node and fails otherwise. In truth, the node is
only preferred, unless __GFP_THISNODE is passed among the gfp flags.
The misleading name has lead to mistakes in the past, see for example
commits 5265047ac301 ("mm, thp: really limit transparent hugepage
allocation to local node") and b360edb43f8e ("mm, mempolicy:
migrate_to_node should only migrate to node").
Another issue with the name is that there's a family of
alloc_pages_exact*() functions where 'exact' means exact size (instead
of page order), which leads to more confusion.
To prevent further mistakes, this patch effectively renames
alloc_pages_exact_node() to __alloc_pages_node() to better convey that
it's an optimized variant of alloc_pages_node() not intended for general
usage. Both functions get described in comments.
It has been also considered to really provide a convenience function for
allocations restricted to a node, but the major opinion seems to be that
__GFP_THISNODE already provides that functionality and we shouldn't
duplicate the API needlessly. The number of users would be small
anyway.
Existing callers of alloc_pages_exact_node() are simply converted to
call __alloc_pages_node(), with the exception of sba_alloc_coherent()
which open-codes the check for NUMA_NO_NODE, so it is converted to use
alloc_pages_node() instead. This means it no longer performs some
VM_BUG_ON checks, and since the current check for nid in
alloc_pages_node() uses a 'nid < 0' comparison (which includes
NUMA_NO_NODE), it may hide wrong values which would be previously
exposed.
Both differences will be rectified by the next patch.
To sum up, this patch makes no functional changes, except temporarily
hiding potentially buggy callers. Restricting the checks in
alloc_pages_node() is left for the next patch which can in turn expose
more existing buggy callers.
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Robin Holt <robinmholt@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Christoph Lameter <cl@linux.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au>
Cc: Mel Gorman <mgorman@suse.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Gleb Natapov <gleb@kernel.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Cliff Whickman <cpw@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-09-08 16:03:50 -06:00
|
|
|
page = __alloc_pages_node(hpage_node,
|
2015-04-14 16:46:58 -06:00
|
|
|
gfp | __GFP_THISNODE, order);
|
2015-02-11 16:27:15 -07:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
nmask = policy_nodemask(gfp, pol);
|
|
|
|
zl = policy_zonelist(gfp, pol, node);
|
2014-10-09 16:27:41 -06:00
|
|
|
mpol_cond_put(pol);
|
2015-02-11 16:27:15 -07:00
|
|
|
page = __alloc_pages_nodemask(gfp, order, zl, nmask);
|
|
|
|
out:
|
2014-04-03 15:47:24 -06:00
|
|
|
if (unlikely(!page && read_mems_allowed_retry(cpuset_mems_cookie)))
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 17:34:11 -06:00
|
|
|
goto retry_cpuset;
|
2010-05-24 15:32:08 -06:00
|
|
|
return page;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* alloc_pages_current - Allocate pages.
|
|
|
|
*
|
|
|
|
* @gfp:
|
|
|
|
* %GFP_USER user allocation,
|
|
|
|
* %GFP_KERNEL kernel allocation,
|
|
|
|
* %GFP_HIGHMEM highmem allocation,
|
|
|
|
* %GFP_FS don't call back into a file system.
|
|
|
|
* %GFP_ATOMIC don't sleep.
|
|
|
|
* @order: Power of two of allocation size in pages. 0 is a single page.
|
|
|
|
*
|
|
|
|
* Allocate a page from the kernel page pool. When not in
|
|
|
|
* interrupt context and apply the current process NUMA policy.
|
|
|
|
* Returns NULL when no page can be allocated.
|
|
|
|
*
|
2006-01-08 02:01:54 -07:00
|
|
|
* Don't call cpuset_update_task_memory_state() unless
|
2005-04-16 16:20:36 -06:00
|
|
|
* 1) it's ok to take cpuset_sem (can WAIT), and
|
|
|
|
* 2) allocating for current task (not interrupt).
|
|
|
|
*/
|
2005-10-07 00:46:04 -06:00
|
|
|
struct page *alloc_pages_current(gfp_t gfp, unsigned order)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2014-10-09 16:27:45 -06:00
|
|
|
struct mempolicy *pol = &default_policy;
|
2010-05-24 15:32:08 -06:00
|
|
|
struct page *page;
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 17:34:11 -06:00
|
|
|
unsigned int cpuset_mems_cookie;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2014-10-09 16:27:45 -06:00
|
|
|
if (!in_interrupt() && !(gfp & __GFP_THISNODE))
|
|
|
|
pol = get_task_policy(current);
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 17:34:11 -06:00
|
|
|
retry_cpuset:
|
2014-04-03 15:47:24 -06:00
|
|
|
cpuset_mems_cookie = read_mems_allowed_begin();
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 17:34:11 -06:00
|
|
|
|
mempolicy: rework mempolicy Reference Counting [yet again]
After further discussion with Christoph Lameter, it has become clear that my
earlier attempts to clean up the mempolicy reference counting were a bit of
overkill in some areas, resulting in superflous ref/unref in what are usually
fast paths. In other areas, further inspection reveals that I botched the
unref for interleave policies.
A separate patch, suitable for upstream/stable trees, fixes up the known
errors in the previous attempt to fix reference counting.
This patch reworks the memory policy referencing counting and, one hopes,
simplifies the code. Maybe I'll get it right this time.
See the update to the numa_memory_policy.txt document for a discussion of
memory policy reference counting that motivates this patch.
Summary:
Lookup of mempolicy, based on (vma, address) need only add a reference for
shared policy, and we need only unref the policy when finished for shared
policies. So, this patch backs out all of the unneeded extra reference
counting added by my previous attempt. It then unrefs only shared policies
when we're finished with them, using the mpol_cond_put() [conditional put]
helper function introduced by this patch.
Note that shmem_swapin() calls read_swap_cache_async() with a dummy vma
containing just the policy. read_swap_cache_async() can call alloc_page_vma()
multiple times, so we can't let alloc_page_vma() unref the shared policy in
this case. To avoid this, we make a copy of any non-null shared policy and
remove the MPOL_F_SHARED flag from the copy. This copy occurs before reading
a page [or multiple pages] from swap, so the overhead should not be an issue
here.
I introduced a new static inline function "mpol_cond_copy()" to copy the
shared policy to an on-stack policy and remove the flags that would require a
conditional free. The current implementation of mpol_cond_copy() assumes that
the struct mempolicy contains no pointers to dynamically allocated structures
that must be duplicated or reference counted during copy.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:16 -06:00
|
|
|
/*
|
|
|
|
* No reference counting needed for current->mempolicy
|
|
|
|
* nor system default_policy
|
|
|
|
*/
|
2008-04-28 03:13:12 -06:00
|
|
|
if (pol->mode == MPOL_INTERLEAVE)
|
2010-05-24 15:32:08 -06:00
|
|
|
page = alloc_page_interleave(gfp, order, interleave_nodes(pol));
|
|
|
|
else
|
|
|
|
page = __alloc_pages_nodemask(gfp, order,
|
2011-03-04 18:36:32 -07:00
|
|
|
policy_zonelist(gfp, pol, numa_node_id()),
|
|
|
|
policy_nodemask(gfp, pol));
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 17:34:11 -06:00
|
|
|
|
2014-04-03 15:47:24 -06:00
|
|
|
if (unlikely(!page && read_mems_allowed_retry(cpuset_mems_cookie)))
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 17:34:11 -06:00
|
|
|
goto retry_cpuset;
|
|
|
|
|
2010-05-24 15:32:08 -06:00
|
|
|
return page;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(alloc_pages_current);
|
|
|
|
|
2013-09-11 15:20:14 -06:00
|
|
|
int vma_dup_policy(struct vm_area_struct *src, struct vm_area_struct *dst)
|
|
|
|
{
|
|
|
|
struct mempolicy *pol = mpol_dup(vma_policy(src));
|
|
|
|
|
|
|
|
if (IS_ERR(pol))
|
|
|
|
return PTR_ERR(pol);
|
|
|
|
dst->vm_policy = pol;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
[PATCH] cpuset: rebind vma mempolicies fix
Fix more of longstanding bug in cpuset/mempolicy interaction.
NUMA mempolicies (mm/mempolicy.c) are constrained by the current tasks cpuset
to just the Memory Nodes allowed by that cpuset. The kernel maintains
internal state for each mempolicy, tracking what nodes are used for the
MPOL_INTERLEAVE, MPOL_BIND or MPOL_PREFERRED policies.
When a tasks cpuset memory placement changes, whether because the cpuset
changed, or because the task was attached to a different cpuset, then the
tasks mempolicies have to be rebound to the new cpuset placement, so as to
preserve the cpuset-relative numbering of the nodes in that policy.
An earlier fix handled such mempolicy rebinding for mempolicies attached to a
task.
This fix rebinds mempolicies attached to vma's (address ranges in a tasks
address space.) Due to the need to hold the task->mm->mmap_sem semaphore while
updating vma's, the rebinding of vma mempolicies has to be done when the
cpuset memory placement is changed, at which time mmap_sem can be safely
acquired. The tasks mempolicy is rebound later, when the task next attempts
to allocate memory and notices that its task->cpuset_mems_generation is
out-of-date with its cpusets mems_generation.
Because walking the tasklist to find all tasks attached to a changing cpuset
requires holding tasklist_lock, a spinlock, one cannot update the vma's of the
affected tasks while doing the tasklist scan. In general, one cannot acquire
a semaphore (which can sleep) while already holding a spinlock (such as
tasklist_lock). So a list of mm references has to be built up during the
tasklist scan, then the tasklist lock dropped, then for each mm, its mmap_sem
acquired, and the vma's in that mm rebound.
Once the tasklist lock is dropped, affected tasks may fork new tasks, before
their mm's are rebound. A kernel global 'cpuset_being_rebound' is set to
point to the cpuset being rebound (there can only be one; cpuset modifications
are done under a global 'manage_sem' semaphore), and the mpol_copy code that
is used to copy a tasks mempolicies during fork catches such forking tasks,
and ensures their children are also rebound.
When a task is moved to a different cpuset, it is easier, as there is only one
task involved. It's mm->vma's are scanned, using the same
mpol_rebind_policy() as used above.
It may happen that both the mpol_copy hook and the update done via the
tasklist scan update the same mm twice. This is ok, as the mempolicies of
each vma in an mm keep track of what mems_allowed they are relative to, and
safely no-op a second request to rebind to the same nodes.
Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 02:01:59 -07:00
|
|
|
/*
|
2008-04-28 03:13:09 -06:00
|
|
|
* If mpol_dup() sees current->cpuset == cpuset_being_rebound, then it
|
[PATCH] cpuset: rebind vma mempolicies fix
Fix more of longstanding bug in cpuset/mempolicy interaction.
NUMA mempolicies (mm/mempolicy.c) are constrained by the current tasks cpuset
to just the Memory Nodes allowed by that cpuset. The kernel maintains
internal state for each mempolicy, tracking what nodes are used for the
MPOL_INTERLEAVE, MPOL_BIND or MPOL_PREFERRED policies.
When a tasks cpuset memory placement changes, whether because the cpuset
changed, or because the task was attached to a different cpuset, then the
tasks mempolicies have to be rebound to the new cpuset placement, so as to
preserve the cpuset-relative numbering of the nodes in that policy.
An earlier fix handled such mempolicy rebinding for mempolicies attached to a
task.
This fix rebinds mempolicies attached to vma's (address ranges in a tasks
address space.) Due to the need to hold the task->mm->mmap_sem semaphore while
updating vma's, the rebinding of vma mempolicies has to be done when the
cpuset memory placement is changed, at which time mmap_sem can be safely
acquired. The tasks mempolicy is rebound later, when the task next attempts
to allocate memory and notices that its task->cpuset_mems_generation is
out-of-date with its cpusets mems_generation.
Because walking the tasklist to find all tasks attached to a changing cpuset
requires holding tasklist_lock, a spinlock, one cannot update the vma's of the
affected tasks while doing the tasklist scan. In general, one cannot acquire
a semaphore (which can sleep) while already holding a spinlock (such as
tasklist_lock). So a list of mm references has to be built up during the
tasklist scan, then the tasklist lock dropped, then for each mm, its mmap_sem
acquired, and the vma's in that mm rebound.
Once the tasklist lock is dropped, affected tasks may fork new tasks, before
their mm's are rebound. A kernel global 'cpuset_being_rebound' is set to
point to the cpuset being rebound (there can only be one; cpuset modifications
are done under a global 'manage_sem' semaphore), and the mpol_copy code that
is used to copy a tasks mempolicies during fork catches such forking tasks,
and ensures their children are also rebound.
When a task is moved to a different cpuset, it is easier, as there is only one
task involved. It's mm->vma's are scanned, using the same
mpol_rebind_policy() as used above.
It may happen that both the mpol_copy hook and the update done via the
tasklist scan update the same mm twice. This is ok, as the mempolicies of
each vma in an mm keep track of what mems_allowed they are relative to, and
safely no-op a second request to rebind to the same nodes.
Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 02:01:59 -07:00
|
|
|
* rebinds the mempolicy its copying by calling mpol_rebind_policy()
|
|
|
|
* with the mems_allowed returned by cpuset_mems_allowed(). This
|
|
|
|
* keeps mempolicies cpuset relative after its cpuset moves. See
|
|
|
|
* further kernel/cpuset.c update_nodemask().
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
*
|
|
|
|
* current's mempolicy may be rebinded by the other task(the task that changes
|
|
|
|
* cpuset's mems), so we needn't do rebind work for current task.
|
[PATCH] cpuset: rebind vma mempolicies fix
Fix more of longstanding bug in cpuset/mempolicy interaction.
NUMA mempolicies (mm/mempolicy.c) are constrained by the current tasks cpuset
to just the Memory Nodes allowed by that cpuset. The kernel maintains
internal state for each mempolicy, tracking what nodes are used for the
MPOL_INTERLEAVE, MPOL_BIND or MPOL_PREFERRED policies.
When a tasks cpuset memory placement changes, whether because the cpuset
changed, or because the task was attached to a different cpuset, then the
tasks mempolicies have to be rebound to the new cpuset placement, so as to
preserve the cpuset-relative numbering of the nodes in that policy.
An earlier fix handled such mempolicy rebinding for mempolicies attached to a
task.
This fix rebinds mempolicies attached to vma's (address ranges in a tasks
address space.) Due to the need to hold the task->mm->mmap_sem semaphore while
updating vma's, the rebinding of vma mempolicies has to be done when the
cpuset memory placement is changed, at which time mmap_sem can be safely
acquired. The tasks mempolicy is rebound later, when the task next attempts
to allocate memory and notices that its task->cpuset_mems_generation is
out-of-date with its cpusets mems_generation.
Because walking the tasklist to find all tasks attached to a changing cpuset
requires holding tasklist_lock, a spinlock, one cannot update the vma's of the
affected tasks while doing the tasklist scan. In general, one cannot acquire
a semaphore (which can sleep) while already holding a spinlock (such as
tasklist_lock). So a list of mm references has to be built up during the
tasklist scan, then the tasklist lock dropped, then for each mm, its mmap_sem
acquired, and the vma's in that mm rebound.
Once the tasklist lock is dropped, affected tasks may fork new tasks, before
their mm's are rebound. A kernel global 'cpuset_being_rebound' is set to
point to the cpuset being rebound (there can only be one; cpuset modifications
are done under a global 'manage_sem' semaphore), and the mpol_copy code that
is used to copy a tasks mempolicies during fork catches such forking tasks,
and ensures their children are also rebound.
When a task is moved to a different cpuset, it is easier, as there is only one
task involved. It's mm->vma's are scanned, using the same
mpol_rebind_policy() as used above.
It may happen that both the mpol_copy hook and the update done via the
tasklist scan update the same mm twice. This is ok, as the mempolicies of
each vma in an mm keep track of what mems_allowed they are relative to, and
safely no-op a second request to rebind to the same nodes.
Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 02:01:59 -07:00
|
|
|
*/
|
|
|
|
|
2008-04-28 03:13:09 -06:00
|
|
|
/* Slow path of a mempolicy duplicate */
|
|
|
|
struct mempolicy *__mpol_dup(struct mempolicy *old)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct mempolicy *new = kmem_cache_alloc(policy_cache, GFP_KERNEL);
|
|
|
|
|
|
|
|
if (!new)
|
|
|
|
return ERR_PTR(-ENOMEM);
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
|
|
|
|
/* task's mempolicy is protected by alloc_lock */
|
|
|
|
if (old == current->mempolicy) {
|
|
|
|
task_lock(current);
|
|
|
|
*new = *old;
|
|
|
|
task_unlock(current);
|
|
|
|
} else
|
|
|
|
*new = *old;
|
|
|
|
|
[PATCH] cpuset: rebind vma mempolicies fix
Fix more of longstanding bug in cpuset/mempolicy interaction.
NUMA mempolicies (mm/mempolicy.c) are constrained by the current tasks cpuset
to just the Memory Nodes allowed by that cpuset. The kernel maintains
internal state for each mempolicy, tracking what nodes are used for the
MPOL_INTERLEAVE, MPOL_BIND or MPOL_PREFERRED policies.
When a tasks cpuset memory placement changes, whether because the cpuset
changed, or because the task was attached to a different cpuset, then the
tasks mempolicies have to be rebound to the new cpuset placement, so as to
preserve the cpuset-relative numbering of the nodes in that policy.
An earlier fix handled such mempolicy rebinding for mempolicies attached to a
task.
This fix rebinds mempolicies attached to vma's (address ranges in a tasks
address space.) Due to the need to hold the task->mm->mmap_sem semaphore while
updating vma's, the rebinding of vma mempolicies has to be done when the
cpuset memory placement is changed, at which time mmap_sem can be safely
acquired. The tasks mempolicy is rebound later, when the task next attempts
to allocate memory and notices that its task->cpuset_mems_generation is
out-of-date with its cpusets mems_generation.
Because walking the tasklist to find all tasks attached to a changing cpuset
requires holding tasklist_lock, a spinlock, one cannot update the vma's of the
affected tasks while doing the tasklist scan. In general, one cannot acquire
a semaphore (which can sleep) while already holding a spinlock (such as
tasklist_lock). So a list of mm references has to be built up during the
tasklist scan, then the tasklist lock dropped, then for each mm, its mmap_sem
acquired, and the vma's in that mm rebound.
Once the tasklist lock is dropped, affected tasks may fork new tasks, before
their mm's are rebound. A kernel global 'cpuset_being_rebound' is set to
point to the cpuset being rebound (there can only be one; cpuset modifications
are done under a global 'manage_sem' semaphore), and the mpol_copy code that
is used to copy a tasks mempolicies during fork catches such forking tasks,
and ensures their children are also rebound.
When a task is moved to a different cpuset, it is easier, as there is only one
task involved. It's mm->vma's are scanned, using the same
mpol_rebind_policy() as used above.
It may happen that both the mpol_copy hook and the update done via the
tasklist scan update the same mm twice. This is ok, as the mempolicies of
each vma in an mm keep track of what mems_allowed they are relative to, and
safely no-op a second request to rebind to the same nodes.
Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 02:01:59 -07:00
|
|
|
if (current_cpuset_is_being_rebound()) {
|
|
|
|
nodemask_t mems = cpuset_mems_allowed(current);
|
mempolicy: restructure rebinding-mempolicy functions
Nick Piggin reported that the allocator may see an empty nodemask when
changing cpuset's mems[1]. It happens only on the kernel that do not do
atomic nodemask_t stores. (MAX_NUMNODES > BITS_PER_LONG)
But I found that there is also a problem on the kernel that can do atomic
nodemask_t stores. The problem is that the allocator can't find a node to
alloc page when changing cpuset's mems though there is a lot of free
memory. The reason is like this:
(mpol: mempolicy)
task1 task1's mpol task2
alloc page 1
alloc on node0? NO 1
1 change mems from 1 to 0
1 rebind task1's mpol
0-1 set new bits
0 clear disallowed bits
alloc on node1? NO 0
...
can't alloc page
goto oom
I can use the attached program reproduce it by the following step:
# mkdir /dev/cpuset
# mount -t cpuset cpuset /dev/cpuset
# mkdir /dev/cpuset/1
# echo `cat /dev/cpuset/cpus` > /dev/cpuset/1/cpus
# echo `cat /dev/cpuset/mems` > /dev/cpuset/1/mems
# echo $$ > /dev/cpuset/1/tasks
# numactl --membind=`cat /dev/cpuset/mems` ./cpuset_mem_hog <nr_tasks> &
<nr_tasks> = max(nr_cpus - 1, 1)
# killall -s SIGUSR1 cpuset_mem_hog
# ./change_mems.sh
several hours later, oom will happen though there is a lot of free memory.
This patchset fixes this problem by expanding the nodes range first(set
newly allowed bits) and shrink it lazily(clear newly disallowed bits). So
we use a variable to tell the write-side task that read-side task is
reading nodemask, and the write-side task clears newly disallowed nodes
after read-side task ends the current memory allocation.
This patch:
In order to fix no node to alloc memory, when we want to update mempolicy
and mems_allowed, we expand the set of nodes first (set all the newly
nodes) and shrink the set of nodes lazily(clean disallowed nodes), But the
mempolicy's rebind functions may breaks the expanding.
So we restructure the mempolicy's rebind functions and split the rebind
work to two steps, just like the update of cpuset's mems: The 1st step:
expand the set of the mempolicy's nodes. The 2nd step: shrink the set of
the mempolicy's nodes. It is used when there is no real lock to protect
the mempolicy in the read-side. Otherwise we can do rebind work at once.
In order to implement it, we define
enum mpol_rebind_step {
MPOL_REBIND_ONCE,
MPOL_REBIND_STEP1,
MPOL_REBIND_STEP2,
MPOL_REBIND_NSTEP,
};
If the mempolicy needn't be updated by two steps, we can pass
MPOL_REBIND_ONCE to the rebind functions. Or we can pass
MPOL_REBIND_STEP1 to do the first step of the rebind work and pass
MPOL_REBIND_STEP2 to do the second step work.
Besides that, it maybe long time between these two step and we have to
release the lock that protects mempolicy and mems_allowed. If we hold the
lock once again, we must check whether the current mempolicy is under the
rebinding (the first step has been done) or not, because the task may
alloc a new mempolicy when we don't hold the lock. So we defined the
following flag to identify it:
#define MPOL_F_REBINDING (1 << 2)
The new functions will be used in the next patch.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Paul Menage <menage@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Ravikiran Thirumalai <kiran@scalex86.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-05-24 15:32:07 -06:00
|
|
|
if (new->flags & MPOL_F_REBINDING)
|
|
|
|
mpol_rebind_policy(new, &mems, MPOL_REBIND_STEP2);
|
|
|
|
else
|
|
|
|
mpol_rebind_policy(new, &mems, MPOL_REBIND_ONCE);
|
[PATCH] cpuset: rebind vma mempolicies fix
Fix more of longstanding bug in cpuset/mempolicy interaction.
NUMA mempolicies (mm/mempolicy.c) are constrained by the current tasks cpuset
to just the Memory Nodes allowed by that cpuset. The kernel maintains
internal state for each mempolicy, tracking what nodes are used for the
MPOL_INTERLEAVE, MPOL_BIND or MPOL_PREFERRED policies.
When a tasks cpuset memory placement changes, whether because the cpuset
changed, or because the task was attached to a different cpuset, then the
tasks mempolicies have to be rebound to the new cpuset placement, so as to
preserve the cpuset-relative numbering of the nodes in that policy.
An earlier fix handled such mempolicy rebinding for mempolicies attached to a
task.
This fix rebinds mempolicies attached to vma's (address ranges in a tasks
address space.) Due to the need to hold the task->mm->mmap_sem semaphore while
updating vma's, the rebinding of vma mempolicies has to be done when the
cpuset memory placement is changed, at which time mmap_sem can be safely
acquired. The tasks mempolicy is rebound later, when the task next attempts
to allocate memory and notices that its task->cpuset_mems_generation is
out-of-date with its cpusets mems_generation.
Because walking the tasklist to find all tasks attached to a changing cpuset
requires holding tasklist_lock, a spinlock, one cannot update the vma's of the
affected tasks while doing the tasklist scan. In general, one cannot acquire
a semaphore (which can sleep) while already holding a spinlock (such as
tasklist_lock). So a list of mm references has to be built up during the
tasklist scan, then the tasklist lock dropped, then for each mm, its mmap_sem
acquired, and the vma's in that mm rebound.
Once the tasklist lock is dropped, affected tasks may fork new tasks, before
their mm's are rebound. A kernel global 'cpuset_being_rebound' is set to
point to the cpuset being rebound (there can only be one; cpuset modifications
are done under a global 'manage_sem' semaphore), and the mpol_copy code that
is used to copy a tasks mempolicies during fork catches such forking tasks,
and ensures their children are also rebound.
When a task is moved to a different cpuset, it is easier, as there is only one
task involved. It's mm->vma's are scanned, using the same
mpol_rebind_policy() as used above.
It may happen that both the mpol_copy hook and the update done via the
tasklist scan update the same mm twice. This is ok, as the mempolicies of
each vma in an mm keep track of what mems_allowed they are relative to, and
safely no-op a second request to rebind to the same nodes.
Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 02:01:59 -07:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
atomic_set(&new->refcnt, 1);
|
|
|
|
return new;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Slow path of a mempolicy comparison */
|
2012-01-10 16:08:21 -07:00
|
|
|
bool __mpol_equal(struct mempolicy *a, struct mempolicy *b)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
if (!a || !b)
|
2012-01-10 16:08:21 -07:00
|
|
|
return false;
|
2008-04-28 03:13:12 -06:00
|
|
|
if (a->mode != b->mode)
|
2012-01-10 16:08:21 -07:00
|
|
|
return false;
|
2010-05-24 15:32:01 -06:00
|
|
|
if (a->flags != b->flags)
|
2012-01-10 16:08:21 -07:00
|
|
|
return false;
|
2010-05-24 15:32:01 -06:00
|
|
|
if (mpol_store_user_nodemask(a))
|
|
|
|
if (!nodes_equal(a->w.user_nodemask, b->w.user_nodemask))
|
2012-01-10 16:08:21 -07:00
|
|
|
return false;
|
2010-05-24 15:32:01 -06:00
|
|
|
|
2008-04-28 03:13:12 -06:00
|
|
|
switch (a->mode) {
|
2008-04-28 03:12:18 -06:00
|
|
|
case MPOL_BIND:
|
|
|
|
/* Fall through */
|
2005-04-16 16:20:36 -06:00
|
|
|
case MPOL_INTERLEAVE:
|
2012-01-10 16:08:21 -07:00
|
|
|
return !!nodes_equal(a->v.nodes, b->v.nodes);
|
2005-04-16 16:20:36 -06:00
|
|
|
case MPOL_PREFERRED:
|
2011-03-22 17:33:02 -06:00
|
|
|
return a->v.preferred_node == b->v.preferred_node;
|
2005-04-16 16:20:36 -06:00
|
|
|
default:
|
|
|
|
BUG();
|
2012-01-10 16:08:21 -07:00
|
|
|
return false;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Shared memory backing store policy support.
|
|
|
|
*
|
|
|
|
* Remember policies even when nobody has shared memory mapped.
|
|
|
|
* The policies are kept in Red-Black tree linked from the inode.
|
2016-01-14 16:18:36 -07:00
|
|
|
* They are protected by the sp->lock rwlock, which should be held
|
2005-04-16 16:20:36 -06:00
|
|
|
* for any accesses to the tree.
|
|
|
|
*/
|
|
|
|
|
2016-01-14 16:18:36 -07:00
|
|
|
/*
|
|
|
|
* lookup first element intersecting start-end. Caller holds sp->lock for
|
|
|
|
* reading or for writing
|
|
|
|
*/
|
2005-04-16 16:20:36 -06:00
|
|
|
static struct sp_node *
|
|
|
|
sp_lookup(struct shared_policy *sp, unsigned long start, unsigned long end)
|
|
|
|
{
|
|
|
|
struct rb_node *n = sp->root.rb_node;
|
|
|
|
|
|
|
|
while (n) {
|
|
|
|
struct sp_node *p = rb_entry(n, struct sp_node, nd);
|
|
|
|
|
|
|
|
if (start >= p->end)
|
|
|
|
n = n->rb_right;
|
|
|
|
else if (end <= p->start)
|
|
|
|
n = n->rb_left;
|
|
|
|
else
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (!n)
|
|
|
|
return NULL;
|
|
|
|
for (;;) {
|
|
|
|
struct sp_node *w = NULL;
|
|
|
|
struct rb_node *prev = rb_prev(n);
|
|
|
|
if (!prev)
|
|
|
|
break;
|
|
|
|
w = rb_entry(prev, struct sp_node, nd);
|
|
|
|
if (w->end <= start)
|
|
|
|
break;
|
|
|
|
n = prev;
|
|
|
|
}
|
|
|
|
return rb_entry(n, struct sp_node, nd);
|
|
|
|
}
|
|
|
|
|
2016-01-14 16:18:36 -07:00
|
|
|
/*
|
|
|
|
* Insert a new shared policy into the list. Caller holds sp->lock for
|
|
|
|
* writing.
|
|
|
|
*/
|
2005-04-16 16:20:36 -06:00
|
|
|
static void sp_insert(struct shared_policy *sp, struct sp_node *new)
|
|
|
|
{
|
|
|
|
struct rb_node **p = &sp->root.rb_node;
|
|
|
|
struct rb_node *parent = NULL;
|
|
|
|
struct sp_node *nd;
|
|
|
|
|
|
|
|
while (*p) {
|
|
|
|
parent = *p;
|
|
|
|
nd = rb_entry(parent, struct sp_node, nd);
|
|
|
|
if (new->start < nd->start)
|
|
|
|
p = &(*p)->rb_left;
|
|
|
|
else if (new->end > nd->end)
|
|
|
|
p = &(*p)->rb_right;
|
|
|
|
else
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
rb_link_node(&new->nd, parent, p);
|
|
|
|
rb_insert_color(&new->nd, &sp->root);
|
2007-07-16 00:38:16 -06:00
|
|
|
pr_debug("inserting %lx-%lx: %d\n", new->start, new->end,
|
2008-04-28 03:13:12 -06:00
|
|
|
new->policy ? new->policy->mode : 0);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Find shared policy intersecting idx */
|
|
|
|
struct mempolicy *
|
|
|
|
mpol_shared_policy_lookup(struct shared_policy *sp, unsigned long idx)
|
|
|
|
{
|
|
|
|
struct mempolicy *pol = NULL;
|
|
|
|
struct sp_node *sn;
|
|
|
|
|
|
|
|
if (!sp->root.rb_node)
|
|
|
|
return NULL;
|
2016-01-14 16:18:36 -07:00
|
|
|
read_lock(&sp->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
sn = sp_lookup(sp, idx, idx+1);
|
|
|
|
if (sn) {
|
|
|
|
mpol_get(sn->policy);
|
|
|
|
pol = sn->policy;
|
|
|
|
}
|
2016-01-14 16:18:36 -07:00
|
|
|
read_unlock(&sp->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return pol;
|
|
|
|
}
|
|
|
|
|
2012-10-08 17:29:19 -06:00
|
|
|
static void sp_free(struct sp_node *n)
|
|
|
|
{
|
|
|
|
mpol_put(n->policy);
|
|
|
|
kmem_cache_free(sn_cache, n);
|
|
|
|
}
|
|
|
|
|
2012-10-25 06:16:30 -06:00
|
|
|
/**
|
|
|
|
* mpol_misplaced - check whether current page node is valid in policy
|
|
|
|
*
|
2014-06-04 17:08:18 -06:00
|
|
|
* @page: page to be checked
|
|
|
|
* @vma: vm area where page mapped
|
|
|
|
* @addr: virtual address where page mapped
|
2012-10-25 06:16:30 -06:00
|
|
|
*
|
|
|
|
* Lookup current policy node id for vma,addr and "compare to" page's
|
|
|
|
* node id.
|
|
|
|
*
|
|
|
|
* Returns:
|
|
|
|
* -1 - not misplaced, page is in the right node
|
|
|
|
* node - node id where the page should be
|
|
|
|
*
|
|
|
|
* Policy determination "mimics" alloc_page_vma().
|
|
|
|
* Called from fault path where we know the vma and faulting address.
|
|
|
|
*/
|
|
|
|
int mpol_misplaced(struct page *page, struct vm_area_struct *vma, unsigned long addr)
|
|
|
|
{
|
|
|
|
struct mempolicy *pol;
|
2016-05-19 18:14:10 -06:00
|
|
|
struct zoneref *z;
|
2012-10-25 06:16:30 -06:00
|
|
|
int curnid = page_to_nid(page);
|
|
|
|
unsigned long pgoff;
|
2013-10-07 04:29:20 -06:00
|
|
|
int thiscpu = raw_smp_processor_id();
|
|
|
|
int thisnid = cpu_to_node(thiscpu);
|
2012-10-25 06:16:30 -06:00
|
|
|
int polnid = -1;
|
|
|
|
int ret = -1;
|
|
|
|
|
|
|
|
BUG_ON(!vma);
|
|
|
|
|
2014-10-09 16:27:57 -06:00
|
|
|
pol = get_vma_policy(vma, addr);
|
2012-10-25 06:16:30 -06:00
|
|
|
if (!(pol->flags & MPOL_F_MOF))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
switch (pol->mode) {
|
|
|
|
case MPOL_INTERLEAVE:
|
|
|
|
BUG_ON(addr >= vma->vm_end);
|
|
|
|
BUG_ON(addr < vma->vm_start);
|
|
|
|
|
|
|
|
pgoff = vma->vm_pgoff;
|
|
|
|
pgoff += (addr - vma->vm_start) >> PAGE_SHIFT;
|
|
|
|
polnid = offset_il_node(pol, vma, pgoff);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case MPOL_PREFERRED:
|
|
|
|
if (pol->flags & MPOL_F_LOCAL)
|
|
|
|
polnid = numa_node_id();
|
|
|
|
else
|
|
|
|
polnid = pol->v.preferred_node;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case MPOL_BIND:
|
2016-05-19 18:14:10 -06:00
|
|
|
|
2012-10-25 06:16:30 -06:00
|
|
|
/*
|
|
|
|
* allows binding to multiple nodes.
|
|
|
|
* use current page if in policy nodemask,
|
|
|
|
* else select nearest allowed node, if any.
|
|
|
|
* If no allowed nodes, use current [!misplaced].
|
|
|
|
*/
|
|
|
|
if (node_isset(curnid, pol->v.nodes))
|
|
|
|
goto out;
|
2016-05-19 18:14:10 -06:00
|
|
|
z = first_zones_zonelist(
|
2012-10-25 06:16:30 -06:00
|
|
|
node_zonelist(numa_node_id(), GFP_HIGHUSER),
|
|
|
|
gfp_zone(GFP_HIGHUSER),
|
2016-05-19 18:14:10 -06:00
|
|
|
&pol->v.nodes);
|
|
|
|
polnid = z->zone->node;
|
2012-10-25 06:16:30 -06:00
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
2012-11-02 12:19:13 -06:00
|
|
|
|
|
|
|
/* Migrate the page towards the node whose CPU is referencing it */
|
2012-11-12 02:17:07 -07:00
|
|
|
if (pol->flags & MPOL_F_MORON) {
|
2013-10-07 04:29:20 -06:00
|
|
|
polnid = thisnid;
|
2012-11-02 12:19:13 -06:00
|
|
|
|
2014-01-27 15:03:44 -07:00
|
|
|
if (!should_numa_migrate_memory(current, page, curnid, thiscpu))
|
2013-10-07 04:29:39 -06:00
|
|
|
goto out;
|
2012-11-12 02:17:07 -07:00
|
|
|
}
|
|
|
|
|
2012-10-25 06:16:30 -06:00
|
|
|
if (curnid != polnid)
|
|
|
|
ret = polnid;
|
|
|
|
out:
|
|
|
|
mpol_cond_put(pol);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
static void sp_delete(struct shared_policy *sp, struct sp_node *n)
|
|
|
|
{
|
2007-07-16 00:38:16 -06:00
|
|
|
pr_debug("deleting %lx-l%lx\n", n->start, n->end);
|
2005-04-16 16:20:36 -06:00
|
|
|
rb_erase(&n->nd, &sp->root);
|
2012-10-08 17:29:19 -06:00
|
|
|
sp_free(n);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2012-12-21 16:10:25 -07:00
|
|
|
static void sp_node_init(struct sp_node *node, unsigned long start,
|
|
|
|
unsigned long end, struct mempolicy *pol)
|
|
|
|
{
|
|
|
|
node->start = start;
|
|
|
|
node->end = end;
|
|
|
|
node->policy = pol;
|
|
|
|
}
|
|
|
|
|
2007-10-16 02:26:26 -06:00
|
|
|
static struct sp_node *sp_alloc(unsigned long start, unsigned long end,
|
|
|
|
struct mempolicy *pol)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
struct sp_node *n;
|
|
|
|
struct mempolicy *newpol;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
n = kmem_cache_alloc(sn_cache, GFP_KERNEL);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (!n)
|
|
|
|
return NULL;
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
|
|
|
|
newpol = mpol_dup(pol);
|
|
|
|
if (IS_ERR(newpol)) {
|
|
|
|
kmem_cache_free(sn_cache, n);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
newpol->flags |= MPOL_F_SHARED;
|
2012-12-21 16:10:25 -07:00
|
|
|
sp_node_init(n, start, end, newpol);
|
mempolicy: remove mempolicy sharing
Dave Jones' system call fuzz testing tool "trinity" triggered the
following bug error with slab debugging enabled
=============================================================================
BUG numa_policy (Not tainted): Poison overwritten
-----------------------------------------------------------------------------
INFO: 0xffff880146498250-0xffff880146498250. First byte 0x6a instead of 0x6b
INFO: Allocated in mpol_new+0xa3/0x140 age=46310 cpu=6 pid=32154
__slab_alloc+0x3d3/0x445
kmem_cache_alloc+0x29d/0x2b0
mpol_new+0xa3/0x140
sys_mbind+0x142/0x620
system_call_fastpath+0x16/0x1b
INFO: Freed in __mpol_put+0x27/0x30 age=46268 cpu=6 pid=32154
__slab_free+0x2e/0x1de
kmem_cache_free+0x25a/0x260
__mpol_put+0x27/0x30
remove_vma+0x68/0x90
exit_mmap+0x118/0x140
mmput+0x73/0x110
exit_mm+0x108/0x130
do_exit+0x162/0xb90
do_group_exit+0x4f/0xc0
sys_exit_group+0x17/0x20
system_call_fastpath+0x16/0x1b
INFO: Slab 0xffffea0005192600 objects=27 used=27 fp=0x (null) flags=0x20000000004080
INFO: Object 0xffff880146498250 @offset=592 fp=0xffff88014649b9d0
The problem is that the structure is being prematurely freed due to a
reference count imbalance. In the following case mbind(addr, len) should
replace the memory policies of both vma1 and vma2 and thus they will
become to share the same mempolicy and the new mempolicy will have the
MPOL_F_SHARED flag.
+-------------------+-------------------+
| vma1 | vma2(shmem) |
+-------------------+-------------------+
| |
addr addr+len
alloc_pages_vma() uses get_vma_policy() and mpol_cond_put() pair for
maintaining the mempolicy reference count. The current rule is that
get_vma_policy() only increments refcount for shmem VMA and
mpol_conf_put() only decrements refcount if the policy has
MPOL_F_SHARED.
In above case, vma1 is not shmem vma and vma->policy has MPOL_F_SHARED!
The reference count will be decreased even though was not increased
whenever alloc_page_vma() is called. This has been broken since commit
[52cd3b07: mempolicy: rework mempolicy Reference Counting] in 2008.
There is another serious bug with the sharing of memory policies.
Currently, mempolicy rebind logic (it is called from cpuset rebinding)
ignores a refcount of mempolicy and override it forcibly. Thus, any
mempolicy sharing may cause mempolicy corruption. The bug was
introduced by commit [68860ec1: cpusets: automatic numa mempolicy
rebinding].
Ideally, the shared policy handling would be rewritten to either
properly handle COW of the policy structures or at least reference count
MPOL_F_SHARED based exclusively on information within the policy.
However, this patch takes the easier approach of disabling any policy
sharing between VMAs. Each new range allocated with sp_alloc will
allocate a new policy, set the reference count to 1 and drop the
reference count of the old policy. This increases the memory footprint
but is not expected to be a major problem as mbind() is unlikely to be
used for fine-grained ranges. It is also inefficient because it means
we allocate a new policy even in cases where mbind_range() could use the
new_policy passed to it. However, it is more straight-forward and the
change should be invisible to the user.
[mgorman@suse.de: Edited changelog]
Reported-by: Dave Jones <davej@redhat.com>,
Cc: Christoph Lameter <cl@linux.com>,
Reviewed-by: Christoph Lameter <cl@linux.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Josh Boyer <jwboyer@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-10-08 17:29:16 -06:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
return n;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Replace a policy range. */
|
|
|
|
static int shared_policy_replace(struct shared_policy *sp, unsigned long start,
|
|
|
|
unsigned long end, struct sp_node *new)
|
|
|
|
{
|
2012-10-08 17:29:17 -06:00
|
|
|
struct sp_node *n;
|
2012-12-21 16:10:25 -07:00
|
|
|
struct sp_node *n_new = NULL;
|
|
|
|
struct mempolicy *mpol_new = NULL;
|
2012-10-08 17:29:17 -06:00
|
|
|
int ret = 0;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2012-12-21 16:10:25 -07:00
|
|
|
restart:
|
2016-01-14 16:18:36 -07:00
|
|
|
write_lock(&sp->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
n = sp_lookup(sp, start, end);
|
|
|
|
/* Take care of old policies in the same range. */
|
|
|
|
while (n && n->start < end) {
|
|
|
|
struct rb_node *next = rb_next(&n->nd);
|
|
|
|
if (n->start >= start) {
|
|
|
|
if (n->end <= end)
|
|
|
|
sp_delete(sp, n);
|
|
|
|
else
|
|
|
|
n->start = end;
|
|
|
|
} else {
|
|
|
|
/* Old policy spanning whole new range. */
|
|
|
|
if (n->end > end) {
|
2012-12-21 16:10:25 -07:00
|
|
|
if (!n_new)
|
|
|
|
goto alloc_new;
|
|
|
|
|
|
|
|
*mpol_new = *n->policy;
|
|
|
|
atomic_set(&mpol_new->refcnt, 1);
|
2013-03-08 13:43:29 -07:00
|
|
|
sp_node_init(n_new, end, n->end, mpol_new);
|
2005-04-16 16:20:36 -06:00
|
|
|
n->end = start;
|
2013-03-08 13:43:28 -07:00
|
|
|
sp_insert(sp, n_new);
|
2012-12-21 16:10:25 -07:00
|
|
|
n_new = NULL;
|
|
|
|
mpol_new = NULL;
|
2005-04-16 16:20:36 -06:00
|
|
|
break;
|
|
|
|
} else
|
|
|
|
n->end = start;
|
|
|
|
}
|
|
|
|
if (!next)
|
|
|
|
break;
|
|
|
|
n = rb_entry(next, struct sp_node, nd);
|
|
|
|
}
|
|
|
|
if (new)
|
|
|
|
sp_insert(sp, new);
|
2016-01-14 16:18:36 -07:00
|
|
|
write_unlock(&sp->lock);
|
2012-12-21 16:10:25 -07:00
|
|
|
ret = 0;
|
|
|
|
|
|
|
|
err_out:
|
|
|
|
if (mpol_new)
|
|
|
|
mpol_put(mpol_new);
|
|
|
|
if (n_new)
|
|
|
|
kmem_cache_free(sn_cache, n_new);
|
|
|
|
|
2012-10-08 17:29:17 -06:00
|
|
|
return ret;
|
2012-12-21 16:10:25 -07:00
|
|
|
|
|
|
|
alloc_new:
|
2016-01-14 16:18:36 -07:00
|
|
|
write_unlock(&sp->lock);
|
2012-12-21 16:10:25 -07:00
|
|
|
ret = -ENOMEM;
|
|
|
|
n_new = kmem_cache_alloc(sn_cache, GFP_KERNEL);
|
|
|
|
if (!n_new)
|
|
|
|
goto err_out;
|
|
|
|
mpol_new = kmem_cache_alloc(policy_cache, GFP_KERNEL);
|
|
|
|
if (!mpol_new)
|
|
|
|
goto err_out;
|
|
|
|
goto restart;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
/**
|
|
|
|
* mpol_shared_policy_init - initialize shared policy for inode
|
|
|
|
* @sp: pointer to inode shared policy
|
|
|
|
* @mpol: struct mempolicy to install
|
|
|
|
*
|
|
|
|
* Install non-NULL @mpol in inode's shared policy rb-tree.
|
|
|
|
* On entry, the current task has a reference on a non-NULL @mpol.
|
|
|
|
* This must be released on exit.
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
* This is called at get_inode() calls and we can use GFP_KERNEL.
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
*/
|
|
|
|
void mpol_shared_policy_init(struct shared_policy *sp, struct mempolicy *mpol)
|
|
|
|
{
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
int ret;
|
|
|
|
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
sp->root = RB_ROOT; /* empty tree == default mempolicy */
|
2016-01-14 16:18:36 -07:00
|
|
|
rwlock_init(&sp->lock);
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
|
|
|
|
if (mpol) {
|
|
|
|
struct vm_area_struct pvma;
|
|
|
|
struct mempolicy *new;
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
NODEMASK_SCRATCH(scratch);
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
if (!scratch)
|
2010-06-29 16:05:30 -06:00
|
|
|
goto put_mpol;
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
/* contextualize the tmpfs mount point mempolicy */
|
|
|
|
new = mpol_new(mpol->mode, mpol->flags, &mpol->w.user_nodemask);
|
2010-05-24 15:32:04 -06:00
|
|
|
if (IS_ERR(new))
|
2010-05-26 00:42:58 -06:00
|
|
|
goto free_scratch; /* no valid nodemask intersection */
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
|
|
|
|
task_lock(current);
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
ret = mpol_set_nodemask(new, &mpol->w.user_nodemask, scratch);
|
cpuset,mm: update tasks' mems_allowed in time
Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.
In order to update tasks' mems_allowed in time, we must modify the code of
memory policy. Because the memory policy is applied in the process's
context originally. After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.
But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression. But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set. In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.
[lee.schermerhorn@hp.com:
The rework of mpol_new() to extract the adjusting of the node mask to
apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
allocation. Fix this by adding the check for MPOL_PREFERRED and empty
node mask to mpol_new_mpolicy().
Remove the now unneeded 'nodes = NULL' from mpol_new().
Note that mpol_new_mempolicy() is always called with a non-NULL
'nodes' parameter now that it has been removed from mpol_new().
Therefore, we don't need to test nodes for NULL before testing it for
'empty'. However, just to be extra paranoid, add a VM_BUG_ON() to
verify this assumption.]
[lee.schermerhorn@hp.com:
I don't think the function name 'mpol_new_mempolicy' is descriptive
enough to differentiate it from mpol_new().
This function applies cpuset set context, usually constraining nodes
to those allowed by the cpuset. However, when the 'RELATIVE_NODES flag
is set, it also translates the nodes. So I settled on
'mpol_set_nodemask()', because the comment block for mpol_new() mentions
that we need to call this function to "set nodes".
Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-06-16 16:31:49 -06:00
|
|
|
task_unlock(current);
|
2010-05-24 15:32:04 -06:00
|
|
|
if (ret)
|
2010-06-29 16:05:30 -06:00
|
|
|
goto put_new;
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
|
|
|
|
/* Create pseudo-vma that contains just the policy */
|
|
|
|
memset(&pvma, 0, sizeof(struct vm_area_struct));
|
|
|
|
pvma.vm_end = TASK_SIZE; /* policy covers entire file */
|
|
|
|
mpol_set_shared_policy(sp, &pvma, new); /* adds ref */
|
2010-05-24 15:32:04 -06:00
|
|
|
|
2010-06-29 16:05:30 -06:00
|
|
|
put_new:
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
mpol_put(new); /* drop initial ref */
|
2010-05-26 00:42:58 -06:00
|
|
|
free_scratch:
|
mm: make set_mempolicy(MPOL_INTERLEAV) N_HIGH_MEMORY aware
At first, init_task's mems_allowed is initialized as this.
init_task->mems_allowed == node_state[N_POSSIBLE]
And cpuset's top_cpuset mask is initialized as this
top_cpuset->mems_allowed = node_state[N_HIGH_MEMORY]
Before 2.6.29:
policy's mems_allowed is initialized as this.
1. update tasks->mems_allowed by its cpuset->mems_allowed.
2. policy->mems_allowed = nodes_and(tasks->mems_allowed, user's mask)
Updating task's mems_allowed in reference to top_cpuset's one.
cpuset's mems_allowed is aware of N_HIGH_MEMORY, always.
In 2.6.30: After commit 58568d2a8215cb6f55caf2332017d7bdff954e1c
("cpuset,mm: update tasks' mems_allowed in time"), policy's mems_allowed
is initialized as this.
1. policy->mems_allowd = nodes_and(task->mems_allowed, user's mask)
Here, if task is in top_cpuset, task->mems_allowed is not updated from
init's one. Assume user excutes command as #numactrl --interleave=all
,....
policy->mems_allowd = nodes_and(N_POSSIBLE, ALL_SET_MASK)
Then, policy's mems_allowd can includes a possible node, which has no pgdat.
MPOL's INTERLEAVE just scans nodemask of task->mems_allowd and access this
directly.
NODE_DATA(nid)->zonelist even if NODE_DATA(nid)==NULL
Then, what's we need is making policy->mems_allowed be aware of
N_HIGH_MEMORY. This patch does that. But to do so, extra nodemask will
be on statck. Because I know cpumask has a new interface of
CPUMASK_ALLOC(), I added it to node.
This patch stands on old behavior. But I feel this fix itself is just a
Band-Aid. But to do fundametal fix, we have to take care of memory
hotplug and it takes time. (task->mems_allowd should be N_HIGH_MEMORY, I
think.)
mpol_set_nodemask() should be aware of N_HIGH_MEMORY and policy's nodemask
should be includes only online nodes.
In old behavior, this is guaranteed by frequent reference to cpuset's
code. Now, most of them are removed and mempolicy has to check it by
itself.
To do check, a few nodemask_t will be used for calculating nodemask. But,
size of nodemask_t can be big and it's not good to allocate them on stack.
Now, cpumask_t has CPUMASK_ALLOC/FREE an easy code for get scratch area.
NODEMASK_ALLOC/FREE shoudl be there.
[akpm@linux-foundation.org: cleanups & tweaks]
Tested-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-08-06 16:07:33 -06:00
|
|
|
NODEMASK_SCRATCH_FREE(scratch);
|
2010-06-29 16:05:30 -06:00
|
|
|
put_mpol:
|
|
|
|
mpol_put(mpol); /* drop our incoming ref on sb mpol */
|
2006-01-14 14:20:48 -07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
int mpol_set_shared_policy(struct shared_policy *info,
|
|
|
|
struct vm_area_struct *vma, struct mempolicy *npol)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
struct sp_node *new = NULL;
|
|
|
|
unsigned long sz = vma_pages(vma);
|
|
|
|
|
2008-04-28 03:12:25 -06:00
|
|
|
pr_debug("set_shared_policy %lx sz %lu %d %d %lx\n",
|
2005-04-16 16:20:36 -06:00
|
|
|
vma->vm_pgoff,
|
2008-04-28 03:13:12 -06:00
|
|
|
sz, npol ? npol->mode : -1,
|
2008-04-28 03:12:25 -06:00
|
|
|
npol ? npol->flags : -1,
|
2013-02-22 17:35:36 -07:00
|
|
|
npol ? nodes_addr(npol->v.nodes)[0] : NUMA_NO_NODE);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (npol) {
|
|
|
|
new = sp_alloc(vma->vm_pgoff, vma->vm_pgoff + sz, npol);
|
|
|
|
if (!new)
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
err = shared_policy_replace(info, vma->vm_pgoff, vma->vm_pgoff+sz, new);
|
|
|
|
if (err && new)
|
2012-10-08 17:29:19 -06:00
|
|
|
sp_free(new);
|
2005-04-16 16:20:36 -06:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Free a backing policy store on inode delete. */
|
|
|
|
void mpol_free_shared_policy(struct shared_policy *p)
|
|
|
|
{
|
|
|
|
struct sp_node *n;
|
|
|
|
struct rb_node *next;
|
|
|
|
|
|
|
|
if (!p->root.rb_node)
|
|
|
|
return;
|
2016-01-14 16:18:36 -07:00
|
|
|
write_lock(&p->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
next = rb_first(&p->root);
|
|
|
|
while (next) {
|
|
|
|
n = rb_entry(next, struct sp_node, nd);
|
|
|
|
next = rb_next(&n->nd);
|
2012-10-08 17:29:19 -06:00
|
|
|
sp_delete(p, n);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
2016-01-14 16:18:36 -07:00
|
|
|
write_unlock(&p->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2012-11-22 04:16:36 -07:00
|
|
|
#ifdef CONFIG_NUMA_BALANCING
|
2014-01-29 15:05:42 -07:00
|
|
|
static int __initdata numabalancing_override;
|
2012-11-22 04:16:36 -07:00
|
|
|
|
|
|
|
static void __init check_numabalancing_enable(void)
|
|
|
|
{
|
|
|
|
bool numabalancing_default = false;
|
|
|
|
|
|
|
|
if (IS_ENABLED(CONFIG_NUMA_BALANCING_DEFAULT_ENABLED))
|
|
|
|
numabalancing_default = true;
|
|
|
|
|
2014-01-29 15:05:42 -07:00
|
|
|
/* Parsed by setup_numabalancing. override == 1 enables, -1 disables */
|
|
|
|
if (numabalancing_override)
|
|
|
|
set_numabalancing_state(numabalancing_override == 1);
|
|
|
|
|
2015-05-14 16:17:09 -06:00
|
|
|
if (num_online_nodes() > 1 && !numabalancing_override) {
|
2016-03-17 15:19:47 -06:00
|
|
|
pr_info("%s automatic NUMA balancing. Configure with numa_balancing= or the kernel.numa_balancing sysctl\n",
|
2014-01-29 15:05:42 -07:00
|
|
|
numabalancing_default ? "Enabling" : "Disabling");
|
2012-11-22 04:16:36 -07:00
|
|
|
set_numabalancing_state(numabalancing_default);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int __init setup_numabalancing(char *str)
|
|
|
|
{
|
|
|
|
int ret = 0;
|
|
|
|
if (!str)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
if (!strcmp(str, "enable")) {
|
2014-01-29 15:05:42 -07:00
|
|
|
numabalancing_override = 1;
|
2012-11-22 04:16:36 -07:00
|
|
|
ret = 1;
|
|
|
|
} else if (!strcmp(str, "disable")) {
|
2014-01-29 15:05:42 -07:00
|
|
|
numabalancing_override = -1;
|
2012-11-22 04:16:36 -07:00
|
|
|
ret = 1;
|
|
|
|
}
|
|
|
|
out:
|
|
|
|
if (!ret)
|
2014-01-29 15:05:43 -07:00
|
|
|
pr_warn("Unable to parse numa_balancing=\n");
|
2012-11-22 04:16:36 -07:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
__setup("numa_balancing=", setup_numabalancing);
|
|
|
|
#else
|
|
|
|
static inline void __init check_numabalancing_enable(void)
|
|
|
|
{
|
|
|
|
}
|
|
|
|
#endif /* CONFIG_NUMA_BALANCING */
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/* assumes fs == KERNEL_DS */
|
|
|
|
void __init numa_policy_init(void)
|
|
|
|
{
|
2007-07-16 00:38:15 -06:00
|
|
|
nodemask_t interleave_nodes;
|
|
|
|
unsigned long largest = 0;
|
|
|
|
int nid, prefer = 0;
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
policy_cache = kmem_cache_create("numa_policy",
|
|
|
|
sizeof(struct mempolicy),
|
2007-07-19 19:11:58 -06:00
|
|
|
0, SLAB_PANIC, NULL);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
sn_cache = kmem_cache_create("shared_policy_node",
|
|
|
|
sizeof(struct sp_node),
|
2007-07-19 19:11:58 -06:00
|
|
|
0, SLAB_PANIC, NULL);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2012-11-02 12:19:13 -06:00
|
|
|
for_each_node(nid) {
|
|
|
|
preferred_node_policy[nid] = (struct mempolicy) {
|
|
|
|
.refcnt = ATOMIC_INIT(1),
|
|
|
|
.mode = MPOL_PREFERRED,
|
|
|
|
.flags = MPOL_F_MOF | MPOL_F_MORON,
|
|
|
|
.v = { .preferred_node = nid, },
|
|
|
|
};
|
|
|
|
}
|
|
|
|
|
2007-07-16 00:38:15 -06:00
|
|
|
/*
|
|
|
|
* Set interleaving policy for system init. Interleaving is only
|
|
|
|
* enabled across suitably sized nodes (default is >= 16MB), or
|
|
|
|
* fall back to the largest node if they're all smaller.
|
|
|
|
*/
|
|
|
|
nodes_clear(interleave_nodes);
|
2012-12-12 14:51:33 -07:00
|
|
|
for_each_node_state(nid, N_MEMORY) {
|
2007-07-16 00:38:15 -06:00
|
|
|
unsigned long total_pages = node_present_pages(nid);
|
|
|
|
|
|
|
|
/* Preserve the largest node */
|
|
|
|
if (largest < total_pages) {
|
|
|
|
largest = total_pages;
|
|
|
|
prefer = nid;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Interleave this node? */
|
|
|
|
if ((total_pages << PAGE_SHIFT) >= (16 << 20))
|
|
|
|
node_set(nid, interleave_nodes);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* All too small, use the largest */
|
|
|
|
if (unlikely(nodes_empty(interleave_nodes)))
|
|
|
|
node_set(prefer, interleave_nodes);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2008-04-28 03:12:25 -06:00
|
|
|
if (do_set_mempolicy(MPOL_INTERLEAVE, 0, &interleave_nodes))
|
2014-06-06 15:38:30 -06:00
|
|
|
pr_err("%s: interleaving failed\n", __func__);
|
2012-11-22 04:16:36 -07:00
|
|
|
|
|
|
|
check_numabalancing_enable();
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2005-10-29 19:16:59 -06:00
|
|
|
/* Reset policy of current process to default */
|
2005-04-16 16:20:36 -06:00
|
|
|
void numa_default_policy(void)
|
|
|
|
{
|
2008-04-28 03:12:25 -06:00
|
|
|
do_set_mempolicy(MPOL_DEFAULT, 0, NULL);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
[PATCH] cpusets: automatic numa mempolicy rebinding
This patch automatically updates a tasks NUMA mempolicy when its cpuset
memory placement changes. It does so within the context of the task,
without any need to support low level external mempolicy manipulation.
If a system is not using cpusets, or if running on a system with just the
root (all-encompassing) cpuset, then this remap is a no-op. Only when a
task is moved between cpusets, or a cpusets memory placement is changed
does the following apply. Otherwise, the main routine below,
rebind_policy() is not even called.
When mixing cpusets, scheduler affinity, and NUMA mempolicies, the
essential role of cpusets is to place jobs (several related tasks) on a set
of CPUs and Memory Nodes, the essential role of sched_setaffinity is to
manage a jobs processor placement within its allowed cpuset, and the
essential role of NUMA mempolicy (mbind, set_mempolicy) is to manage a jobs
memory placement within its allowed cpuset.
However, CPU affinity and NUMA memory placement are managed within the
kernel using absolute system wide numbering, not cpuset relative numbering.
This is ok until a job is migrated to a different cpuset, or what's the
same, a jobs cpuset is moved to different CPUs and Memory Nodes.
Then the CPU affinity and NUMA memory placement of the tasks in the job
need to be updated, to preserve their cpuset-relative position. This can
be done for CPU affinity using sched_setaffinity() from user code, as one
task can modify anothers CPU affinity. This cannot be done from an
external task for NUMA memory placement, as that can only be modified in
the context of the task using it.
However, it easy enough to remap a tasks NUMA mempolicy automatically when
a task is migrated, using the existing cpuset mechanism to trigger a
refresh of a tasks memory placement after its cpuset has changed. All that
is needed is the old and new nodemask, and notice to the task that it needs
to rebind its mempolicy. The tasks mems_allowed has the old mask, the
tasks cpuset has the new mask, and the existing
cpuset_update_current_mems_allowed() mechanism provides the notice. The
bitmap/cpumask/nodemask remap operators provide the cpuset relative
calculations.
This patch leaves open a couple of issues:
1) Updating vma and shmfs/tmpfs/hugetlbfs memory policies:
These mempolicies may reference nodes outside of those allowed to
the current task by its cpuset. Tasks are migrated as part of jobs,
which reside on what might be several cpusets in a subtree. When such
a job is migrated, all NUMA memory policy references to nodes within
that cpuset subtree should be translated, and references to any nodes
outside that subtree should be left untouched. A future patch will
provide the cpuset mechanism needed to mark such subtrees. With that
patch, we will be able to correctly migrate these other memory policies
across a job migration.
2) Updating cpuset, affinity and memory policies in user space:
This is harder. Any placement state stored in user space using
system-wide numbering will be invalidated across a migration. More
work will be required to provide user code with a migration-safe means
to manage its cpuset relative placement, while preserving the current
API's that pass system wide numbers, not cpuset relative numbers across
the kernel-user boundary.
Signed-off-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-30 16:02:36 -07:00
|
|
|
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
/*
|
|
|
|
* Parse and format mempolicy from/to strings
|
|
|
|
*/
|
|
|
|
|
2006-01-08 02:01:02 -07:00
|
|
|
/*
|
tmpfs mempolicy: fix /proc/mounts corrupting memory
Recently I suggested using "mount -o remount,mpol=local /tmp" in NUMA
mempolicy testing. Very nasty. Reading /proc/mounts, /proc/pid/mounts
or /proc/pid/mountinfo may then corrupt one bit of kernel memory, often
in a page table (causing "Bad swap" or "Bad page map" warning or "Bad
pagetable" oops), sometimes in a vm_area_struct or rbnode or somewhere
worse. "mpol=prefer" and "mpol=prefer:Node" are equally toxic.
Recent NUMA enhancements are not to blame: this dates back to 2.6.35,
when commit e17f74af351c "mempolicy: don't call mpol_set_nodemask() when
no_context" skipped mpol_parse_str()'s call to mpol_set_nodemask(),
which used to initialize v.preferred_node, or set MPOL_F_LOCAL in flags.
With slab poisoning, you can then rely on mpol_to_str() to set the bit
for node 0x6b6b, probably in the next page above the caller's stack.
mpol_parse_str() is only called from shmem_parse_options(): no_context
is always true, so call it unused for now, and remove !no_context code.
Set v.nodes or v.preferred_node or MPOL_F_LOCAL as mpol_to_str() might
expect. Then mpol_to_str() can ignore its no_context argument also,
the mpol being appropriately initialized whether contextualized or not.
Rename its no_context unused too, and let subsequent patch remove them
(that's not needed for stable backporting, which would involve rejects).
I don't understand why MPOL_LOCAL is described as a pseudo-policy:
it's a reasonable policy which suffers from a confusing implementation
in terms of MPOL_PREFERRED with MPOL_F_LOCAL. I believe this would be
much more robust if MPOL_LOCAL were recognized in switch statements
throughout, MPOL_F_LOCAL deleted, and MPOL_PREFERRED use the (possibly
empty) nodes mask like everyone else, instead of its preferred_node
variant (I presume an optimization from the days before MPOL_LOCAL).
But that would take me too long to get right and fully tested.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-02 03:01:33 -07:00
|
|
|
* "local" is implemented internally by MPOL_PREFERRED with MPOL_F_LOCAL flag.
|
2006-01-08 02:01:02 -07:00
|
|
|
*/
|
2010-05-24 15:32:04 -06:00
|
|
|
static const char * const policy_modes[] =
|
|
|
|
{
|
|
|
|
[MPOL_DEFAULT] = "default",
|
|
|
|
[MPOL_PREFERRED] = "prefer",
|
|
|
|
[MPOL_BIND] = "bind",
|
|
|
|
[MPOL_INTERLEAVE] = "interleave",
|
2012-10-25 06:16:29 -06:00
|
|
|
[MPOL_LOCAL] = "local",
|
2010-05-24 15:32:04 -06:00
|
|
|
};
|
2006-01-08 02:01:02 -07:00
|
|
|
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
|
|
|
|
#ifdef CONFIG_TMPFS
|
|
|
|
/**
|
tmpfs mempolicy: fix /proc/mounts corrupting memory
Recently I suggested using "mount -o remount,mpol=local /tmp" in NUMA
mempolicy testing. Very nasty. Reading /proc/mounts, /proc/pid/mounts
or /proc/pid/mountinfo may then corrupt one bit of kernel memory, often
in a page table (causing "Bad swap" or "Bad page map" warning or "Bad
pagetable" oops), sometimes in a vm_area_struct or rbnode or somewhere
worse. "mpol=prefer" and "mpol=prefer:Node" are equally toxic.
Recent NUMA enhancements are not to blame: this dates back to 2.6.35,
when commit e17f74af351c "mempolicy: don't call mpol_set_nodemask() when
no_context" skipped mpol_parse_str()'s call to mpol_set_nodemask(),
which used to initialize v.preferred_node, or set MPOL_F_LOCAL in flags.
With slab poisoning, you can then rely on mpol_to_str() to set the bit
for node 0x6b6b, probably in the next page above the caller's stack.
mpol_parse_str() is only called from shmem_parse_options(): no_context
is always true, so call it unused for now, and remove !no_context code.
Set v.nodes or v.preferred_node or MPOL_F_LOCAL as mpol_to_str() might
expect. Then mpol_to_str() can ignore its no_context argument also,
the mpol being appropriately initialized whether contextualized or not.
Rename its no_context unused too, and let subsequent patch remove them
(that's not needed for stable backporting, which would involve rejects).
I don't understand why MPOL_LOCAL is described as a pseudo-policy:
it's a reasonable policy which suffers from a confusing implementation
in terms of MPOL_PREFERRED with MPOL_F_LOCAL. I believe this would be
much more robust if MPOL_LOCAL were recognized in switch statements
throughout, MPOL_F_LOCAL deleted, and MPOL_PREFERRED use the (possibly
empty) nodes mask like everyone else, instead of its preferred_node
variant (I presume an optimization from the days before MPOL_LOCAL).
But that would take me too long to get right and fully tested.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-02 03:01:33 -07:00
|
|
|
* mpol_parse_str - parse string to mempolicy, for tmpfs mpol mount option.
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
* @str: string containing mempolicy to parse
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
* @mpol: pointer to struct mempolicy pointer, returned on success.
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
*
|
|
|
|
* Format of input:
|
|
|
|
* <mode>[=<flags>][:<nodelist>]
|
|
|
|
*
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
* On success, returns 0, else 1
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
*/
|
2013-01-02 03:04:23 -07:00
|
|
|
int mpol_parse_str(char *str, struct mempolicy **mpol)
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
{
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
struct mempolicy *new = NULL;
|
2010-05-24 15:32:03 -06:00
|
|
|
unsigned short mode;
|
tmpfs mempolicy: fix /proc/mounts corrupting memory
Recently I suggested using "mount -o remount,mpol=local /tmp" in NUMA
mempolicy testing. Very nasty. Reading /proc/mounts, /proc/pid/mounts
or /proc/pid/mountinfo may then corrupt one bit of kernel memory, often
in a page table (causing "Bad swap" or "Bad page map" warning or "Bad
pagetable" oops), sometimes in a vm_area_struct or rbnode or somewhere
worse. "mpol=prefer" and "mpol=prefer:Node" are equally toxic.
Recent NUMA enhancements are not to blame: this dates back to 2.6.35,
when commit e17f74af351c "mempolicy: don't call mpol_set_nodemask() when
no_context" skipped mpol_parse_str()'s call to mpol_set_nodemask(),
which used to initialize v.preferred_node, or set MPOL_F_LOCAL in flags.
With slab poisoning, you can then rely on mpol_to_str() to set the bit
for node 0x6b6b, probably in the next page above the caller's stack.
mpol_parse_str() is only called from shmem_parse_options(): no_context
is always true, so call it unused for now, and remove !no_context code.
Set v.nodes or v.preferred_node or MPOL_F_LOCAL as mpol_to_str() might
expect. Then mpol_to_str() can ignore its no_context argument also,
the mpol being appropriately initialized whether contextualized or not.
Rename its no_context unused too, and let subsequent patch remove them
(that's not needed for stable backporting, which would involve rejects).
I don't understand why MPOL_LOCAL is described as a pseudo-policy:
it's a reasonable policy which suffers from a confusing implementation
in terms of MPOL_PREFERRED with MPOL_F_LOCAL. I believe this would be
much more robust if MPOL_LOCAL were recognized in switch statements
throughout, MPOL_F_LOCAL deleted, and MPOL_PREFERRED use the (possibly
empty) nodes mask like everyone else, instead of its preferred_node
variant (I presume an optimization from the days before MPOL_LOCAL).
But that would take me too long to get right and fully tested.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-02 03:01:33 -07:00
|
|
|
unsigned short mode_flags;
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
nodemask_t nodes;
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
char *nodelist = strchr(str, ':');
|
|
|
|
char *flags = strchr(str, '=');
|
|
|
|
int err = 1;
|
|
|
|
|
|
|
|
if (nodelist) {
|
|
|
|
/* NUL-terminate mode or flags string */
|
|
|
|
*nodelist++ = '\0';
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
if (nodelist_parse(nodelist, nodes))
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
goto out;
|
2012-12-12 14:51:33 -07:00
|
|
|
if (!nodes_subset(nodes, node_states[N_MEMORY]))
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
goto out;
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
} else
|
|
|
|
nodes_clear(nodes);
|
|
|
|
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
if (flags)
|
|
|
|
*flags++ = '\0'; /* terminate mode string */
|
|
|
|
|
2012-10-25 06:16:28 -06:00
|
|
|
for (mode = 0; mode < MPOL_MAX; mode++) {
|
2010-05-24 15:32:04 -06:00
|
|
|
if (!strcmp(str, policy_modes[mode])) {
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2012-11-16 02:37:58 -07:00
|
|
|
if (mode >= MPOL_MAX)
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
goto out;
|
|
|
|
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
switch (mode) {
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
case MPOL_PREFERRED:
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
/*
|
|
|
|
* Insist on a nodelist of one node only
|
|
|
|
*/
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
if (nodelist) {
|
|
|
|
char *rest = nodelist;
|
|
|
|
while (isdigit(*rest))
|
|
|
|
rest++;
|
2010-03-23 14:35:32 -06:00
|
|
|
if (*rest)
|
|
|
|
goto out;
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
}
|
|
|
|
break;
|
|
|
|
case MPOL_INTERLEAVE:
|
|
|
|
/*
|
|
|
|
* Default to online nodes with memory if no nodelist
|
|
|
|
*/
|
|
|
|
if (!nodelist)
|
2012-12-12 14:51:33 -07:00
|
|
|
nodes = node_states[N_MEMORY];
|
2008-04-28 03:13:24 -06:00
|
|
|
break;
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
case MPOL_LOCAL:
|
2008-04-28 03:13:24 -06:00
|
|
|
/*
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
* Don't allow a nodelist; mpol_new() checks flags
|
2008-04-28 03:13:24 -06:00
|
|
|
*/
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
if (nodelist)
|
2008-04-28 03:13:24 -06:00
|
|
|
goto out;
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
mode = MPOL_PREFERRED;
|
2008-04-28 03:13:24 -06:00
|
|
|
break;
|
2010-03-23 14:35:28 -06:00
|
|
|
case MPOL_DEFAULT:
|
|
|
|
/*
|
|
|
|
* Insist on a empty nodelist
|
|
|
|
*/
|
|
|
|
if (!nodelist)
|
|
|
|
err = 0;
|
|
|
|
goto out;
|
2010-03-23 14:35:30 -06:00
|
|
|
case MPOL_BIND:
|
|
|
|
/*
|
|
|
|
* Insist on a nodelist
|
|
|
|
*/
|
|
|
|
if (!nodelist)
|
|
|
|
goto out;
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
}
|
|
|
|
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
mode_flags = 0;
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
if (flags) {
|
|
|
|
/*
|
|
|
|
* Currently, we only support two mutually exclusive
|
|
|
|
* mode flags.
|
|
|
|
*/
|
|
|
|
if (!strcmp(flags, "static"))
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
mode_flags |= MPOL_F_STATIC_NODES;
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
else if (!strcmp(flags, "relative"))
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
mode_flags |= MPOL_F_RELATIVE_NODES;
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
else
|
2010-03-23 14:35:32 -06:00
|
|
|
goto out;
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
}
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
|
|
|
|
new = mpol_new(mode, mode_flags, &nodes);
|
|
|
|
if (IS_ERR(new))
|
2010-03-23 14:35:32 -06:00
|
|
|
goto out;
|
|
|
|
|
tmpfs mempolicy: fix /proc/mounts corrupting memory
Recently I suggested using "mount -o remount,mpol=local /tmp" in NUMA
mempolicy testing. Very nasty. Reading /proc/mounts, /proc/pid/mounts
or /proc/pid/mountinfo may then corrupt one bit of kernel memory, often
in a page table (causing "Bad swap" or "Bad page map" warning or "Bad
pagetable" oops), sometimes in a vm_area_struct or rbnode or somewhere
worse. "mpol=prefer" and "mpol=prefer:Node" are equally toxic.
Recent NUMA enhancements are not to blame: this dates back to 2.6.35,
when commit e17f74af351c "mempolicy: don't call mpol_set_nodemask() when
no_context" skipped mpol_parse_str()'s call to mpol_set_nodemask(),
which used to initialize v.preferred_node, or set MPOL_F_LOCAL in flags.
With slab poisoning, you can then rely on mpol_to_str() to set the bit
for node 0x6b6b, probably in the next page above the caller's stack.
mpol_parse_str() is only called from shmem_parse_options(): no_context
is always true, so call it unused for now, and remove !no_context code.
Set v.nodes or v.preferred_node or MPOL_F_LOCAL as mpol_to_str() might
expect. Then mpol_to_str() can ignore its no_context argument also,
the mpol being appropriately initialized whether contextualized or not.
Rename its no_context unused too, and let subsequent patch remove them
(that's not needed for stable backporting, which would involve rejects).
I don't understand why MPOL_LOCAL is described as a pseudo-policy:
it's a reasonable policy which suffers from a confusing implementation
in terms of MPOL_PREFERRED with MPOL_F_LOCAL. I believe this would be
much more robust if MPOL_LOCAL were recognized in switch statements
throughout, MPOL_F_LOCAL deleted, and MPOL_PREFERRED use the (possibly
empty) nodes mask like everyone else, instead of its preferred_node
variant (I presume an optimization from the days before MPOL_LOCAL).
But that would take me too long to get right and fully tested.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-02 03:01:33 -07:00
|
|
|
/*
|
|
|
|
* Save nodes for mpol_to_str() to show the tmpfs mount options
|
|
|
|
* for /proc/mounts, /proc/pid/mounts and /proc/pid/mountinfo.
|
|
|
|
*/
|
|
|
|
if (mode != MPOL_PREFERRED)
|
|
|
|
new->v.nodes = nodes;
|
|
|
|
else if (nodelist)
|
|
|
|
new->v.preferred_node = first_node(nodes);
|
|
|
|
else
|
|
|
|
new->flags |= MPOL_F_LOCAL;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Save nodes for contextualization: this will be used to "clone"
|
|
|
|
* the mempolicy in a specific context [cpuset] at a later time.
|
|
|
|
*/
|
|
|
|
new->w.user_nodemask = nodes;
|
|
|
|
|
2010-03-23 14:35:32 -06:00
|
|
|
err = 0;
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
out:
|
|
|
|
/* Restore string for error message */
|
|
|
|
if (nodelist)
|
|
|
|
*--nodelist = ':';
|
|
|
|
if (flags)
|
|
|
|
*--flags = '=';
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
if (!err)
|
|
|
|
*mpol = new;
|
mempolicy: rework shmem mpol parsing and display
mm/shmem.c currently contains functions to parse and display memory policy
strings for the tmpfs 'mpol' mount option. Move this to mm/mempolicy.c with
the rest of the mempolicy support. With subsequent patches, we'll be able to
remove knowledge of the details [mode, flags, policy, ...] completely from
shmem.c
1) replace shmem_parse_mpol() in mm/shmem.c with mpol_parse_str() in
mm/mempolicy.c. Rework to use the policy_types[] array [used by
mpol_to_str()] to look up mode by name.
2) use mpol_to_str() to format policy for shmem_show_mpol(). mpol_to_str()
expects a pointer to a struct mempolicy, so temporarily construct one.
This will be replaced with a reference to a struct mempolicy in the tmpfs
superblock in a subsequent patch.
NOTE 1: I changed mpol_to_str() to use a colon ':' rather than an equal
sign '=' as the nodemask delimiter to match mpol_parse_str() and the
tmpfs/shmem mpol mount option formatting that now uses mpol_to_str(). This
is a user visible change to numa_maps, but then the addition of the mode
flags already changed the display. It makes sense to me to have the mounts
and numa_maps display the policy in the same format. However, if anyone
objects strongly, I can pass the desired nodemask delimeter as an arg to
mpol_to_str().
Note 2: Like show_numa_map(), I don't check the return code from
mpol_to_str(). I do use a longer buffer than the one provided by
show_numa_map(), which seems to have sufficed so far.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:23 -06:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
#endif /* CONFIG_TMPFS */
|
|
|
|
|
mempolicy: use struct mempolicy pointer in shmem_sb_info
This patch replaces the mempolicy mode, mode_flags, and nodemask in the
shmem_sb_info struct with a struct mempolicy pointer, initialized to NULL.
This removes dependency on the details of mempolicy from shmem.c and hugetlbfs
inode.c and simplifies the interfaces.
mpol_parse_str() in mempolicy.c is changed to return, via a pointer to a
pointer arg, a struct mempolicy pointer on success. For MPOL_DEFAULT, the
returned pointer is NULL. Further, mpol_parse_str() now takes a 'no_context'
argument that causes the input nodemask to be stored in the w.user_nodemask of
the created mempolicy for use when the mempolicy is installed in a tmpfs inode
shared policy tree. At that time, any cpuset contextualization is applied to
the original input nodemask. This preserves the previous behavior where the
input nodemask was stored in the superblock. We can think of the returned
mempolicy as "context free".
Because mpol_parse_str() is now calling mpol_new(), we can remove from
mpol_to_str() the semantic checks that mpol_new() already performs.
Add 'no_context' parameter to mpol_to_str() to specify that it should format
the nodemask in w.user_nodemask for 'bind' and 'interleave' policies.
Change mpol_shared_policy_init() to take a pointer to a "context free" struct
mempolicy and to create a new, "contextualized" mempolicy using the mode,
mode_flags and user_nodemask from the input mempolicy.
Note: we know that the mempolicy passed to mpol_to_str() or
mpol_shared_policy_init() from a tmpfs superblock is "context free". This
is currently the only instance thereof. However, if we found more uses for
this concept, and introduced any ambiguity as to whether a mempolicy was
context free or not, we could add another internal mode flag to identify
context free mempolicies. Then, we could remove the 'no_context' argument
from mpol_to_str().
Added shmem_get_sbmpol() to return a reference counted superblock mempolicy,
if one exists, to pass to mpol_shared_policy_init(). We must add the
reference under the sb stat_lock to prevent races with replacement of the mpol
by remount. This reference is removed in mpol_shared_policy_init().
[akpm@linux-foundation.org: build fix]
[akpm@linux-foundation.org: another build fix]
[akpm@linux-foundation.org: yet another build fix]
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:26 -06:00
|
|
|
/**
|
|
|
|
* mpol_to_str - format a mempolicy structure for printing
|
|
|
|
* @buffer: to contain formatted mempolicy string
|
|
|
|
* @maxlen: length of @buffer
|
|
|
|
* @pol: pointer to mempolicy to be formatted
|
|
|
|
*
|
2013-11-12 16:07:28 -07:00
|
|
|
* Convert @pol into a string. If @buffer is too short, truncate the string.
|
|
|
|
* Recommend a @maxlen of at least 32 for the longest mode, "interleave", the
|
|
|
|
* longest flag, "relative", and to display at least a few node ids.
|
2006-01-08 02:01:02 -07:00
|
|
|
*/
|
2013-11-12 16:07:28 -07:00
|
|
|
void mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol)
|
2006-01-08 02:01:02 -07:00
|
|
|
{
|
|
|
|
char *p = buffer;
|
2013-11-12 16:07:28 -07:00
|
|
|
nodemask_t nodes = NODE_MASK_NONE;
|
|
|
|
unsigned short mode = MPOL_DEFAULT;
|
|
|
|
unsigned short flags = 0;
|
mempolicy: clean-up mpol-to-str() mempolicy formatting
mpol-to-str() formats memory policies into printable strings. Currently this
is only used to display "numa_maps". A subsequent patch will use
mpol_to_str() for formatting tmpfs [shmem] mpol mount options, allowing us to
remove essentially duplicate code in mm/shmem.c. This patch cleans up
mpol_to_str() generally and in preparation for that patch.
1) show_numa_maps() is not checking the return code from mpol_to_str().
There's not a lot we can do in this context if mpol_to_str() did return the
error [insufficient space in buffer]. Proposed "solution": just check,
under DEBUG_VM, that callers are providing sufficient buffer space for the
policy, flags, and a few nodes. This way, we'll get some display.
show_numa_maps() is providing a 50-byte buffer, so it won't trip this
check. 50-bytes should be sufficient unless one has a large number of
nodes in a very sparse nodemask.
2) The display of the new mode flags ["static" & "relative"] was set up to
display multiple flags, separated by a "bar" '|'. However, this support is
incomplete--e.g., need_bar was never incremented; and currently, these two
flags are mutually exclusive. So remove the "bar" support, for now, and
only display one flag.
3) Use snprint() to format flags, so as not to overflow the buffer. Not
that it's ever happed, AFAIK.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:22 -06:00
|
|
|
|
2014-01-30 16:46:08 -07:00
|
|
|
if (pol && pol != &default_policy && !(pol->flags & MPOL_F_MORON)) {
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
mode = pol->mode;
|
2013-11-12 16:07:28 -07:00
|
|
|
flags = pol->flags;
|
|
|
|
}
|
mempolicy: use MPOL_PREFERRED for system-wide default policy
Currently, when one specifies MPOL_DEFAULT via a NUMA memory policy API
[set_mempolicy(), mbind() and internal versions], the kernel simply installs a
NULL struct mempolicy pointer in the appropriate context: task policy, vma
policy, or shared policy. This causes any use of that policy to "fall back"
to the next most specific policy scope.
The only use of MPOL_DEFAULT to mean "local allocation" is in the system
default policy. This requires extra checks/cases for MPOL_DEFAULT in many
mempolicy.c functions.
There is another, "preferred" way to specify local allocation via the APIs.
That is using the MPOL_PREFERRED policy mode with an empty nodemask.
Internally, the empty nodemask gets converted to a preferred_node id of '-1'.
All internal usage of MPOL_PREFERRED will convert the '-1' to the id of the
node local to the cpu where the allocation occurs.
System default policy, except during boot, is hard-coded to "local
allocation". By using the MPOL_PREFERRED mode with a negative value of
preferred node for system default policy, MPOL_DEFAULT will never occur in the
'policy' member of a struct mempolicy. Thus, we can remove all checks for
MPOL_DEFAULT when converting policy to a node id/zonelist in the allocation
paths.
In slab_node() return local node id when policy pointer is NULL. No need to
set a pol value to take the switch default. Replace switch default with
BUG()--i.e., shouldn't happen.
With this patch MPOL_DEFAULT is only used in the APIs, including internal
calls to do_set_mempolicy() and in the display of policy in
/proc/<pid>/numa_maps. It always means "fall back" to the the next most
specific policy scope. This simplifies the description of memory policies
quite a bit, with no visible change in behavior.
get_mempolicy() continues to return MPOL_DEFAULT and an empty nodemask when
the requested policy [task or vma/shared] is NULL. These are the values one
would supply via set_mempolicy() or mbind() to achieve that condition--default
behavior.
This patch updates Documentation to reflect this change.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:18 -06:00
|
|
|
|
2006-01-08 02:01:02 -07:00
|
|
|
switch (mode) {
|
|
|
|
case MPOL_DEFAULT:
|
|
|
|
break;
|
|
|
|
case MPOL_PREFERRED:
|
2008-04-28 03:13:21 -06:00
|
|
|
if (flags & MPOL_F_LOCAL)
|
tmpfs mempolicy: fix /proc/mounts corrupting memory
Recently I suggested using "mount -o remount,mpol=local /tmp" in NUMA
mempolicy testing. Very nasty. Reading /proc/mounts, /proc/pid/mounts
or /proc/pid/mountinfo may then corrupt one bit of kernel memory, often
in a page table (causing "Bad swap" or "Bad page map" warning or "Bad
pagetable" oops), sometimes in a vm_area_struct or rbnode or somewhere
worse. "mpol=prefer" and "mpol=prefer:Node" are equally toxic.
Recent NUMA enhancements are not to blame: this dates back to 2.6.35,
when commit e17f74af351c "mempolicy: don't call mpol_set_nodemask() when
no_context" skipped mpol_parse_str()'s call to mpol_set_nodemask(),
which used to initialize v.preferred_node, or set MPOL_F_LOCAL in flags.
With slab poisoning, you can then rely on mpol_to_str() to set the bit
for node 0x6b6b, probably in the next page above the caller's stack.
mpol_parse_str() is only called from shmem_parse_options(): no_context
is always true, so call it unused for now, and remove !no_context code.
Set v.nodes or v.preferred_node or MPOL_F_LOCAL as mpol_to_str() might
expect. Then mpol_to_str() can ignore its no_context argument also,
the mpol being appropriately initialized whether contextualized or not.
Rename its no_context unused too, and let subsequent patch remove them
(that's not needed for stable backporting, which would involve rejects).
I don't understand why MPOL_LOCAL is described as a pseudo-policy:
it's a reasonable policy which suffers from a confusing implementation
in terms of MPOL_PREFERRED with MPOL_F_LOCAL. I believe this would be
much more robust if MPOL_LOCAL were recognized in switch statements
throughout, MPOL_F_LOCAL deleted, and MPOL_PREFERRED use the (possibly
empty) nodes mask like everyone else, instead of its preferred_node
variant (I presume an optimization from the days before MPOL_LOCAL).
But that would take me too long to get right and fully tested.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-02 03:01:33 -07:00
|
|
|
mode = MPOL_LOCAL;
|
mempolicy: mPOL_PREFERRED cleanups for "local allocation"
Here are a couple of "cleanups" for MPOL_PREFERRED behavior when
v.preferred_node < 0 -- i.e., "local allocation":
1) [do_]get_mempolicy() calls the now renamed get_policy_nodemask()
to fetch the nodemask associated with a policy. Currently,
get_policy_nodemask() returns the set of nodes with memory, when
the policy 'mode' is 'PREFERRED, and the preferred_node is < 0.
Change to return an empty nodemask, as this is what was specified
to achieve "local allocation".
2) When a task is moved into a [new] cpuset, mpol_rebind_policy() is
called to adjust any task and vma policy nodes to be valid in the
new cpuset. However, when the policy is MPOL_PREFERRED, and the
preferred_node is <0, no rebind is necessary. The "local allocation"
indication is valid in any cpuset. Existing code will "do the right
thing" because node_remap() will just return the argument node when
it is outside of the valid range of node ids. However, I think it is
clearer and cleaner to skip the remap explicitly in this case.
3) mpol_to_str() produces a printable, "human readable" string from a
struct mempolicy. For MPOL_PREFERRED with preferred_node <0, show
"local", as this indicates local allocation, as the task migrates
among nodes. Note that this matches the usage of "local allocation"
in libnuma() and numactl. Without this change, I believe that node_set()
[via set_bit()] will set bit 31, resulting in a misleading display.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:20 -06:00
|
|
|
else
|
2008-04-28 03:13:21 -06:00
|
|
|
node_set(pol->v.preferred_node, nodes);
|
2006-01-08 02:01:02 -07:00
|
|
|
break;
|
|
|
|
case MPOL_BIND:
|
|
|
|
case MPOL_INTERLEAVE:
|
tmpfs mempolicy: fix /proc/mounts corrupting memory
Recently I suggested using "mount -o remount,mpol=local /tmp" in NUMA
mempolicy testing. Very nasty. Reading /proc/mounts, /proc/pid/mounts
or /proc/pid/mountinfo may then corrupt one bit of kernel memory, often
in a page table (causing "Bad swap" or "Bad page map" warning or "Bad
pagetable" oops), sometimes in a vm_area_struct or rbnode or somewhere
worse. "mpol=prefer" and "mpol=prefer:Node" are equally toxic.
Recent NUMA enhancements are not to blame: this dates back to 2.6.35,
when commit e17f74af351c "mempolicy: don't call mpol_set_nodemask() when
no_context" skipped mpol_parse_str()'s call to mpol_set_nodemask(),
which used to initialize v.preferred_node, or set MPOL_F_LOCAL in flags.
With slab poisoning, you can then rely on mpol_to_str() to set the bit
for node 0x6b6b, probably in the next page above the caller's stack.
mpol_parse_str() is only called from shmem_parse_options(): no_context
is always true, so call it unused for now, and remove !no_context code.
Set v.nodes or v.preferred_node or MPOL_F_LOCAL as mpol_to_str() might
expect. Then mpol_to_str() can ignore its no_context argument also,
the mpol being appropriately initialized whether contextualized or not.
Rename its no_context unused too, and let subsequent patch remove them
(that's not needed for stable backporting, which would involve rejects).
I don't understand why MPOL_LOCAL is described as a pseudo-policy:
it's a reasonable policy which suffers from a confusing implementation
in terms of MPOL_PREFERRED with MPOL_F_LOCAL. I believe this would be
much more robust if MPOL_LOCAL were recognized in switch statements
throughout, MPOL_F_LOCAL deleted, and MPOL_PREFERRED use the (possibly
empty) nodes mask like everyone else, instead of its preferred_node
variant (I presume an optimization from the days before MPOL_LOCAL).
But that would take me too long to get right and fully tested.
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-01-02 03:01:33 -07:00
|
|
|
nodes = pol->v.nodes;
|
2006-01-08 02:01:02 -07:00
|
|
|
break;
|
|
|
|
default:
|
2013-11-12 16:07:28 -07:00
|
|
|
WARN_ON_ONCE(1);
|
|
|
|
snprintf(p, maxlen, "unknown");
|
|
|
|
return;
|
2006-01-08 02:01:02 -07:00
|
|
|
}
|
|
|
|
|
2013-11-21 15:32:06 -07:00
|
|
|
p += snprintf(p, maxlen, "%s", policy_modes[mode]);
|
2006-01-08 02:01:02 -07:00
|
|
|
|
2008-04-28 03:13:21 -06:00
|
|
|
if (flags & MPOL_MODE_FLAGS) {
|
2013-11-12 16:07:28 -07:00
|
|
|
p += snprintf(p, buffer + maxlen - p, "=");
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
|
mempolicy: clean-up mpol-to-str() mempolicy formatting
mpol-to-str() formats memory policies into printable strings. Currently this
is only used to display "numa_maps". A subsequent patch will use
mpol_to_str() for formatting tmpfs [shmem] mpol mount options, allowing us to
remove essentially duplicate code in mm/shmem.c. This patch cleans up
mpol_to_str() generally and in preparation for that patch.
1) show_numa_maps() is not checking the return code from mpol_to_str().
There's not a lot we can do in this context if mpol_to_str() did return the
error [insufficient space in buffer]. Proposed "solution": just check,
under DEBUG_VM, that callers are providing sufficient buffer space for the
policy, flags, and a few nodes. This way, we'll get some display.
show_numa_maps() is providing a 50-byte buffer, so it won't trip this
check. 50-bytes should be sufficient unless one has a large number of
nodes in a very sparse nodemask.
2) The display of the new mode flags ["static" & "relative"] was set up to
display multiple flags, separated by a "bar" '|'. However, this support is
incomplete--e.g., need_bar was never incremented; and currently, these two
flags are mutually exclusive. So remove the "bar" support, for now, and
only display one flag.
3) Use snprint() to format flags, so as not to overflow the buffer. Not
that it's ever happed, AFAIK.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:22 -06:00
|
|
|
/*
|
|
|
|
* Currently, the only defined flags are mutually exclusive
|
|
|
|
*/
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
if (flags & MPOL_F_STATIC_NODES)
|
mempolicy: clean-up mpol-to-str() mempolicy formatting
mpol-to-str() formats memory policies into printable strings. Currently this
is only used to display "numa_maps". A subsequent patch will use
mpol_to_str() for formatting tmpfs [shmem] mpol mount options, allowing us to
remove essentially duplicate code in mm/shmem.c. This patch cleans up
mpol_to_str() generally and in preparation for that patch.
1) show_numa_maps() is not checking the return code from mpol_to_str().
There's not a lot we can do in this context if mpol_to_str() did return the
error [insufficient space in buffer]. Proposed "solution": just check,
under DEBUG_VM, that callers are providing sufficient buffer space for the
policy, flags, and a few nodes. This way, we'll get some display.
show_numa_maps() is providing a 50-byte buffer, so it won't trip this
check. 50-bytes should be sufficient unless one has a large number of
nodes in a very sparse nodemask.
2) The display of the new mode flags ["static" & "relative"] was set up to
display multiple flags, separated by a "bar" '|'. However, this support is
incomplete--e.g., need_bar was never incremented; and currently, these two
flags are mutually exclusive. So remove the "bar" support, for now, and
only display one flag.
3) Use snprint() to format flags, so as not to overflow the buffer. Not
that it's ever happed, AFAIK.
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:13:22 -06:00
|
|
|
p += snprintf(p, buffer + maxlen - p, "static");
|
|
|
|
else if (flags & MPOL_F_RELATIVE_NODES)
|
|
|
|
p += snprintf(p, buffer + maxlen - p, "relative");
|
mempolicy: add MPOL_F_STATIC_NODES flag
Add an optional mempolicy mode flag, MPOL_F_STATIC_NODES, that suppresses the
node remap when the policy is rebound.
Adds another member to struct mempolicy, nodemask_t user_nodemask, as part of
a union with cpuset_mems_allowed:
struct mempolicy {
...
union {
nodemask_t cpuset_mems_allowed;
nodemask_t user_nodemask;
} w;
}
that stores the the nodemask that the user passed when he or she created the
mempolicy via set_mempolicy() or mbind(). When using MPOL_F_STATIC_NODES,
which is passed with any mempolicy mode, the user's passed nodemask
intersected with the VMA or task's allowed nodes is always used when
determining the preferred node, setting the MPOL_BIND zonelist, or creating
the interleave nodemask. This happens whenever the policy is rebound,
including when a task's cpuset assignment changes or the cpuset's mems are
changed.
This creates an interesting side-effect in that it allows the mempolicy
"intent" to lie dormant and uneffected until it has access to the node(s) that
it desires. For example, if you currently ask for an interleaved policy over
a set of nodes that you do not have access to, the mempolicy is not created
and the task continues to use the previous policy. With this change, however,
it is possible to create the same mempolicy; it is only effected when access
to nodes in the nodemask is acquired.
It is also possible to mount tmpfs with the static nodemask behavior when
specifying a node or nodemask. To do this, simply add "=static" immediately
following the mempolicy mode at mount time:
mount -o remount mpol=interleave=static:1-3
Also removes mpol_check_policy() and folds its logic into mpol_new() since it
is now obsoleted. The unused vma_mpol_equal() is also removed.
Cc: Paul Jackson <pj@sgi.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-04-28 03:12:27 -06:00
|
|
|
}
|
|
|
|
|
2015-02-13 15:38:02 -07:00
|
|
|
if (!nodes_empty(nodes))
|
|
|
|
p += scnprintf(p, buffer + maxlen - p, ":%*pbl",
|
|
|
|
nodemask_pr_args(&nodes));
|
2006-01-08 02:01:02 -07:00
|
|
|
}
|