License cleanup: add SPDX GPL-2.0 license identifier to files with no license
Many source files in the tree are missing licensing information, which
makes it harder for compliance tools to determine the correct license.
By default all files without license information are under the default
license of the kernel, which is GPL version 2.
Update the files which contain no license information with the 'GPL-2.0'
SPDX license identifier. The SPDX identifier is a legally binding
shorthand, which can be used instead of the full boiler plate text.
This patch is based on work done by Thomas Gleixner and Kate Stewart and
Philippe Ombredanne.
How this work was done:
Patches were generated and checked against linux-4.14-rc6 for a subset of
the use cases:
- file had no licensing information it it.
- file was a */uapi/* one with no licensing information in it,
- file was a */uapi/* one with existing licensing information,
Further patches will be generated in subsequent months to fix up cases
where non-standard license headers were used, and references to license
had to be inferred by heuristics based on keywords.
The analysis to determine which SPDX License Identifier to be applied to
a file was done in a spreadsheet of side by side results from of the
output of two independent scanners (ScanCode & Windriver) producing SPDX
tag:value files created by Philippe Ombredanne. Philippe prepared the
base worksheet, and did an initial spot review of a few 1000 files.
The 4.13 kernel was the starting point of the analysis with 60,537 files
assessed. Kate Stewart did a file by file comparison of the scanner
results in the spreadsheet to determine which SPDX license identifier(s)
to be applied to the file. She confirmed any determination that was not
immediately clear with lawyers working with the Linux Foundation.
Criteria used to select files for SPDX license identifier tagging was:
- Files considered eligible had to be source code files.
- Make and config files were included as candidates if they contained >5
lines of source
- File already had some variant of a license header in it (even if <5
lines).
All documentation files were explicitly excluded.
The following heuristics were used to determine which SPDX license
identifiers to apply.
- when both scanners couldn't find any license traces, file was
considered to have no license information in it, and the top level
COPYING file license applied.
For non */uapi/* files that summary was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 11139
and resulted in the first patch in this series.
If that file was a */uapi/* path one, it was "GPL-2.0 WITH
Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 WITH Linux-syscall-note 930
and resulted in the second patch in this series.
- if a file had some form of licensing information in it, and was one
of the */uapi/* ones, it was denoted with the Linux-syscall-note if
any GPL family license was found in the file or had no licensing in
it (per prior point). Results summary:
SPDX license identifier # files
---------------------------------------------------|------
GPL-2.0 WITH Linux-syscall-note 270
GPL-2.0+ WITH Linux-syscall-note 169
((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21
((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17
LGPL-2.1+ WITH Linux-syscall-note 15
GPL-1.0+ WITH Linux-syscall-note 14
((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5
LGPL-2.0+ WITH Linux-syscall-note 4
LGPL-2.1 WITH Linux-syscall-note 3
((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3
((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1
and that resulted in the third patch in this series.
- when the two scanners agreed on the detected license(s), that became
the concluded license(s).
- when there was disagreement between the two scanners (one detected a
license but the other didn't, or they both detected different
licenses) a manual inspection of the file occurred.
- In most cases a manual inspection of the information in the file
resulted in a clear resolution of the license that should apply (and
which scanner probably needed to revisit its heuristics).
- When it was not immediately clear, the license identifier was
confirmed with lawyers working with the Linux Foundation.
- If there was any question as to the appropriate license identifier,
the file was flagged for further research and to be revisited later
in time.
In total, over 70 hours of logged manual review was done on the
spreadsheet to determine the SPDX license identifiers to apply to the
source files by Kate, Philippe, Thomas and, in some cases, confirmation
by lawyers working with the Linux Foundation.
Kate also obtained a third independent scan of the 4.13 code base from
FOSSology, and compared selected files where the other two scanners
disagreed against that SPDX file, to see if there was new insights. The
Windriver scanner is based on an older version of FOSSology in part, so
they are related.
Thomas did random spot checks in about 500 files from the spreadsheets
for the uapi headers and agreed with SPDX license identifier in the
files he inspected. For the non-uapi files Thomas did random spot checks
in about 15000 files.
In initial set of patches against 4.14-rc6, 3 files were found to have
copy/paste license identifier errors, and have been fixed to reflect the
correct identifier.
Additionally Philippe spent 10 hours this week doing a detailed manual
inspection and review of the 12,461 patched files from the initial patch
version early this week with:
- a full scancode scan run, collecting the matched texts, detected
license ids and scores
- reviewing anything where there was a license detected (about 500+
files) to ensure that the applied SPDX license was correct
- reviewing anything where there was no detection but the patch license
was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied
SPDX license was correct
This produced a worksheet with 20 files needing minor correction. This
worksheet was then exported into 3 different .csv files for the
different types of files to be modified.
These .csv files were then reviewed by Greg. Thomas wrote a script to
parse the csv files and add the proper SPDX tag to the file, in the
format that the file expected. This script was further refined by Greg
based on the output to detect more types of files automatically and to
distinguish between header and source .c files (which need different
comment types.) Finally Greg ran the script using the .csv files to
generate the patches.
Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org>
Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 08:07:57 -06:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* linux/ipc/sem.c
|
|
|
|
* Copyright (C) 1992 Krishna Balasubramanian
|
|
|
|
* Copyright (C) 1995 Eric Schenk, Bruno Haible
|
|
|
|
*
|
|
|
|
* /proc/sysvipc/sem support (c) 1999 Dragos Acostachioaie <dragos@iname.com>
|
|
|
|
*
|
|
|
|
* SMP-threaded, sysctl's added
|
2006-01-14 18:43:54 -07:00
|
|
|
* (c) 1999 Manfred Spraul <manfred@colorfullife.com>
|
2005-04-16 16:20:36 -06:00
|
|
|
* Enforced range limit on SEM_UNDO
|
2009-01-05 07:06:29 -07:00
|
|
|
* (c) 2001 Red Hat Inc
|
2005-04-16 16:20:36 -06:00
|
|
|
* Lockless wakeup
|
|
|
|
* (c) 2003 Manfred Spraul <manfred@colorfullife.com>
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
* (c) 2016 Davidlohr Bueso <dave@stgolabs.net>
|
2010-05-26 15:43:43 -06:00
|
|
|
* Further wakeup optimizations, documentation
|
|
|
|
* (c) 2010 Manfred Spraul <manfred@colorfullife.com>
|
2006-04-02 15:07:33 -06:00
|
|
|
*
|
|
|
|
* support for audit of ipc object properties and permission changes
|
|
|
|
* Dustin Kirkland <dustin.kirkland@us.ibm.com>
|
2006-10-02 03:18:22 -06:00
|
|
|
*
|
|
|
|
* namespaces support
|
|
|
|
* OpenVZ, SWsoft Inc.
|
|
|
|
* Pavel Emelianov <xemul@openvz.org>
|
2010-05-26 15:43:43 -06:00
|
|
|
*
|
|
|
|
* Implementation notes: (May 2010)
|
|
|
|
* This file implements System V semaphores.
|
|
|
|
*
|
|
|
|
* User space visible behavior:
|
|
|
|
* - FIFO ordering for semop() operations (just FIFO, not starvation
|
|
|
|
* protection)
|
|
|
|
* - multiple semaphore operations that alter the same semaphore in
|
|
|
|
* one semop() are handled.
|
|
|
|
* - sem_ctime (time of last semctl()) is updated in the IPC_SET, SETVAL and
|
|
|
|
* SETALL calls.
|
|
|
|
* - two Linux specific semctl() commands: SEM_STAT, SEM_INFO.
|
|
|
|
* - undo adjustments at process exit are limited to 0..SEMVMX.
|
|
|
|
* - namespace are supported.
|
|
|
|
* - SEMMSL, SEMMNS, SEMOPM and SEMMNI can be configured at runtine by writing
|
|
|
|
* to /proc/sys/kernel/sem.
|
|
|
|
* - statistics about the usage are reported in /proc/sysvipc/sem.
|
|
|
|
*
|
|
|
|
* Internals:
|
|
|
|
* - scalability:
|
|
|
|
* - all global variables are read-mostly.
|
|
|
|
* - semop() calls and semctl(RMID) are synchronized by RCU.
|
|
|
|
* - most operations do write operations (actually: spin_lock calls) to
|
|
|
|
* the per-semaphore array structure.
|
|
|
|
* Thus: Perfect SMP scaling between independent semaphore arrays.
|
|
|
|
* If multiple semaphores in one array are used, then cache line
|
|
|
|
* trashing on the semaphore array spinlock will limit the scaling.
|
2014-06-06 15:37:48 -06:00
|
|
|
* - semncnt and semzcnt are calculated on demand in count_semcnt()
|
2010-05-26 15:43:43 -06:00
|
|
|
* - the task that performs a successful semop() scans the list of all
|
|
|
|
* sleeping tasks and completes any pending operations that can be fulfilled.
|
|
|
|
* Semaphores are actively given to waiting tasks (necessary for FIFO).
|
|
|
|
* (see update_queue())
|
|
|
|
* - To improve the scalability, the actual wake-up calls are performed after
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
* dropping all locks. (see wake_up_sem_queue_prepare())
|
2010-05-26 15:43:43 -06:00
|
|
|
* - All work is done by the waker, the woken up task does not have to do
|
|
|
|
* anything - not even acquiring a lock or dropping a refcount.
|
|
|
|
* - A woken up task may not even touch the semaphore array anymore, it may
|
|
|
|
* have been destroyed already by a semctl(RMID).
|
|
|
|
* - UNDO values are stored in an array (one per process and per
|
|
|
|
* semaphore array, lazily allocated). For backwards compatibility, multiple
|
|
|
|
* modes for the UNDO variables are supported (per process, per thread)
|
|
|
|
* (see copy_semundo, CLONE_SYSVSEM)
|
|
|
|
* - There are two lists of the pending operations: a per-array list
|
|
|
|
* and per-semaphore list (stored in the array). This allows to achieve FIFO
|
|
|
|
* ordering without always scanning all pending operations.
|
|
|
|
* The worst-case behavior is nevertheless O(N^2) for N wakeups.
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
|
|
|
|
2018-04-13 05:58:23 -06:00
|
|
|
#include <linux/compat.h>
|
2005-04-16 16:20:36 -06:00
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/spinlock.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/proc_fs.h>
|
|
|
|
#include <linux/time.h>
|
|
|
|
#include <linux/security.h>
|
|
|
|
#include <linux/syscalls.h>
|
|
|
|
#include <linux/audit.h>
|
2006-01-11 13:17:46 -07:00
|
|
|
#include <linux/capability.h>
|
2005-09-06 16:17:10 -06:00
|
|
|
#include <linux/seq_file.h>
|
2007-10-19 00:40:54 -06:00
|
|
|
#include <linux/rwsem.h>
|
2006-10-02 03:18:22 -06:00
|
|
|
#include <linux/nsproxy.h>
|
2008-02-08 05:18:22 -07:00
|
|
|
#include <linux/ipc_namespace.h>
|
2017-02-01 08:36:40 -07:00
|
|
|
#include <linux/sched/wake_q.h>
|
2018-06-14 16:27:51 -06:00
|
|
|
#include <linux/nospec.h>
|
2018-06-17 20:52:50 -06:00
|
|
|
#include <linux/rhashtable.h>
|
2006-03-26 02:37:17 -07:00
|
|
|
|
2014-06-06 15:37:37 -06:00
|
|
|
#include <linux/uaccess.h>
|
2005-04-16 16:20:36 -06:00
|
|
|
#include "util.h"
|
|
|
|
|
2018-03-22 20:30:56 -06:00
|
|
|
/* One semaphore structure for each semaphore in the system. */
|
|
|
|
struct sem {
|
|
|
|
int semval; /* current value */
|
|
|
|
/*
|
|
|
|
* PID of the process that last modified the semaphore. For
|
|
|
|
* Linux, specifically these are:
|
|
|
|
* - semop
|
|
|
|
* - semctl, via SETVAL and SETALL.
|
|
|
|
* - at task exit when performing undo adjustments (see exit_sem).
|
|
|
|
*/
|
2018-03-23 00:11:29 -06:00
|
|
|
struct pid *sempid;
|
2018-03-22 20:30:56 -06:00
|
|
|
spinlock_t lock; /* spinlock for fine-grained semtimedop */
|
|
|
|
struct list_head pending_alter; /* pending single-sop operations */
|
|
|
|
/* that alter the semaphore */
|
|
|
|
struct list_head pending_const; /* pending single-sop operations */
|
|
|
|
/* that do not alter the semaphore*/
|
2018-04-12 07:19:44 -06:00
|
|
|
time64_t sem_otime; /* candidate for sem_otime */
|
2018-03-22 20:30:56 -06:00
|
|
|
} ____cacheline_aligned_in_smp;
|
|
|
|
|
|
|
|
/* One sem_array data structure for each set of semaphores in the system. */
|
|
|
|
struct sem_array {
|
|
|
|
struct kern_ipc_perm sem_perm; /* permissions .. see ipc.h */
|
|
|
|
time64_t sem_ctime; /* create/last semctl() time */
|
|
|
|
struct list_head pending_alter; /* pending operations */
|
|
|
|
/* that alter the array */
|
|
|
|
struct list_head pending_const; /* pending complex operations */
|
|
|
|
/* that do not alter semvals */
|
|
|
|
struct list_head list_id; /* undo requests on this array */
|
|
|
|
int sem_nsems; /* no. of semaphores in array */
|
|
|
|
int complex_count; /* pending complex operations */
|
|
|
|
unsigned int use_global_lock;/* >0: global lock required */
|
|
|
|
|
|
|
|
struct sem sems[];
|
|
|
|
} __randomize_layout;
|
2011-11-02 14:38:54 -06:00
|
|
|
|
|
|
|
/* One queue for each sleeping process in the system. */
|
|
|
|
struct sem_queue {
|
|
|
|
struct list_head list; /* queue of pending operations */
|
|
|
|
struct task_struct *sleeper; /* this process */
|
|
|
|
struct sem_undo *undo; /* undo structure */
|
2018-03-23 00:11:29 -06:00
|
|
|
struct pid *pid; /* process id of requesting process */
|
2011-11-02 14:38:54 -06:00
|
|
|
int status; /* completion status of operation */
|
|
|
|
struct sembuf *sops; /* array of pending operations */
|
2014-06-06 15:37:49 -06:00
|
|
|
struct sembuf *blocking; /* the operation that blocked */
|
2011-11-02 14:38:54 -06:00
|
|
|
int nsops; /* number of operations */
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
bool alter; /* does *sops alter the array? */
|
|
|
|
bool dupsop; /* sops on more than one sem_num */
|
2011-11-02 14:38:54 -06:00
|
|
|
};
|
|
|
|
|
|
|
|
/* Each task has a list of undo requests. They are executed automatically
|
|
|
|
* when the process exits.
|
|
|
|
*/
|
|
|
|
struct sem_undo {
|
|
|
|
struct list_head list_proc; /* per-process list: *
|
|
|
|
* all undos from one process
|
|
|
|
* rcu protected */
|
|
|
|
struct rcu_head rcu; /* rcu struct for sem_undo */
|
|
|
|
struct sem_undo_list *ulp; /* back ptr to sem_undo_list */
|
|
|
|
struct list_head list_id; /* per semaphore array list:
|
|
|
|
* all undos for one array */
|
|
|
|
int semid; /* semaphore set identifier */
|
|
|
|
short *semadj; /* array of adjustments */
|
|
|
|
/* one per semaphore */
|
|
|
|
};
|
|
|
|
|
|
|
|
/* sem_undo_list controls shared access to the list of sem_undo structures
|
|
|
|
* that may be shared among all a CLONE_SYSVSEM task group.
|
|
|
|
*/
|
|
|
|
struct sem_undo_list {
|
2017-09-08 17:17:42 -06:00
|
|
|
refcount_t refcnt;
|
2011-11-02 14:38:54 -06:00
|
|
|
spinlock_t lock;
|
|
|
|
struct list_head list_proc;
|
|
|
|
};
|
|
|
|
|
|
|
|
|
2008-02-08 05:18:57 -07:00
|
|
|
#define sem_ids(ns) ((ns)->ids[IPC_SEM_IDS])
|
2006-10-02 03:18:22 -06:00
|
|
|
|
2007-10-19 00:40:49 -06:00
|
|
|
static int newary(struct ipc_namespace *, struct ipc_params *);
|
2008-02-08 05:18:57 -07:00
|
|
|
static void freeary(struct ipc_namespace *, struct kern_ipc_perm *);
|
2005-04-16 16:20:36 -06:00
|
|
|
#ifdef CONFIG_PROC_FS
|
2005-09-06 16:17:10 -06:00
|
|
|
static int sysvipc_sem_proc_show(struct seq_file *s, void *it);
|
2005-04-16 16:20:36 -06:00
|
|
|
#endif
|
|
|
|
|
|
|
|
#define SEMMSL_FAST 256 /* 512 bytes on stack */
|
|
|
|
#define SEMOPM_FAST 64 /* ~ 372 bytes on stack */
|
|
|
|
|
2017-02-27 15:28:18 -07:00
|
|
|
/*
|
|
|
|
* Switching from the mode suitable for simple ops
|
|
|
|
* to the mode for complex ops is costly. Therefore:
|
|
|
|
* use some hysteresis
|
|
|
|
*/
|
|
|
|
#define USE_GLOBAL_LOCK_HYSTERESIS 10
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
2013-07-08 17:01:26 -06:00
|
|
|
* Locking:
|
2016-10-11 14:54:50 -06:00
|
|
|
* a) global sem_lock() for read/write
|
2005-04-16 16:20:36 -06:00
|
|
|
* sem_undo.id_next,
|
2013-07-08 17:01:26 -06:00
|
|
|
* sem_array.complex_count,
|
2016-10-11 14:54:50 -06:00
|
|
|
* sem_array.pending{_alter,_const},
|
|
|
|
* sem_array.sem_undo
|
2014-06-06 15:37:37 -06:00
|
|
|
*
|
2016-10-11 14:54:50 -06:00
|
|
|
* b) global or semaphore sem_lock() for read/write:
|
2017-07-12 15:34:38 -06:00
|
|
|
* sem_array.sems[i].pending_{const,alter}:
|
2016-10-11 14:54:50 -06:00
|
|
|
*
|
|
|
|
* c) special:
|
|
|
|
* sem_undo_list.list_proc:
|
|
|
|
* * undo_list->lock for write
|
|
|
|
* * rcu for read
|
2017-02-27 15:28:18 -07:00
|
|
|
* use_global_lock:
|
|
|
|
* * global sem_lock() for write
|
|
|
|
* * either local or global sem_lock() for read.
|
|
|
|
*
|
|
|
|
* Memory ordering:
|
|
|
|
* Most ordering is enforced by using spin_lock() and spin_unlock().
|
2020-02-03 18:34:42 -07:00
|
|
|
*
|
|
|
|
* Exceptions:
|
|
|
|
* 1) use_global_lock: (SEM_BARRIER_1)
|
2017-02-27 15:28:18 -07:00
|
|
|
* Setting it from non-zero to 0 is a RELEASE, this is ensured by
|
2020-02-03 18:34:42 -07:00
|
|
|
* using smp_store_release(): Immediately after setting it to 0,
|
|
|
|
* a simple op can start.
|
2017-02-27 15:28:18 -07:00
|
|
|
* Testing if it is non-zero is an ACQUIRE, this is ensured by using
|
|
|
|
* smp_load_acquire().
|
|
|
|
* Setting it from 0 to non-zero must be ordered with regards to
|
|
|
|
* this smp_load_acquire(), this is guaranteed because the smp_load_acquire()
|
|
|
|
* is inside a spin_lock() and after a write from 0 to non-zero a
|
|
|
|
* spin_lock()+spin_unlock() is done.
|
2020-02-03 18:34:42 -07:00
|
|
|
*
|
|
|
|
* 2) queue.status: (SEM_BARRIER_2)
|
|
|
|
* Initialization is done while holding sem_lock(), so no further barrier is
|
|
|
|
* required.
|
|
|
|
* Setting it to a result code is a RELEASE, this is ensured by both a
|
|
|
|
* smp_store_release() (for case a) and while holding sem_lock()
|
|
|
|
* (for case b).
|
|
|
|
* The AQUIRE when reading the result code without holding sem_lock() is
|
|
|
|
* achieved by using READ_ONCE() + smp_acquire__after_ctrl_dep().
|
|
|
|
* (case a above).
|
|
|
|
* Reading the result code while holding sem_lock() needs no further barriers,
|
|
|
|
* the locks inside sem_lock() enforce ordering (case b above)
|
|
|
|
*
|
|
|
|
* 3) current->state:
|
|
|
|
* current->state is set to TASK_INTERRUPTIBLE while holding sem_lock().
|
|
|
|
* The wakeup is handled using the wake_q infrastructure. wake_q wakeups may
|
|
|
|
* happen immediately after calling wake_q_add. As wake_q_add_safe() is called
|
|
|
|
* when holding sem_lock(), no further barriers are required.
|
|
|
|
*
|
|
|
|
* See also ipc/mqueue.c for more details on the covered races.
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
|
|
|
|
2006-10-02 03:18:22 -06:00
|
|
|
#define sc_semmsl sem_ctls[0]
|
|
|
|
#define sc_semmns sem_ctls[1]
|
|
|
|
#define sc_semopm sem_ctls[2]
|
|
|
|
#define sc_semmni sem_ctls[3]
|
|
|
|
|
2018-08-21 23:01:56 -06:00
|
|
|
void sem_init_ns(struct ipc_namespace *ns)
|
2006-10-02 03:18:22 -06:00
|
|
|
{
|
|
|
|
ns->sc_semmsl = SEMMSL;
|
|
|
|
ns->sc_semmns = SEMMNS;
|
|
|
|
ns->sc_semopm = SEMOPM;
|
|
|
|
ns->sc_semmni = SEMMNI;
|
|
|
|
ns->used_sems = 0;
|
2018-08-21 23:01:56 -06:00
|
|
|
ipc_init_ids(&ns->ids[IPC_SEM_IDS]);
|
2006-10-02 03:18:22 -06:00
|
|
|
}
|
|
|
|
|
2008-02-08 05:18:22 -07:00
|
|
|
#ifdef CONFIG_IPC_NS
|
2006-10-02 03:18:22 -06:00
|
|
|
void sem_exit_ns(struct ipc_namespace *ns)
|
|
|
|
{
|
2008-02-08 05:18:57 -07:00
|
|
|
free_ipcs(ns, &sem_ids(ns), freeary);
|
2009-12-15 17:47:27 -07:00
|
|
|
idr_destroy(&ns->ids[IPC_SEM_IDS].ipcs_idr);
|
ipc: optimize semget/shmget/msgget for lots of keys
ipc_findkey() used to scan all objects to look for the wanted key. This
is slow when using a high number of keys. This change adds an rhashtable
of kern_ipc_perm objects in ipc_ids, so that one lookup cease to be O(n).
This change gives a 865% improvement of benchmark reaim.jobs_per_min on a
56 threads Intel(R) Xeon(R) CPU E5-2695 v3 @ 2.30GHz with 256G memory [1]
Other (more micro) benchmark results, by the author: On an i5 laptop, the
following loop executed right after a reboot took, without and with this
change:
for (int i = 0, k=0x424242; i < KEYS; ++i)
semget(k++, 1, IPC_CREAT | 0600);
total total max single max single
KEYS without with call without call with
1 3.5 4.9 µs 3.5 4.9
10 7.6 8.6 µs 3.7 4.7
32 16.2 15.9 µs 4.3 5.3
100 72.9 41.8 µs 3.7 4.7
1000 5,630.0 502.0 µs * *
10000 1,340,000.0 7,240.0 µs * *
31900 17,600,000.0 22,200.0 µs * *
*: unreliable measure: high variance
The duration for a lookup-only usage was obtained by the same loop once
the keys are present:
total total max single max single
KEYS without with call without call with
1 2.1 2.5 µs 2.1 2.5
10 4.5 4.8 µs 2.2 2.3
32 13.0 10.8 µs 2.3 2.8
100 82.9 25.1 µs * 2.3
1000 5,780.0 217.0 µs * *
10000 1,470,000.0 2,520.0 µs * *
31900 17,400,000.0 7,810.0 µs * *
Finally, executing each semget() in a new process gave, when still
summing only the durations of these syscalls:
creation:
total total
KEYS without with
1 3.7 5.0 µs
10 32.9 36.7 µs
32 125.0 109.0 µs
100 523.0 353.0 µs
1000 20,300.0 3,280.0 µs
10000 2,470,000.0 46,700.0 µs
31900 27,800,000.0 219,000.0 µs
lookup-only:
total total
KEYS without with
1 2.5 2.7 µs
10 25.4 24.4 µs
32 106.0 72.6 µs
100 591.0 352.0 µs
1000 22,400.0 2,250.0 µs
10000 2,510,000.0 25,700.0 µs
31900 28,200,000.0 115,000.0 µs
[1] http://lkml.kernel.org/r/20170814060507.GE23258@yexl-desktop
Link: http://lkml.kernel.org/r/20170815194954.ck32ta2z35yuzpwp@debix
Signed-off-by: Guillaume Knispel <guillaume.knispel@supersonicimagine.com>
Reviewed-by: Marc Pardo <marc.pardo@supersonicimagine.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Kees Cook <keescook@chromium.org>
Cc: Manfred Spraul <manfred@colorfullife.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "Peter Zijlstra (Intel)" <peterz@infradead.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Serge Hallyn <serge@hallyn.com>
Cc: Andrey Vagin <avagin@openvz.org>
Cc: Guillaume Knispel <guillaume.knispel@supersonicimagine.com>
Cc: Marc Pardo <marc.pardo@supersonicimagine.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-08 17:17:55 -06:00
|
|
|
rhashtable_destroy(&ns->ids[IPC_SEM_IDS].key_ht);
|
2006-10-02 03:18:22 -06:00
|
|
|
}
|
2008-02-08 05:18:22 -07:00
|
|
|
#endif
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2018-08-21 23:01:56 -06:00
|
|
|
void __init sem_init(void)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2018-08-21 23:01:56 -06:00
|
|
|
sem_init_ns(&init_ipc_ns);
|
2005-09-06 16:17:10 -06:00
|
|
|
ipc_init_proc_interface("sysvipc/sem",
|
|
|
|
" key semid perms nsems uid gid cuid cgid otime ctime\n",
|
2006-10-02 03:18:22 -06:00
|
|
|
IPC_SEM_IDS, sysvipc_sem_proc_show);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2013-07-08 17:01:24 -06:00
|
|
|
/**
|
|
|
|
* unmerge_queues - unmerge queues, if possible.
|
|
|
|
* @sma: semaphore array
|
|
|
|
*
|
|
|
|
* The function unmerges the wait queues if complex_count is 0.
|
|
|
|
* It must be called prior to dropping the global semaphore array lock.
|
|
|
|
*/
|
|
|
|
static void unmerge_queues(struct sem_array *sma)
|
|
|
|
{
|
|
|
|
struct sem_queue *q, *tq;
|
|
|
|
|
|
|
|
/* complex operations still around? */
|
|
|
|
if (sma->complex_count)
|
|
|
|
return;
|
|
|
|
/*
|
|
|
|
* We will switch back to simple mode.
|
|
|
|
* Move all pending operation back into the per-semaphore
|
|
|
|
* queues.
|
|
|
|
*/
|
|
|
|
list_for_each_entry_safe(q, tq, &sma->pending_alter, list) {
|
|
|
|
struct sem *curr;
|
2017-07-12 15:34:38 -06:00
|
|
|
curr = &sma->sems[q->sops[0].sem_num];
|
2013-07-08 17:01:24 -06:00
|
|
|
|
|
|
|
list_add_tail(&q->list, &curr->pending_alter);
|
|
|
|
}
|
|
|
|
INIT_LIST_HEAD(&sma->pending_alter);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2014-01-27 18:07:05 -07:00
|
|
|
* merge_queues - merge single semop queues into global queue
|
2013-07-08 17:01:24 -06:00
|
|
|
* @sma: semaphore array
|
|
|
|
*
|
|
|
|
* This function merges all per-semaphore queues into the global queue.
|
|
|
|
* It is necessary to achieve FIFO ordering for the pending single-sop
|
|
|
|
* operations when a multi-semop operation must sleep.
|
|
|
|
* Only the alter operations must be moved, the const operations can stay.
|
|
|
|
*/
|
|
|
|
static void merge_queues(struct sem_array *sma)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for (i = 0; i < sma->sem_nsems; i++) {
|
2017-07-12 15:34:38 -06:00
|
|
|
struct sem *sem = &sma->sems[i];
|
2013-07-08 17:01:24 -06:00
|
|
|
|
|
|
|
list_splice_init(&sem->pending_alter, &sma->pending_alter);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
ipc: fix race with LSMs
Currently, IPC mechanisms do security and auditing related checks under
RCU. However, since security modules can free the security structure,
for example, through selinux_[sem,msg_queue,shm]_free_security(), we can
race if the structure is freed before other tasks are done with it,
creating a use-after-free condition. Manfred illustrates this nicely,
for instance with shared mem and selinux:
-> do_shmat calls rcu_read_lock()
-> do_shmat calls shm_object_check().
Checks that the object is still valid - but doesn't acquire any locks.
Then it returns.
-> do_shmat calls security_shm_shmat (e.g. selinux_shm_shmat)
-> selinux_shm_shmat calls ipc_has_perm()
-> ipc_has_perm accesses ipc_perms->security
shm_close()
-> shm_close acquires rw_mutex & shm_lock
-> shm_close calls shm_destroy
-> shm_destroy calls security_shm_free (e.g. selinux_shm_free_security)
-> selinux_shm_free_security calls ipc_free_security(&shp->shm_perm)
-> ipc_free_security calls kfree(ipc_perms->security)
This patch delays the freeing of the security structures after all RCU
readers are done. Furthermore it aligns the security life cycle with
that of the rest of IPC - freeing them based on the reference counter.
For situations where we need not free security, the current behavior is
kept. Linus states:
"... the old behavior was suspect for another reason too: having the
security blob go away from under a user sounds like it could cause
various other problems anyway, so I think the old code was at least
_prone_ to bugs even if it didn't have catastrophic behavior."
I have tested this patch with IPC testcases from LTP on both my
quad-core laptop and on a 64 core NUMA server. In both cases selinux is
enabled, and tests pass for both voluntary and forced preemption models.
While the mentioned races are theoretical (at least no one as reported
them), I wanted to make sure that this new logic doesn't break anything
we weren't aware of.
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-23 18:04:45 -06:00
|
|
|
static void sem_rcu_free(struct rcu_head *head)
|
|
|
|
{
|
2017-07-12 15:34:41 -06:00
|
|
|
struct kern_ipc_perm *p = container_of(head, struct kern_ipc_perm, rcu);
|
|
|
|
struct sem_array *sma = container_of(p, struct sem_array, sem_perm);
|
ipc: fix race with LSMs
Currently, IPC mechanisms do security and auditing related checks under
RCU. However, since security modules can free the security structure,
for example, through selinux_[sem,msg_queue,shm]_free_security(), we can
race if the structure is freed before other tasks are done with it,
creating a use-after-free condition. Manfred illustrates this nicely,
for instance with shared mem and selinux:
-> do_shmat calls rcu_read_lock()
-> do_shmat calls shm_object_check().
Checks that the object is still valid - but doesn't acquire any locks.
Then it returns.
-> do_shmat calls security_shm_shmat (e.g. selinux_shm_shmat)
-> selinux_shm_shmat calls ipc_has_perm()
-> ipc_has_perm accesses ipc_perms->security
shm_close()
-> shm_close acquires rw_mutex & shm_lock
-> shm_close calls shm_destroy
-> shm_destroy calls security_shm_free (e.g. selinux_shm_free_security)
-> selinux_shm_free_security calls ipc_free_security(&shp->shm_perm)
-> ipc_free_security calls kfree(ipc_perms->security)
This patch delays the freeing of the security structures after all RCU
readers are done. Furthermore it aligns the security life cycle with
that of the rest of IPC - freeing them based on the reference counter.
For situations where we need not free security, the current behavior is
kept. Linus states:
"... the old behavior was suspect for another reason too: having the
security blob go away from under a user sounds like it could cause
various other problems anyway, so I think the old code was at least
_prone_ to bugs even if it didn't have catastrophic behavior."
I have tested this patch with IPC testcases from LTP on both my
quad-core laptop and on a 64 core NUMA server. In both cases selinux is
enabled, and tests pass for both voluntary and forced preemption models.
While the mentioned races are theoretical (at least no one as reported
them), I wanted to make sure that this new logic doesn't break anything
we weren't aware of.
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-23 18:04:45 -06:00
|
|
|
|
2018-03-22 19:52:43 -06:00
|
|
|
security_sem_free(&sma->sem_perm);
|
2017-07-12 15:35:31 -06:00
|
|
|
kvfree(sma);
|
ipc: fix race with LSMs
Currently, IPC mechanisms do security and auditing related checks under
RCU. However, since security modules can free the security structure,
for example, through selinux_[sem,msg_queue,shm]_free_security(), we can
race if the structure is freed before other tasks are done with it,
creating a use-after-free condition. Manfred illustrates this nicely,
for instance with shared mem and selinux:
-> do_shmat calls rcu_read_lock()
-> do_shmat calls shm_object_check().
Checks that the object is still valid - but doesn't acquire any locks.
Then it returns.
-> do_shmat calls security_shm_shmat (e.g. selinux_shm_shmat)
-> selinux_shm_shmat calls ipc_has_perm()
-> ipc_has_perm accesses ipc_perms->security
shm_close()
-> shm_close acquires rw_mutex & shm_lock
-> shm_close calls shm_destroy
-> shm_destroy calls security_shm_free (e.g. selinux_shm_free_security)
-> selinux_shm_free_security calls ipc_free_security(&shp->shm_perm)
-> ipc_free_security calls kfree(ipc_perms->security)
This patch delays the freeing of the security structures after all RCU
readers are done. Furthermore it aligns the security life cycle with
that of the rest of IPC - freeing them based on the reference counter.
For situations where we need not free security, the current behavior is
kept. Linus states:
"... the old behavior was suspect for another reason too: having the
security blob go away from under a user sounds like it could cause
various other problems anyway, so I think the old code was at least
_prone_ to bugs even if it didn't have catastrophic behavior."
I have tested this patch with IPC testcases from LTP on both my
quad-core laptop and on a 64 core NUMA server. In both cases selinux is
enabled, and tests pass for both voluntary and forced preemption models.
While the mentioned races are theoretical (at least no one as reported
them), I wanted to make sure that this new logic doesn't break anything
we weren't aware of.
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-23 18:04:45 -06:00
|
|
|
}
|
|
|
|
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
/*
|
2016-10-11 14:54:50 -06:00
|
|
|
* Enter the mode suitable for non-simple operations:
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
* Caller must own sem_perm.lock.
|
|
|
|
*/
|
2016-10-11 14:54:50 -06:00
|
|
|
static void complexmode_enter(struct sem_array *sma)
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
struct sem *sem;
|
|
|
|
|
2017-02-27 15:28:18 -07:00
|
|
|
if (sma->use_global_lock > 0) {
|
|
|
|
/*
|
|
|
|
* We are already in global lock mode.
|
|
|
|
* Nothing to do, just reset the
|
|
|
|
* counter until we return to simple mode.
|
|
|
|
*/
|
|
|
|
sma->use_global_lock = USE_GLOBAL_LOCK_HYSTERESIS;
|
2013-09-30 14:45:06 -06:00
|
|
|
return;
|
|
|
|
}
|
2017-02-27 15:28:18 -07:00
|
|
|
sma->use_global_lock = USE_GLOBAL_LOCK_HYSTERESIS;
|
2016-10-11 14:54:50 -06:00
|
|
|
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
for (i = 0; i < sma->sem_nsems; i++) {
|
2017-07-12 15:34:38 -06:00
|
|
|
sem = &sma->sems[i];
|
2017-02-27 15:28:15 -07:00
|
|
|
spin_lock(&sem->lock);
|
|
|
|
spin_unlock(&sem->lock);
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
}
|
2016-10-11 14:54:50 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Try to leave the mode that disallows simple operations:
|
|
|
|
* Caller must own sem_perm.lock.
|
|
|
|
*/
|
|
|
|
static void complexmode_tryleave(struct sem_array *sma)
|
|
|
|
{
|
|
|
|
if (sma->complex_count) {
|
|
|
|
/* Complex ops are sleeping.
|
|
|
|
* We must stay in complex mode
|
|
|
|
*/
|
|
|
|
return;
|
|
|
|
}
|
2017-02-27 15:28:18 -07:00
|
|
|
if (sma->use_global_lock == 1) {
|
2020-02-03 18:34:42 -07:00
|
|
|
|
|
|
|
/* See SEM_BARRIER_1 for purpose/pairing */
|
2017-02-27 15:28:18 -07:00
|
|
|
smp_store_release(&sma->use_global_lock, 0);
|
|
|
|
} else {
|
|
|
|
sma->use_global_lock--;
|
|
|
|
}
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
}
|
|
|
|
|
2016-10-11 14:54:50 -06:00
|
|
|
#define SEM_GLOBAL_LOCK (-1)
|
2013-04-30 20:15:44 -06:00
|
|
|
/*
|
|
|
|
* If the request contains only one semaphore operation, and there are
|
|
|
|
* no complex transactions pending, lock only the semaphore involved.
|
|
|
|
* Otherwise, lock the entire semaphore array, since we either have
|
|
|
|
* multiple semaphores in our own semops, or we need to look at
|
|
|
|
* semaphores from other pending complex operations.
|
|
|
|
*/
|
|
|
|
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
|
|
|
|
int nsops)
|
|
|
|
{
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
struct sem *sem;
|
2018-06-14 16:27:51 -06:00
|
|
|
int idx;
|
2013-04-30 20:15:44 -06:00
|
|
|
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
if (nsops != 1) {
|
|
|
|
/* Complex operation - acquire a full lock */
|
|
|
|
ipc_lock_object(&sma->sem_perm);
|
2013-04-30 20:15:44 -06:00
|
|
|
|
2016-10-11 14:54:50 -06:00
|
|
|
/* Prevent parallel simple ops */
|
|
|
|
complexmode_enter(sma);
|
|
|
|
return SEM_GLOBAL_LOCK;
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Only one semaphore affected - try to optimize locking.
|
2016-10-11 14:54:50 -06:00
|
|
|
* Optimized locking is possible if no complex operation
|
|
|
|
* is either enqueued or processed right now.
|
|
|
|
*
|
2017-02-27 15:28:18 -07:00
|
|
|
* Both facts are tracked by use_global_mode.
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
*/
|
2018-06-14 16:27:51 -06:00
|
|
|
idx = array_index_nospec(sops->sem_num, sma->sem_nsems);
|
|
|
|
sem = &sma->sems[idx];
|
2013-04-30 20:15:44 -06:00
|
|
|
|
2016-10-11 14:54:50 -06:00
|
|
|
/*
|
2017-02-27 15:28:18 -07:00
|
|
|
* Initial check for use_global_lock. Just an optimization,
|
2016-10-11 14:54:50 -06:00
|
|
|
* no locking, no memory barrier.
|
|
|
|
*/
|
2017-02-27 15:28:18 -07:00
|
|
|
if (!sma->use_global_lock) {
|
2013-04-30 20:15:44 -06:00
|
|
|
/*
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
* It appears that no complex operation is around.
|
|
|
|
* Acquire the per-semaphore lock.
|
2013-04-30 20:15:44 -06:00
|
|
|
*/
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
spin_lock(&sem->lock);
|
|
|
|
|
2020-02-03 18:34:42 -07:00
|
|
|
/* see SEM_BARRIER_1 for purpose/pairing */
|
2017-02-27 15:28:18 -07:00
|
|
|
if (!smp_load_acquire(&sma->use_global_lock)) {
|
2016-10-11 14:54:50 -06:00
|
|
|
/* fast path successful! */
|
|
|
|
return sops->sem_num;
|
2013-04-30 20:15:44 -06:00
|
|
|
}
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
spin_unlock(&sem->lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* slow path: acquire the full lock */
|
|
|
|
ipc_lock_object(&sma->sem_perm);
|
2013-04-30 20:15:44 -06:00
|
|
|
|
2017-02-27 15:28:18 -07:00
|
|
|
if (sma->use_global_lock == 0) {
|
|
|
|
/*
|
|
|
|
* The use_global_lock mode ended while we waited for
|
|
|
|
* sma->sem_perm.lock. Thus we must switch to locking
|
|
|
|
* with sem->lock.
|
|
|
|
* Unlike in the fast path, there is no need to recheck
|
|
|
|
* sma->use_global_lock after we have acquired sem->lock:
|
|
|
|
* We own sma->sem_perm.lock, thus use_global_lock cannot
|
|
|
|
* change.
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
*/
|
|
|
|
spin_lock(&sem->lock);
|
2017-02-27 15:28:18 -07:00
|
|
|
|
ipc/sem.c: fix race in sem_lock()
The exclusion of complex operations in sem_lock() is insufficient: after
acquiring the per-semaphore lock, a simple op must first check that
sem_perm.lock is not locked and only after that test check
complex_count. The current code does it the other way around - and that
creates a race. Details are below.
The patch is a complete rewrite of sem_lock(), based in part on the code
from Mike Galbraith. It removes all gotos and all loops and thus the
risk of livelocks.
I have tested the patch (together with the next one) on my i3 laptop and
it didn't cause any problems.
The bug is probably also present in 3.10 and 3.11, but for these kernels
it might be simpler just to move the test of sma->complex_count after
the spin_is_locked() test.
Details of the bug:
Assume:
- sma->complex_count = 0.
- Thread 1: semtimedop(complex op that must sleep)
- Thread 2: semtimedop(simple op).
Pseudo-Trace:
Thread 1: sem_lock(): acquire sem_perm.lock
Thread 1: sem_lock(): check for ongoing simple ops
Nothing ongoing, thread 2 is still before sem_lock().
Thread 1: try_atomic_semop()
<<< preempted.
Thread 2: sem_lock():
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
int nsops)
{
int locknum;
again:
if (nsops == 1 && !sma->complex_count) {
struct sem *sem = sma->sem_base + sops->sem_num;
/* Lock just the semaphore we are interested in. */
spin_lock(&sem->lock);
/*
* If sma->complex_count was set while we were spinning,
* we may need to look at things we did not lock here.
*/
if (unlikely(sma->complex_count)) {
spin_unlock(&sem->lock);
goto lock_array;
}
<<<<<<<<<
<<< complex_count is still 0.
<<<
<<< Here it is preempted
<<<<<<<<<
Thread 1: try_atomic_semop() returns, notices that it must sleep.
Thread 1: increases sma->complex_count.
Thread 1: drops sem_perm.lock
Thread 2:
/*
* Another process is holding the global lock on the
* sem_array; we cannot enter our critical section,
* but have to wait for the global lock to be released.
*/
if (unlikely(spin_is_locked(&sma->sem_perm.lock))) {
spin_unlock(&sem->lock);
spin_unlock_wait(&sma->sem_perm.lock);
goto again;
}
<<< sem_perm.lock already dropped, thus no "goto again;"
locknum = sops->sem_num;
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Mike Galbraith <bitbucket@online.de>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-09-30 14:45:04 -06:00
|
|
|
ipc_unlock_object(&sma->sem_perm);
|
|
|
|
return sops->sem_num;
|
2013-04-30 20:15:44 -06:00
|
|
|
} else {
|
2017-02-27 15:28:18 -07:00
|
|
|
/*
|
|
|
|
* Not a false alarm, thus continue to use the global lock
|
|
|
|
* mode. No need for complexmode_enter(), this was done by
|
|
|
|
* the caller that has set use_global_mode to non-zero.
|
2013-04-30 20:15:44 -06:00
|
|
|
*/
|
2016-10-11 14:54:50 -06:00
|
|
|
return SEM_GLOBAL_LOCK;
|
2013-04-30 20:15:44 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void sem_unlock(struct sem_array *sma, int locknum)
|
|
|
|
{
|
2016-10-11 14:54:50 -06:00
|
|
|
if (locknum == SEM_GLOBAL_LOCK) {
|
2013-07-08 17:01:24 -06:00
|
|
|
unmerge_queues(sma);
|
2016-10-11 14:54:50 -06:00
|
|
|
complexmode_tryleave(sma);
|
2013-07-08 17:01:11 -06:00
|
|
|
ipc_unlock_object(&sma->sem_perm);
|
2013-04-30 20:15:44 -06:00
|
|
|
} else {
|
2017-07-12 15:34:38 -06:00
|
|
|
struct sem *sem = &sma->sems[locknum];
|
2013-04-30 20:15:44 -06:00
|
|
|
spin_unlock(&sem->lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-10-19 00:40:54 -06:00
|
|
|
/*
|
2013-09-11 15:26:24 -06:00
|
|
|
* sem_lock_(check_) routines are called in the paths where the rwsem
|
2007-10-19 00:40:54 -06:00
|
|
|
* is not held.
|
2013-05-04 11:47:57 -06:00
|
|
|
*
|
|
|
|
* The caller holds the RCU read lock.
|
2007-10-19 00:40:54 -06:00
|
|
|
*/
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
static inline struct sem_array *sem_obtain_object(struct ipc_namespace *ns, int id)
|
|
|
|
{
|
2015-06-30 15:58:42 -06:00
|
|
|
struct kern_ipc_perm *ipcp = ipc_obtain_object_idr(&sem_ids(ns), id);
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
|
|
|
|
if (IS_ERR(ipcp))
|
|
|
|
return ERR_CAST(ipcp);
|
|
|
|
|
|
|
|
return container_of(ipcp, struct sem_array, sem_perm);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline struct sem_array *sem_obtain_object_check(struct ipc_namespace *ns,
|
|
|
|
int id)
|
|
|
|
{
|
|
|
|
struct kern_ipc_perm *ipcp = ipc_obtain_object_check(&sem_ids(ns), id);
|
|
|
|
|
|
|
|
if (IS_ERR(ipcp))
|
|
|
|
return ERR_CAST(ipcp);
|
IPC: fix error check in all new xxx_lock() and xxx_exit_ns() functions
In the new implementation of the [sem|shm|msg]_lock[_check]() routines, we
use the return value of ipc_lock() in container_of() without any check.
But ipc_lock may return a errcode. The use of this errcode in
container_of() may alter this errcode, and we don't want this.
And in xxx_exit_ns, the pointer return by idr_find is of type 'struct
kern_ipc_per'...
Today, the code will work as is because the member used in these
container_of() is the first member of its container (offset == 0), the
errcode isn't changed then. But in the general case, we can't count on
this assumption and this may lead later to a real bug if we don't correct
this.
Again, the proposed solution is simple and correct. But, as pointed by
Nadia, with this solution, the same check will be done several times (in
all sub-callers...), what is not very funny/optimal...
Signed-off-by: Pierre Peiffer <pierre.peiffer@bull.net>
Cc: Nadia Derbey <Nadia.Derbey@bull.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-06 02:36:23 -07:00
|
|
|
|
2007-10-19 00:40:51 -06:00
|
|
|
return container_of(ipcp, struct sem_array, sem_perm);
|
2007-10-19 00:40:51 -06:00
|
|
|
}
|
|
|
|
|
2008-04-29 02:00:46 -06:00
|
|
|
static inline void sem_lock_and_putref(struct sem_array *sma)
|
|
|
|
{
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_lock(sma, NULL, -1);
|
2017-07-12 15:34:41 -06:00
|
|
|
ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
|
2008-04-29 02:00:46 -06:00
|
|
|
}
|
|
|
|
|
2007-10-19 00:40:48 -06:00
|
|
|
static inline void sem_rmid(struct ipc_namespace *ns, struct sem_array *s)
|
|
|
|
{
|
|
|
|
ipc_rmid(&sem_ids(ns), &s->sem_perm);
|
|
|
|
}
|
|
|
|
|
2017-07-12 15:35:02 -06:00
|
|
|
static struct sem_array *sem_alloc(size_t nsems)
|
|
|
|
{
|
|
|
|
struct sem_array *sma;
|
|
|
|
|
|
|
|
if (nsems > (INT_MAX - sizeof(*sma)) / sizeof(sma->sems[0]))
|
|
|
|
return NULL;
|
|
|
|
|
2019-03-07 17:30:26 -07:00
|
|
|
sma = kvzalloc(struct_size(sma, sems, nsems), GFP_KERNEL);
|
2017-07-12 15:35:02 -06:00
|
|
|
if (unlikely(!sma))
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
return sma;
|
|
|
|
}
|
|
|
|
|
2007-10-19 00:40:53 -06:00
|
|
|
/**
|
|
|
|
* newary - Create a new semaphore set
|
|
|
|
* @ns: namespace
|
|
|
|
* @params: ptr to the structure that contains key, semflg and nsems
|
|
|
|
*
|
2013-09-11 15:26:24 -06:00
|
|
|
* Called with sem_ids.rwsem held (as a writer)
|
2007-10-19 00:40:53 -06:00
|
|
|
*/
|
2007-10-19 00:40:49 -06:00
|
|
|
static int newary(struct ipc_namespace *ns, struct ipc_params *params)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
int retval;
|
|
|
|
struct sem_array *sma;
|
2007-10-19 00:40:49 -06:00
|
|
|
key_t key = params->key;
|
|
|
|
int nsems = params->u.nsems;
|
|
|
|
int semflg = params->flg;
|
2009-12-15 17:47:32 -07:00
|
|
|
int i;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (!nsems)
|
|
|
|
return -EINVAL;
|
2006-10-02 03:18:22 -06:00
|
|
|
if (ns->used_sems + nsems > ns->sc_semmns)
|
2005-04-16 16:20:36 -06:00
|
|
|
return -ENOSPC;
|
|
|
|
|
2017-07-12 15:35:02 -06:00
|
|
|
sma = sem_alloc(nsems);
|
2014-01-27 18:07:06 -07:00
|
|
|
if (!sma)
|
2005-04-16 16:20:36 -06:00
|
|
|
return -ENOMEM;
|
2014-01-27 18:07:06 -07:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
sma->sem_perm.mode = (semflg & S_IRWXUGO);
|
|
|
|
sma->sem_perm.key = key;
|
|
|
|
|
|
|
|
sma->sem_perm.security = NULL;
|
2018-03-22 19:52:43 -06:00
|
|
|
retval = security_sem_alloc(&sma->sem_perm);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (retval) {
|
2017-07-12 15:35:31 -06:00
|
|
|
kvfree(sma);
|
2005-04-16 16:20:36 -06:00
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
2013-04-30 20:15:44 -06:00
|
|
|
for (i = 0; i < nsems; i++) {
|
2017-07-12 15:34:38 -06:00
|
|
|
INIT_LIST_HEAD(&sma->sems[i].pending_alter);
|
|
|
|
INIT_LIST_HEAD(&sma->sems[i].pending_const);
|
|
|
|
spin_lock_init(&sma->sems[i].lock);
|
2013-04-30 20:15:44 -06:00
|
|
|
}
|
2009-12-15 17:47:32 -07:00
|
|
|
|
|
|
|
sma->complex_count = 0;
|
2017-02-27 15:28:18 -07:00
|
|
|
sma->use_global_lock = USE_GLOBAL_LOCK_HYSTERESIS;
|
2013-07-08 17:01:23 -06:00
|
|
|
INIT_LIST_HEAD(&sma->pending_alter);
|
|
|
|
INIT_LIST_HEAD(&sma->pending_const);
|
2008-07-25 02:48:04 -06:00
|
|
|
INIT_LIST_HEAD(&sma->list_id);
|
2005-04-16 16:20:36 -06:00
|
|
|
sma->sem_nsems = nsems;
|
2017-08-02 20:51:13 -06:00
|
|
|
sma->sem_ctime = ktime_get_real_seconds();
|
2014-12-02 16:59:34 -07:00
|
|
|
|
2017-11-17 16:31:11 -07:00
|
|
|
/* ipc_addid() locks sma upon success. */
|
2017-07-12 15:35:13 -06:00
|
|
|
retval = ipc_addid(&sem_ids(ns), &sma->sem_perm, ns->sc_semmni);
|
|
|
|
if (retval < 0) {
|
2018-08-21 23:01:29 -06:00
|
|
|
ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
|
2017-07-12 15:35:13 -06:00
|
|
|
return retval;
|
2014-12-02 16:59:34 -07:00
|
|
|
}
|
|
|
|
ns->used_sems += nsems;
|
|
|
|
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2007-10-19 00:40:48 -06:00
|
|
|
return sma->sem_perm.id;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2007-10-19 00:40:49 -06:00
|
|
|
|
2007-10-19 00:40:53 -06:00
|
|
|
/*
|
2013-09-11 15:26:24 -06:00
|
|
|
* Called with sem_ids.rwsem and ipcp locked.
|
2007-10-19 00:40:53 -06:00
|
|
|
*/
|
2020-08-11 19:37:05 -06:00
|
|
|
static int sem_more_checks(struct kern_ipc_perm *ipcp, struct ipc_params *params)
|
2007-10-19 00:40:49 -06:00
|
|
|
{
|
2007-10-19 00:40:51 -06:00
|
|
|
struct sem_array *sma;
|
|
|
|
|
|
|
|
sma = container_of(ipcp, struct sem_array, sem_perm);
|
|
|
|
if (params->u.nsems > sma->sem_nsems)
|
2007-10-19 00:40:49 -06:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2018-03-20 12:53:58 -06:00
|
|
|
long ksys_semget(key_t key, int nsems, int semflg)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-10-02 03:18:22 -06:00
|
|
|
struct ipc_namespace *ns;
|
2014-06-06 15:37:36 -06:00
|
|
|
static const struct ipc_ops sem_ops = {
|
|
|
|
.getnew = newary,
|
2018-03-23 22:41:55 -06:00
|
|
|
.associate = security_sem_associate,
|
2014-06-06 15:37:36 -06:00
|
|
|
.more_checks = sem_more_checks,
|
|
|
|
};
|
2007-10-19 00:40:49 -06:00
|
|
|
struct ipc_params sem_params;
|
2006-10-02 03:18:22 -06:00
|
|
|
|
|
|
|
ns = current->nsproxy->ipc_ns;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-10-02 03:18:22 -06:00
|
|
|
if (nsems < 0 || nsems > ns->sc_semmsl)
|
2005-04-16 16:20:36 -06:00
|
|
|
return -EINVAL;
|
2007-10-19 00:40:48 -06:00
|
|
|
|
2007-10-19 00:40:49 -06:00
|
|
|
sem_params.key = key;
|
|
|
|
sem_params.flg = semflg;
|
|
|
|
sem_params.u.nsems = nsems;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2007-10-19 00:40:49 -06:00
|
|
|
return ipcget(ns, &sem_ids(ns), &sem_ops, &sem_params);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2018-03-20 12:53:58 -06:00
|
|
|
SYSCALL_DEFINE3(semget, key_t, key, int, nsems, int, semflg)
|
|
|
|
{
|
|
|
|
return ksys_semget(key, nsems, semflg);
|
|
|
|
}
|
|
|
|
|
ipc/sem.c: avoid overflow of semop undo (semadj) value
When trying to understand semop code, I found a small mistake in the check
for semadj (undo) value overflow. The new undo value is not stored
immediately and next potential checks are done against the old value.
The failing scenario is not much practical. One semop call has to do more
operations on the same semaphore. Also semval and semadj must have
different values, so there has to be some operations without SEM_UNDO
flag. For example:
struct sembuf depositor_op[1];
struct sembuf collector_op[2];
depositor_op[0].sem_num = 0;
depositor_op[0].sem_op = 20000;
depositor_op[0].sem_flg = 0;
collector_op[0].sem_num = 0;
collector_op[0].sem_op = -10000;
collector_op[0].sem_flg = SEM_UNDO;
collector_op[1].sem_num = 0;
collector_op[1].sem_op = -10000;
collector_op[1].sem_flg = SEM_UNDO;
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 1st deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 1st collect"); return 1; }
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 2nd deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 2nd collect"); return 1; }
return 0;
It passes without error now but the semadj value has overflown in the 2nd
collector operation.
[akpm@linux-foundation.org: restore lessened scope of local `undo']
[davidlohr@hp.com: correct header comment for perform_atomic_semop]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-27 18:07:00 -07:00
|
|
|
/**
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
* perform_atomic_semop[_slow] - Attempt to perform semaphore
|
|
|
|
* operations on a given array.
|
2013-07-08 17:01:26 -06:00
|
|
|
* @sma: semaphore array
|
2014-06-06 15:37:49 -06:00
|
|
|
* @q: struct sem_queue that describes the operation
|
2013-07-08 17:01:26 -06:00
|
|
|
*
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
* Caller blocking are as follows, based the value
|
|
|
|
* indicated by the semaphore operation (sem_op):
|
|
|
|
*
|
|
|
|
* (1) >0 never blocks.
|
|
|
|
* (2) 0 (wait-for-zero operation): semval is non-zero.
|
|
|
|
* (3) <0 attempting to decrement semval to a value smaller than zero.
|
|
|
|
*
|
2013-07-08 17:01:26 -06:00
|
|
|
* Returns 0 if the operation was possible.
|
|
|
|
* Returns 1 if the operation is impossible, the caller must sleep.
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
* Returns <0 for error codes.
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
static int perform_atomic_semop_slow(struct sem_array *sma, struct sem_queue *q)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2018-03-23 00:11:29 -06:00
|
|
|
int result, sem_op, nsops;
|
|
|
|
struct pid *pid;
|
2005-04-16 16:20:36 -06:00
|
|
|
struct sembuf *sop;
|
2014-01-27 18:07:04 -07:00
|
|
|
struct sem *curr;
|
2014-06-06 15:37:49 -06:00
|
|
|
struct sembuf *sops;
|
|
|
|
struct sem_undo *un;
|
|
|
|
|
|
|
|
sops = q->sops;
|
|
|
|
nsops = q->nsops;
|
|
|
|
un = q->undo;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
for (sop = sops; sop < sops + nsops; sop++) {
|
2018-06-14 16:27:51 -06:00
|
|
|
int idx = array_index_nospec(sop->sem_num, sma->sem_nsems);
|
|
|
|
curr = &sma->sems[idx];
|
2005-04-16 16:20:36 -06:00
|
|
|
sem_op = sop->sem_op;
|
|
|
|
result = curr->semval;
|
ipc/sem.c: avoid overflow of semop undo (semadj) value
When trying to understand semop code, I found a small mistake in the check
for semadj (undo) value overflow. The new undo value is not stored
immediately and next potential checks are done against the old value.
The failing scenario is not much practical. One semop call has to do more
operations on the same semaphore. Also semval and semadj must have
different values, so there has to be some operations without SEM_UNDO
flag. For example:
struct sembuf depositor_op[1];
struct sembuf collector_op[2];
depositor_op[0].sem_num = 0;
depositor_op[0].sem_op = 20000;
depositor_op[0].sem_flg = 0;
collector_op[0].sem_num = 0;
collector_op[0].sem_op = -10000;
collector_op[0].sem_flg = SEM_UNDO;
collector_op[1].sem_num = 0;
collector_op[1].sem_op = -10000;
collector_op[1].sem_flg = SEM_UNDO;
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 1st deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 1st collect"); return 1; }
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 2nd deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 2nd collect"); return 1; }
return 0;
It passes without error now but the semadj value has overflown in the 2nd
collector operation.
[akpm@linux-foundation.org: restore lessened scope of local `undo']
[davidlohr@hp.com: correct header comment for perform_atomic_semop]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-27 18:07:00 -07:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
if (!sem_op && result)
|
|
|
|
goto would_block;
|
|
|
|
|
|
|
|
result += sem_op;
|
|
|
|
if (result < 0)
|
|
|
|
goto would_block;
|
|
|
|
if (result > SEMVMX)
|
|
|
|
goto out_of_range;
|
ipc/sem.c: avoid overflow of semop undo (semadj) value
When trying to understand semop code, I found a small mistake in the check
for semadj (undo) value overflow. The new undo value is not stored
immediately and next potential checks are done against the old value.
The failing scenario is not much practical. One semop call has to do more
operations on the same semaphore. Also semval and semadj must have
different values, so there has to be some operations without SEM_UNDO
flag. For example:
struct sembuf depositor_op[1];
struct sembuf collector_op[2];
depositor_op[0].sem_num = 0;
depositor_op[0].sem_op = 20000;
depositor_op[0].sem_flg = 0;
collector_op[0].sem_num = 0;
collector_op[0].sem_op = -10000;
collector_op[0].sem_flg = SEM_UNDO;
collector_op[1].sem_num = 0;
collector_op[1].sem_op = -10000;
collector_op[1].sem_flg = SEM_UNDO;
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 1st deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 1st collect"); return 1; }
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 2nd deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 2nd collect"); return 1; }
return 0;
It passes without error now but the semadj value has overflown in the 2nd
collector operation.
[akpm@linux-foundation.org: restore lessened scope of local `undo']
[davidlohr@hp.com: correct header comment for perform_atomic_semop]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-27 18:07:00 -07:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
if (sop->sem_flg & SEM_UNDO) {
|
|
|
|
int undo = un->semadj[sop->sem_num] - sem_op;
|
ipc/sem.c: avoid overflow of semop undo (semadj) value
When trying to understand semop code, I found a small mistake in the check
for semadj (undo) value overflow. The new undo value is not stored
immediately and next potential checks are done against the old value.
The failing scenario is not much practical. One semop call has to do more
operations on the same semaphore. Also semval and semadj must have
different values, so there has to be some operations without SEM_UNDO
flag. For example:
struct sembuf depositor_op[1];
struct sembuf collector_op[2];
depositor_op[0].sem_num = 0;
depositor_op[0].sem_op = 20000;
depositor_op[0].sem_flg = 0;
collector_op[0].sem_num = 0;
collector_op[0].sem_op = -10000;
collector_op[0].sem_flg = SEM_UNDO;
collector_op[1].sem_num = 0;
collector_op[1].sem_op = -10000;
collector_op[1].sem_flg = SEM_UNDO;
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 1st deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 1st collect"); return 1; }
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 2nd deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 2nd collect"); return 1; }
return 0;
It passes without error now but the semadj value has overflown in the 2nd
collector operation.
[akpm@linux-foundation.org: restore lessened scope of local `undo']
[davidlohr@hp.com: correct header comment for perform_atomic_semop]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-27 18:07:00 -07:00
|
|
|
/* Exceeding the undo range is an error. */
|
2005-04-16 16:20:36 -06:00
|
|
|
if (undo < (-SEMAEM - 1) || undo > SEMAEM)
|
|
|
|
goto out_of_range;
|
ipc/sem.c: avoid overflow of semop undo (semadj) value
When trying to understand semop code, I found a small mistake in the check
for semadj (undo) value overflow. The new undo value is not stored
immediately and next potential checks are done against the old value.
The failing scenario is not much practical. One semop call has to do more
operations on the same semaphore. Also semval and semadj must have
different values, so there has to be some operations without SEM_UNDO
flag. For example:
struct sembuf depositor_op[1];
struct sembuf collector_op[2];
depositor_op[0].sem_num = 0;
depositor_op[0].sem_op = 20000;
depositor_op[0].sem_flg = 0;
collector_op[0].sem_num = 0;
collector_op[0].sem_op = -10000;
collector_op[0].sem_flg = SEM_UNDO;
collector_op[1].sem_num = 0;
collector_op[1].sem_op = -10000;
collector_op[1].sem_flg = SEM_UNDO;
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 1st deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 1st collect"); return 1; }
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 2nd deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 2nd collect"); return 1; }
return 0;
It passes without error now but the semadj value has overflown in the 2nd
collector operation.
[akpm@linux-foundation.org: restore lessened scope of local `undo']
[davidlohr@hp.com: correct header comment for perform_atomic_semop]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-27 18:07:00 -07:00
|
|
|
un->semadj[sop->sem_num] = undo;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
ipc/sem.c: avoid overflow of semop undo (semadj) value
When trying to understand semop code, I found a small mistake in the check
for semadj (undo) value overflow. The new undo value is not stored
immediately and next potential checks are done against the old value.
The failing scenario is not much practical. One semop call has to do more
operations on the same semaphore. Also semval and semadj must have
different values, so there has to be some operations without SEM_UNDO
flag. For example:
struct sembuf depositor_op[1];
struct sembuf collector_op[2];
depositor_op[0].sem_num = 0;
depositor_op[0].sem_op = 20000;
depositor_op[0].sem_flg = 0;
collector_op[0].sem_num = 0;
collector_op[0].sem_op = -10000;
collector_op[0].sem_flg = SEM_UNDO;
collector_op[1].sem_num = 0;
collector_op[1].sem_op = -10000;
collector_op[1].sem_flg = SEM_UNDO;
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 1st deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 1st collect"); return 1; }
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 2nd deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 2nd collect"); return 1; }
return 0;
It passes without error now but the semadj value has overflown in the 2nd
collector operation.
[akpm@linux-foundation.org: restore lessened scope of local `undo']
[davidlohr@hp.com: correct header comment for perform_atomic_semop]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-27 18:07:00 -07:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
curr->semval = result;
|
|
|
|
}
|
|
|
|
|
|
|
|
sop--;
|
2014-06-06 15:37:49 -06:00
|
|
|
pid = q->pid;
|
2005-04-16 16:20:36 -06:00
|
|
|
while (sop >= sops) {
|
2018-03-23 00:11:29 -06:00
|
|
|
ipc_update_pid(&sma->sems[sop->sem_num].sempid, pid);
|
2005-04-16 16:20:36 -06:00
|
|
|
sop--;
|
|
|
|
}
|
ipc/sem.c: avoid overflow of semop undo (semadj) value
When trying to understand semop code, I found a small mistake in the check
for semadj (undo) value overflow. The new undo value is not stored
immediately and next potential checks are done against the old value.
The failing scenario is not much practical. One semop call has to do more
operations on the same semaphore. Also semval and semadj must have
different values, so there has to be some operations without SEM_UNDO
flag. For example:
struct sembuf depositor_op[1];
struct sembuf collector_op[2];
depositor_op[0].sem_num = 0;
depositor_op[0].sem_op = 20000;
depositor_op[0].sem_flg = 0;
collector_op[0].sem_num = 0;
collector_op[0].sem_op = -10000;
collector_op[0].sem_flg = SEM_UNDO;
collector_op[1].sem_num = 0;
collector_op[1].sem_op = -10000;
collector_op[1].sem_flg = SEM_UNDO;
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 1st deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 1st collect"); return 1; }
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 2nd deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 2nd collect"); return 1; }
return 0;
It passes without error now but the semadj value has overflown in the 2nd
collector operation.
[akpm@linux-foundation.org: restore lessened scope of local `undo']
[davidlohr@hp.com: correct header comment for perform_atomic_semop]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-27 18:07:00 -07:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
out_of_range:
|
|
|
|
result = -ERANGE;
|
|
|
|
goto undo;
|
|
|
|
|
|
|
|
would_block:
|
2014-06-06 15:37:49 -06:00
|
|
|
q->blocking = sop;
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
if (sop->sem_flg & IPC_NOWAIT)
|
|
|
|
result = -EAGAIN;
|
|
|
|
else
|
|
|
|
result = 1;
|
|
|
|
|
|
|
|
undo:
|
|
|
|
sop--;
|
|
|
|
while (sop >= sops) {
|
ipc/sem.c: avoid overflow of semop undo (semadj) value
When trying to understand semop code, I found a small mistake in the check
for semadj (undo) value overflow. The new undo value is not stored
immediately and next potential checks are done against the old value.
The failing scenario is not much practical. One semop call has to do more
operations on the same semaphore. Also semval and semadj must have
different values, so there has to be some operations without SEM_UNDO
flag. For example:
struct sembuf depositor_op[1];
struct sembuf collector_op[2];
depositor_op[0].sem_num = 0;
depositor_op[0].sem_op = 20000;
depositor_op[0].sem_flg = 0;
collector_op[0].sem_num = 0;
collector_op[0].sem_op = -10000;
collector_op[0].sem_flg = SEM_UNDO;
collector_op[1].sem_num = 0;
collector_op[1].sem_op = -10000;
collector_op[1].sem_flg = SEM_UNDO;
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 1st deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 1st collect"); return 1; }
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 2nd deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 2nd collect"); return 1; }
return 0;
It passes without error now but the semadj value has overflown in the 2nd
collector operation.
[akpm@linux-foundation.org: restore lessened scope of local `undo']
[davidlohr@hp.com: correct header comment for perform_atomic_semop]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-27 18:07:00 -07:00
|
|
|
sem_op = sop->sem_op;
|
2017-07-12 15:34:38 -06:00
|
|
|
sma->sems[sop->sem_num].semval -= sem_op;
|
ipc/sem.c: avoid overflow of semop undo (semadj) value
When trying to understand semop code, I found a small mistake in the check
for semadj (undo) value overflow. The new undo value is not stored
immediately and next potential checks are done against the old value.
The failing scenario is not much practical. One semop call has to do more
operations on the same semaphore. Also semval and semadj must have
different values, so there has to be some operations without SEM_UNDO
flag. For example:
struct sembuf depositor_op[1];
struct sembuf collector_op[2];
depositor_op[0].sem_num = 0;
depositor_op[0].sem_op = 20000;
depositor_op[0].sem_flg = 0;
collector_op[0].sem_num = 0;
collector_op[0].sem_op = -10000;
collector_op[0].sem_flg = SEM_UNDO;
collector_op[1].sem_num = 0;
collector_op[1].sem_op = -10000;
collector_op[1].sem_flg = SEM_UNDO;
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 1st deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 1st collect"); return 1; }
if (semop(semid, depositor_op, 1) == -1)
{ perror("Failed to do 2nd deposit"); return 1; }
if (semop(semid, collector_op, 2) == -1)
{ perror("Failed to do 2nd collect"); return 1; }
return 0;
It passes without error now but the semadj value has overflown in the 2nd
collector operation.
[akpm@linux-foundation.org: restore lessened scope of local `undo']
[davidlohr@hp.com: correct header comment for perform_atomic_semop]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Davidlohr Bueso <davidlohr@hp.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-01-27 18:07:00 -07:00
|
|
|
if (sop->sem_flg & SEM_UNDO)
|
|
|
|
un->semadj[sop->sem_num] += sem_op;
|
2005-04-16 16:20:36 -06:00
|
|
|
sop--;
|
|
|
|
}
|
|
|
|
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
static int perform_atomic_semop(struct sem_array *sma, struct sem_queue *q)
|
|
|
|
{
|
|
|
|
int result, sem_op, nsops;
|
|
|
|
struct sembuf *sop;
|
|
|
|
struct sem *curr;
|
|
|
|
struct sembuf *sops;
|
|
|
|
struct sem_undo *un;
|
|
|
|
|
|
|
|
sops = q->sops;
|
|
|
|
nsops = q->nsops;
|
|
|
|
un = q->undo;
|
|
|
|
|
|
|
|
if (unlikely(q->dupsop))
|
|
|
|
return perform_atomic_semop_slow(sma, q);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We scan the semaphore set twice, first to ensure that the entire
|
|
|
|
* operation can succeed, therefore avoiding any pointless writes
|
|
|
|
* to shared memory and having to undo such changes in order to block
|
|
|
|
* until the operations can go through.
|
|
|
|
*/
|
|
|
|
for (sop = sops; sop < sops + nsops; sop++) {
|
2018-06-14 16:27:51 -06:00
|
|
|
int idx = array_index_nospec(sop->sem_num, sma->sem_nsems);
|
|
|
|
|
|
|
|
curr = &sma->sems[idx];
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
sem_op = sop->sem_op;
|
|
|
|
result = curr->semval;
|
|
|
|
|
|
|
|
if (!sem_op && result)
|
|
|
|
goto would_block; /* wait-for-zero */
|
|
|
|
|
|
|
|
result += sem_op;
|
|
|
|
if (result < 0)
|
|
|
|
goto would_block;
|
|
|
|
|
|
|
|
if (result > SEMVMX)
|
|
|
|
return -ERANGE;
|
|
|
|
|
|
|
|
if (sop->sem_flg & SEM_UNDO) {
|
|
|
|
int undo = un->semadj[sop->sem_num] - sem_op;
|
|
|
|
|
|
|
|
/* Exceeding the undo range is an error. */
|
|
|
|
if (undo < (-SEMAEM - 1) || undo > SEMAEM)
|
|
|
|
return -ERANGE;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
for (sop = sops; sop < sops + nsops; sop++) {
|
2017-07-12 15:34:38 -06:00
|
|
|
curr = &sma->sems[sop->sem_num];
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
sem_op = sop->sem_op;
|
|
|
|
result = curr->semval;
|
|
|
|
|
|
|
|
if (sop->sem_flg & SEM_UNDO) {
|
|
|
|
int undo = un->semadj[sop->sem_num] - sem_op;
|
|
|
|
|
|
|
|
un->semadj[sop->sem_num] = undo;
|
|
|
|
}
|
|
|
|
curr->semval += sem_op;
|
2018-03-23 00:11:29 -06:00
|
|
|
ipc_update_pid(&curr->sempid, q->pid);
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
would_block:
|
|
|
|
q->blocking = sop;
|
|
|
|
return sop->sem_flg & IPC_NOWAIT ? -EAGAIN : 1;
|
|
|
|
}
|
|
|
|
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
static inline void wake_up_sem_queue_prepare(struct sem_queue *q, int error,
|
|
|
|
struct wake_q_head *wake_q)
|
2010-05-26 15:43:41 -06:00
|
|
|
{
|
2020-02-03 18:34:42 -07:00
|
|
|
get_task_struct(q->sleeper);
|
|
|
|
|
|
|
|
/* see SEM_BARRIER_2 for purpuse/pairing */
|
|
|
|
smp_store_release(&q->status, error);
|
|
|
|
|
|
|
|
wake_q_add_safe(wake_q, q->sleeper);
|
2009-12-15 17:47:30 -07:00
|
|
|
}
|
|
|
|
|
2009-12-15 17:47:32 -07:00
|
|
|
static void unlink_queue(struct sem_array *sma, struct sem_queue *q)
|
|
|
|
{
|
|
|
|
list_del(&q->list);
|
2013-04-30 20:15:39 -06:00
|
|
|
if (q->nsops > 1)
|
2009-12-15 17:47:32 -07:00
|
|
|
sma->complex_count--;
|
|
|
|
}
|
|
|
|
|
2010-05-26 15:43:40 -06:00
|
|
|
/** check_restart(sma, q)
|
|
|
|
* @sma: semaphore array
|
|
|
|
* @q: the operation that just completed
|
|
|
|
*
|
|
|
|
* update_queue is O(N^2) when it restarts scanning the whole queue of
|
|
|
|
* waiting operations. Therefore this function checks if the restart is
|
|
|
|
* really necessary. It is called after a previously waiting operation
|
2013-07-08 17:01:23 -06:00
|
|
|
* modified the array.
|
|
|
|
* Note that wait-for-zero operations are handled without restart.
|
2010-05-26 15:43:40 -06:00
|
|
|
*/
|
2016-12-14 16:06:40 -07:00
|
|
|
static inline int check_restart(struct sem_array *sma, struct sem_queue *q)
|
2010-05-26 15:43:40 -06:00
|
|
|
{
|
2013-07-08 17:01:23 -06:00
|
|
|
/* pending complex alter operations are too difficult to analyse */
|
|
|
|
if (!list_empty(&sma->pending_alter))
|
2010-05-26 15:43:40 -06:00
|
|
|
return 1;
|
|
|
|
|
|
|
|
/* we were a sleeping complex operation. Too difficult */
|
|
|
|
if (q->nsops > 1)
|
|
|
|
return 1;
|
|
|
|
|
2013-07-08 17:01:23 -06:00
|
|
|
/* It is impossible that someone waits for the new value:
|
|
|
|
* - complex operations always restart.
|
|
|
|
* - wait-for-zero are handled seperately.
|
|
|
|
* - q is a previously sleeping simple operation that
|
|
|
|
* altered the array. It must be a decrement, because
|
|
|
|
* simple increments never sleep.
|
|
|
|
* - If there are older (higher priority) decrements
|
|
|
|
* in the queue, then they have observed the original
|
|
|
|
* semval value and couldn't proceed. The operation
|
|
|
|
* decremented to value - thus they won't proceed either.
|
|
|
|
*/
|
|
|
|
return 0;
|
|
|
|
}
|
2010-05-26 15:43:40 -06:00
|
|
|
|
2013-07-08 17:01:23 -06:00
|
|
|
/**
|
2014-01-27 18:07:05 -07:00
|
|
|
* wake_const_ops - wake up non-alter tasks
|
2013-07-08 17:01:23 -06:00
|
|
|
* @sma: semaphore array.
|
|
|
|
* @semnum: semaphore that was modified.
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
* @wake_q: lockless wake-queue head.
|
2013-07-08 17:01:23 -06:00
|
|
|
*
|
|
|
|
* wake_const_ops must be called after a semaphore in a semaphore array
|
|
|
|
* was set to 0. If complex const operations are pending, wake_const_ops must
|
|
|
|
* be called with semnum = -1, as well as with the number of each modified
|
|
|
|
* semaphore.
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
* The tasks that must be woken up are added to @wake_q. The return code
|
2013-07-08 17:01:23 -06:00
|
|
|
* is stored in q->pid.
|
|
|
|
* The function returns 1 if at least one operation was completed successfully.
|
|
|
|
*/
|
|
|
|
static int wake_const_ops(struct sem_array *sma, int semnum,
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
struct wake_q_head *wake_q)
|
2013-07-08 17:01:23 -06:00
|
|
|
{
|
2016-12-14 16:06:43 -07:00
|
|
|
struct sem_queue *q, *tmp;
|
2013-07-08 17:01:23 -06:00
|
|
|
struct list_head *pending_list;
|
|
|
|
int semop_completed = 0;
|
|
|
|
|
|
|
|
if (semnum == -1)
|
|
|
|
pending_list = &sma->pending_const;
|
|
|
|
else
|
2017-07-12 15:34:38 -06:00
|
|
|
pending_list = &sma->sems[semnum].pending_const;
|
2010-05-26 15:43:40 -06:00
|
|
|
|
2016-12-14 16:06:43 -07:00
|
|
|
list_for_each_entry_safe(q, tmp, pending_list, list) {
|
|
|
|
int error = perform_atomic_semop(sma, q);
|
2013-07-08 17:01:23 -06:00
|
|
|
|
2016-12-14 16:06:43 -07:00
|
|
|
if (error > 0)
|
|
|
|
continue;
|
|
|
|
/* operation completed, remove from queue & wakeup */
|
|
|
|
unlink_queue(sma, q);
|
2013-07-08 17:01:23 -06:00
|
|
|
|
2016-12-14 16:06:43 -07:00
|
|
|
wake_up_sem_queue_prepare(q, error, wake_q);
|
|
|
|
if (error == 0)
|
|
|
|
semop_completed = 1;
|
2013-07-08 17:01:23 -06:00
|
|
|
}
|
2016-12-14 16:06:43 -07:00
|
|
|
|
2013-07-08 17:01:23 -06:00
|
|
|
return semop_completed;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2014-01-27 18:07:05 -07:00
|
|
|
* do_smart_wakeup_zero - wakeup all wait for zero tasks
|
2013-07-08 17:01:23 -06:00
|
|
|
* @sma: semaphore array
|
|
|
|
* @sops: operations that were performed
|
|
|
|
* @nsops: number of operations
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
* @wake_q: lockless wake-queue head
|
2013-07-08 17:01:23 -06:00
|
|
|
*
|
2014-01-27 18:07:05 -07:00
|
|
|
* Checks all required queue for wait-for-zero operations, based
|
|
|
|
* on the actual changes that were performed on the semaphore array.
|
2013-07-08 17:01:23 -06:00
|
|
|
* The function returns 1 if at least one operation was completed successfully.
|
|
|
|
*/
|
|
|
|
static int do_smart_wakeup_zero(struct sem_array *sma, struct sembuf *sops,
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
int nsops, struct wake_q_head *wake_q)
|
2013-07-08 17:01:23 -06:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
int semop_completed = 0;
|
|
|
|
int got_zero = 0;
|
|
|
|
|
|
|
|
/* first: the per-semaphore queues, if known */
|
|
|
|
if (sops) {
|
|
|
|
for (i = 0; i < nsops; i++) {
|
|
|
|
int num = sops[i].sem_num;
|
|
|
|
|
2017-07-12 15:34:38 -06:00
|
|
|
if (sma->sems[num].semval == 0) {
|
2013-07-08 17:01:23 -06:00
|
|
|
got_zero = 1;
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
semop_completed |= wake_const_ops(sma, num, wake_q);
|
2013-07-08 17:01:23 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* No sops means modified semaphores not known.
|
|
|
|
* Assume all were changed.
|
2010-05-26 15:43:40 -06:00
|
|
|
*/
|
2013-07-08 17:01:23 -06:00
|
|
|
for (i = 0; i < sma->sem_nsems; i++) {
|
2017-07-12 15:34:38 -06:00
|
|
|
if (sma->sems[i].semval == 0) {
|
2013-07-08 17:01:23 -06:00
|
|
|
got_zero = 1;
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
semop_completed |= wake_const_ops(sma, i, wake_q);
|
2013-07-08 17:01:23 -06:00
|
|
|
}
|
|
|
|
}
|
2010-05-26 15:43:40 -06:00
|
|
|
}
|
|
|
|
/*
|
2013-07-08 17:01:23 -06:00
|
|
|
* If one of the modified semaphores got 0,
|
|
|
|
* then check the global queue, too.
|
2010-05-26 15:43:40 -06:00
|
|
|
*/
|
2013-07-08 17:01:23 -06:00
|
|
|
if (got_zero)
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
semop_completed |= wake_const_ops(sma, -1, wake_q);
|
2010-05-26 15:43:40 -06:00
|
|
|
|
2013-07-08 17:01:23 -06:00
|
|
|
return semop_completed;
|
2010-05-26 15:43:40 -06:00
|
|
|
}
|
|
|
|
|
2009-12-15 17:47:33 -07:00
|
|
|
|
|
|
|
/**
|
2014-01-27 18:07:05 -07:00
|
|
|
* update_queue - look for tasks that can be completed.
|
2009-12-15 17:47:33 -07:00
|
|
|
* @sma: semaphore array.
|
|
|
|
* @semnum: semaphore that was modified.
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
* @wake_q: lockless wake-queue head.
|
2009-12-15 17:47:33 -07:00
|
|
|
*
|
|
|
|
* update_queue must be called after a semaphore in a semaphore array
|
2013-04-30 20:15:39 -06:00
|
|
|
* was modified. If multiple semaphores were modified, update_queue must
|
|
|
|
* be called with semnum = -1, as well as with the number of each modified
|
|
|
|
* semaphore.
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
* The tasks that must be woken up are added to @wake_q. The return code
|
2010-05-26 15:43:41 -06:00
|
|
|
* is stored in q->pid.
|
2013-07-08 17:01:23 -06:00
|
|
|
* The function internally checks if const operations can now succeed.
|
|
|
|
*
|
2010-05-26 15:43:41 -06:00
|
|
|
* The function return 1 if at least one semop was completed successfully.
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
static int update_queue(struct sem_array *sma, int semnum, struct wake_q_head *wake_q)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2016-12-14 16:06:43 -07:00
|
|
|
struct sem_queue *q, *tmp;
|
2009-12-15 17:47:33 -07:00
|
|
|
struct list_head *pending_list;
|
2010-05-26 15:43:41 -06:00
|
|
|
int semop_completed = 0;
|
2009-12-15 17:47:33 -07:00
|
|
|
|
2013-04-30 20:15:39 -06:00
|
|
|
if (semnum == -1)
|
2013-07-08 17:01:23 -06:00
|
|
|
pending_list = &sma->pending_alter;
|
2013-04-30 20:15:39 -06:00
|
|
|
else
|
2017-07-12 15:34:38 -06:00
|
|
|
pending_list = &sma->sems[semnum].pending_alter;
|
2009-12-15 17:47:29 -07:00
|
|
|
|
|
|
|
again:
|
2016-12-14 16:06:43 -07:00
|
|
|
list_for_each_entry_safe(q, tmp, pending_list, list) {
|
2010-05-26 15:43:40 -06:00
|
|
|
int error, restart;
|
2009-12-15 17:47:33 -07:00
|
|
|
|
2009-12-15 17:47:34 -07:00
|
|
|
/* If we are scanning the single sop, per-semaphore list of
|
|
|
|
* one semaphore and that semaphore is 0, then it is not
|
2013-07-08 17:01:23 -06:00
|
|
|
* necessary to scan further: simple increments
|
2009-12-15 17:47:34 -07:00
|
|
|
* that affect only one entry succeed immediately and cannot
|
|
|
|
* be in the per semaphore pending queue, and decrements
|
|
|
|
* cannot be successful if the value is already 0.
|
|
|
|
*/
|
2017-07-12 15:34:38 -06:00
|
|
|
if (semnum != -1 && sma->sems[semnum].semval == 0)
|
2009-12-15 17:47:34 -07:00
|
|
|
break;
|
|
|
|
|
2014-06-06 15:37:49 -06:00
|
|
|
error = perform_atomic_semop(sma, q);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
/* Does q->sleeper still need to sleep? */
|
2009-12-15 17:47:29 -07:00
|
|
|
if (error > 0)
|
|
|
|
continue;
|
|
|
|
|
2009-12-15 17:47:32 -07:00
|
|
|
unlink_queue(sma, q);
|
2009-12-15 17:47:29 -07:00
|
|
|
|
2010-05-26 15:43:41 -06:00
|
|
|
if (error) {
|
2010-05-26 15:43:40 -06:00
|
|
|
restart = 0;
|
2010-05-26 15:43:41 -06:00
|
|
|
} else {
|
|
|
|
semop_completed = 1;
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
do_smart_wakeup_zero(sma, q->sops, q->nsops, wake_q);
|
2010-05-26 15:43:40 -06:00
|
|
|
restart = check_restart(sma, q);
|
2010-05-26 15:43:41 -06:00
|
|
|
}
|
2010-05-26 15:43:40 -06:00
|
|
|
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
wake_up_sem_queue_prepare(q, error, wake_q);
|
2010-05-26 15:43:40 -06:00
|
|
|
if (restart)
|
2009-12-15 17:47:29 -07:00
|
|
|
goto again;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
2010-05-26 15:43:41 -06:00
|
|
|
return semop_completed;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2013-09-30 14:45:25 -06:00
|
|
|
/**
|
2014-01-27 18:07:05 -07:00
|
|
|
* set_semotime - set sem_otime
|
2013-09-30 14:45:25 -06:00
|
|
|
* @sma: semaphore array
|
|
|
|
* @sops: operations that modified the array, may be NULL
|
|
|
|
*
|
|
|
|
* sem_otime is replicated to avoid cache line trashing.
|
|
|
|
* This function sets one instance to the current time.
|
|
|
|
*/
|
|
|
|
static void set_semotime(struct sem_array *sma, struct sembuf *sops)
|
|
|
|
{
|
|
|
|
if (sops == NULL) {
|
2018-04-12 07:19:44 -06:00
|
|
|
sma->sems[0].sem_otime = ktime_get_real_seconds();
|
2013-09-30 14:45:25 -06:00
|
|
|
} else {
|
2017-07-12 15:34:38 -06:00
|
|
|
sma->sems[sops[0].sem_num].sem_otime =
|
2018-04-12 07:19:44 -06:00
|
|
|
ktime_get_real_seconds();
|
2013-09-30 14:45:25 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2010-05-26 15:43:41 -06:00
|
|
|
/**
|
2014-01-27 18:07:05 -07:00
|
|
|
* do_smart_update - optimized update_queue
|
2010-05-26 15:43:40 -06:00
|
|
|
* @sma: semaphore array
|
|
|
|
* @sops: operations that were performed
|
|
|
|
* @nsops: number of operations
|
2010-05-26 15:43:41 -06:00
|
|
|
* @otime: force setting otime
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
* @wake_q: lockless wake-queue head
|
2010-05-26 15:43:40 -06:00
|
|
|
*
|
2013-07-08 17:01:23 -06:00
|
|
|
* do_smart_update() does the required calls to update_queue and wakeup_zero,
|
|
|
|
* based on the actual changes that were performed on the semaphore array.
|
2010-05-26 15:43:41 -06:00
|
|
|
* Note that the function does not do the actual wake-up: the caller is
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
* responsible for calling wake_up_q().
|
2010-05-26 15:43:41 -06:00
|
|
|
* It is safe to perform this call after dropping all locks.
|
2010-05-26 15:43:40 -06:00
|
|
|
*/
|
2010-05-26 15:43:41 -06:00
|
|
|
static void do_smart_update(struct sem_array *sma, struct sembuf *sops, int nsops,
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
int otime, struct wake_q_head *wake_q)
|
2010-05-26 15:43:40 -06:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
otime |= do_smart_wakeup_zero(sma, sops, nsops, wake_q);
|
2013-07-08 17:01:23 -06:00
|
|
|
|
2013-07-08 17:01:24 -06:00
|
|
|
if (!list_empty(&sma->pending_alter)) {
|
|
|
|
/* semaphore array uses the global queue - just process it. */
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
otime |= update_queue(sma, -1, wake_q);
|
2013-07-08 17:01:24 -06:00
|
|
|
} else {
|
|
|
|
if (!sops) {
|
|
|
|
/*
|
|
|
|
* No sops, thus the modified semaphores are not
|
|
|
|
* known. Check all.
|
|
|
|
*/
|
|
|
|
for (i = 0; i < sma->sem_nsems; i++)
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
otime |= update_queue(sma, i, wake_q);
|
2013-07-08 17:01:24 -06:00
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* Check the semaphores that were increased:
|
|
|
|
* - No complex ops, thus all sleeping ops are
|
|
|
|
* decrease.
|
|
|
|
* - if we decreased the value, then any sleeping
|
|
|
|
* semaphore ops wont be able to run: If the
|
|
|
|
* previous value was too small, then the new
|
|
|
|
* value will be too small, too.
|
|
|
|
*/
|
|
|
|
for (i = 0; i < nsops; i++) {
|
|
|
|
if (sops[i].sem_op > 0) {
|
|
|
|
otime |= update_queue(sma,
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
sops[i].sem_num, wake_q);
|
2013-07-08 17:01:24 -06:00
|
|
|
}
|
2013-05-26 03:08:52 -06:00
|
|
|
}
|
2013-04-30 20:15:39 -06:00
|
|
|
}
|
2010-05-26 15:43:40 -06:00
|
|
|
}
|
2013-09-30 14:45:25 -06:00
|
|
|
if (otime)
|
|
|
|
set_semotime(sma, sops);
|
2010-05-26 15:43:40 -06:00
|
|
|
}
|
|
|
|
|
2014-06-06 15:37:48 -06:00
|
|
|
/*
|
ipc/sem.c: make semctl(,,{GETNCNT,GETZCNT}) standard compliant
SUSv4 clearly defines how semncnt and semzcnt must be calculated: A task
waits on exactly one semaphore: The semaphore from the first operation
in the sop array that cannot proceed.
The Linux implementation never followed the standard, it tried to count
all semaphores that might be the reason why a task sleeps.
This patch fixes that.
Note:
a) The implementation assumes that GETNCNT and GETZCNT are rare operations,
therefore the code counts them only on demand.
(If they wouldn't be rare, then the non-compliance would have
been found earlier)
b) compared to the initial version of the patch, the BUG_ONs were removed
and it was clarified that the new behavior conforms to SUS.
Back-compatibility concerns:
Manfred:
: - there is no application in Fedora that uses GETNCNT or GETZCNT.
:
: - application that use only single-sop semop() are also safe, the
: difference only affects complex apps.
:
: - portable application are also safe, the new behavior is standard
: compliant.
:
: But that's it. The old behavior existed in Linux from 0.99.something
: until now.
Michael:
: * These operations seem to be very little used. Grepping the public
: source that is contained Fedora 20 source DVD, there appear to be no
: uses. Of course, this says nothing about uses in private /
: non-mainstream FOSS code, but it seems likely that the same pattern
: is followed there.
:
: * The existing behavior is hard enough to understand that I suspect
: that no one understood it well enough to rely on it anyway
: (especially as that behavior contradicted both man page and POSIX).
:
: So, there's a chance of breakage, but I estimate that it's minute.
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-06 15:37:51 -06:00
|
|
|
* check_qop: Test if a queued operation sleeps on the semaphore semnum
|
2014-06-06 15:37:48 -06:00
|
|
|
*/
|
|
|
|
static int check_qop(struct sem_array *sma, int semnum, struct sem_queue *q,
|
|
|
|
bool count_zero)
|
|
|
|
{
|
ipc/sem.c: make semctl(,,{GETNCNT,GETZCNT}) standard compliant
SUSv4 clearly defines how semncnt and semzcnt must be calculated: A task
waits on exactly one semaphore: The semaphore from the first operation
in the sop array that cannot proceed.
The Linux implementation never followed the standard, it tried to count
all semaphores that might be the reason why a task sleeps.
This patch fixes that.
Note:
a) The implementation assumes that GETNCNT and GETZCNT are rare operations,
therefore the code counts them only on demand.
(If they wouldn't be rare, then the non-compliance would have
been found earlier)
b) compared to the initial version of the patch, the BUG_ONs were removed
and it was clarified that the new behavior conforms to SUS.
Back-compatibility concerns:
Manfred:
: - there is no application in Fedora that uses GETNCNT or GETZCNT.
:
: - application that use only single-sop semop() are also safe, the
: difference only affects complex apps.
:
: - portable application are also safe, the new behavior is standard
: compliant.
:
: But that's it. The old behavior existed in Linux from 0.99.something
: until now.
Michael:
: * These operations seem to be very little used. Grepping the public
: source that is contained Fedora 20 source DVD, there appear to be no
: uses. Of course, this says nothing about uses in private /
: non-mainstream FOSS code, but it seems likely that the same pattern
: is followed there.
:
: * The existing behavior is hard enough to understand that I suspect
: that no one understood it well enough to rely on it anyway
: (especially as that behavior contradicted both man page and POSIX).
:
: So, there's a chance of breakage, but I estimate that it's minute.
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-06 15:37:51 -06:00
|
|
|
struct sembuf *sop = q->blocking;
|
2014-06-06 15:37:48 -06:00
|
|
|
|
2014-06-06 15:37:52 -06:00
|
|
|
/*
|
|
|
|
* Linux always (since 0.99.10) reported a task as sleeping on all
|
|
|
|
* semaphores. This violates SUS, therefore it was changed to the
|
|
|
|
* standard compliant behavior.
|
|
|
|
* Give the administrators a chance to notice that an application
|
|
|
|
* might misbehave because it relies on the Linux behavior.
|
|
|
|
*/
|
|
|
|
pr_info_once("semctl(GETNCNT/GETZCNT) is since 3.16 Single Unix Specification compliant.\n"
|
|
|
|
"The task %s (%d) triggered the difference, watch for misbehavior.\n",
|
|
|
|
current->comm, task_pid_nr(current));
|
|
|
|
|
ipc/sem.c: make semctl(,,{GETNCNT,GETZCNT}) standard compliant
SUSv4 clearly defines how semncnt and semzcnt must be calculated: A task
waits on exactly one semaphore: The semaphore from the first operation
in the sop array that cannot proceed.
The Linux implementation never followed the standard, it tried to count
all semaphores that might be the reason why a task sleeps.
This patch fixes that.
Note:
a) The implementation assumes that GETNCNT and GETZCNT are rare operations,
therefore the code counts them only on demand.
(If they wouldn't be rare, then the non-compliance would have
been found earlier)
b) compared to the initial version of the patch, the BUG_ONs were removed
and it was clarified that the new behavior conforms to SUS.
Back-compatibility concerns:
Manfred:
: - there is no application in Fedora that uses GETNCNT or GETZCNT.
:
: - application that use only single-sop semop() are also safe, the
: difference only affects complex apps.
:
: - portable application are also safe, the new behavior is standard
: compliant.
:
: But that's it. The old behavior existed in Linux from 0.99.something
: until now.
Michael:
: * These operations seem to be very little used. Grepping the public
: source that is contained Fedora 20 source DVD, there appear to be no
: uses. Of course, this says nothing about uses in private /
: non-mainstream FOSS code, but it seems likely that the same pattern
: is followed there.
:
: * The existing behavior is hard enough to understand that I suspect
: that no one understood it well enough to rely on it anyway
: (especially as that behavior contradicted both man page and POSIX).
:
: So, there's a chance of breakage, but I estimate that it's minute.
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-06 15:37:51 -06:00
|
|
|
if (sop->sem_num != semnum)
|
|
|
|
return 0;
|
2014-06-06 15:37:48 -06:00
|
|
|
|
ipc/sem.c: make semctl(,,{GETNCNT,GETZCNT}) standard compliant
SUSv4 clearly defines how semncnt and semzcnt must be calculated: A task
waits on exactly one semaphore: The semaphore from the first operation
in the sop array that cannot proceed.
The Linux implementation never followed the standard, it tried to count
all semaphores that might be the reason why a task sleeps.
This patch fixes that.
Note:
a) The implementation assumes that GETNCNT and GETZCNT are rare operations,
therefore the code counts them only on demand.
(If they wouldn't be rare, then the non-compliance would have
been found earlier)
b) compared to the initial version of the patch, the BUG_ONs were removed
and it was clarified that the new behavior conforms to SUS.
Back-compatibility concerns:
Manfred:
: - there is no application in Fedora that uses GETNCNT or GETZCNT.
:
: - application that use only single-sop semop() are also safe, the
: difference only affects complex apps.
:
: - portable application are also safe, the new behavior is standard
: compliant.
:
: But that's it. The old behavior existed in Linux from 0.99.something
: until now.
Michael:
: * These operations seem to be very little used. Grepping the public
: source that is contained Fedora 20 source DVD, there appear to be no
: uses. Of course, this says nothing about uses in private /
: non-mainstream FOSS code, but it seems likely that the same pattern
: is followed there.
:
: * The existing behavior is hard enough to understand that I suspect
: that no one understood it well enough to rely on it anyway
: (especially as that behavior contradicted both man page and POSIX).
:
: So, there's a chance of breakage, but I estimate that it's minute.
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-06 15:37:51 -06:00
|
|
|
if (count_zero && sop->sem_op == 0)
|
|
|
|
return 1;
|
|
|
|
if (!count_zero && sop->sem_op < 0)
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
return 0;
|
2014-06-06 15:37:48 -06:00
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/* The following counts are associated to each semaphore:
|
|
|
|
* semncnt number of tasks waiting on semval being nonzero
|
|
|
|
* semzcnt number of tasks waiting on semval being zero
|
ipc/sem.c: make semctl(,,{GETNCNT,GETZCNT}) standard compliant
SUSv4 clearly defines how semncnt and semzcnt must be calculated: A task
waits on exactly one semaphore: The semaphore from the first operation
in the sop array that cannot proceed.
The Linux implementation never followed the standard, it tried to count
all semaphores that might be the reason why a task sleeps.
This patch fixes that.
Note:
a) The implementation assumes that GETNCNT and GETZCNT are rare operations,
therefore the code counts them only on demand.
(If they wouldn't be rare, then the non-compliance would have
been found earlier)
b) compared to the initial version of the patch, the BUG_ONs were removed
and it was clarified that the new behavior conforms to SUS.
Back-compatibility concerns:
Manfred:
: - there is no application in Fedora that uses GETNCNT or GETZCNT.
:
: - application that use only single-sop semop() are also safe, the
: difference only affects complex apps.
:
: - portable application are also safe, the new behavior is standard
: compliant.
:
: But that's it. The old behavior existed in Linux from 0.99.something
: until now.
Michael:
: * These operations seem to be very little used. Grepping the public
: source that is contained Fedora 20 source DVD, there appear to be no
: uses. Of course, this says nothing about uses in private /
: non-mainstream FOSS code, but it seems likely that the same pattern
: is followed there.
:
: * The existing behavior is hard enough to understand that I suspect
: that no one understood it well enough to rely on it anyway
: (especially as that behavior contradicted both man page and POSIX).
:
: So, there's a chance of breakage, but I estimate that it's minute.
Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-06 15:37:51 -06:00
|
|
|
*
|
|
|
|
* Per definition, a task waits only on the semaphore of the first semop
|
|
|
|
* that cannot proceed, even if additional operation would block, too.
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
2014-06-06 15:37:48 -06:00
|
|
|
static int count_semcnt(struct sem_array *sma, ushort semnum,
|
|
|
|
bool count_zero)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2014-06-06 15:37:48 -06:00
|
|
|
struct list_head *l;
|
2014-01-27 18:07:04 -07:00
|
|
|
struct sem_queue *q;
|
2014-06-06 15:37:48 -06:00
|
|
|
int semcnt;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2014-06-06 15:37:48 -06:00
|
|
|
semcnt = 0;
|
|
|
|
/* First: check the simple operations. They are easy to evaluate */
|
|
|
|
if (count_zero)
|
2017-07-12 15:34:38 -06:00
|
|
|
l = &sma->sems[semnum].pending_const;
|
2014-06-06 15:37:48 -06:00
|
|
|
else
|
2017-07-12 15:34:38 -06:00
|
|
|
l = &sma->sems[semnum].pending_alter;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2014-06-06 15:37:48 -06:00
|
|
|
list_for_each_entry(q, l, list) {
|
|
|
|
/* all task on a per-semaphore list sleep on exactly
|
|
|
|
* that semaphore
|
|
|
|
*/
|
|
|
|
semcnt++;
|
2013-05-09 14:53:28 -06:00
|
|
|
}
|
|
|
|
|
2014-06-06 15:37:48 -06:00
|
|
|
/* Then: check the complex operations. */
|
2014-06-06 15:37:47 -06:00
|
|
|
list_for_each_entry(q, &sma->pending_alter, list) {
|
2014-06-06 15:37:48 -06:00
|
|
|
semcnt += check_qop(sma, semnum, q, count_zero);
|
|
|
|
}
|
|
|
|
if (count_zero) {
|
|
|
|
list_for_each_entry(q, &sma->pending_const, list) {
|
|
|
|
semcnt += check_qop(sma, semnum, q, count_zero);
|
|
|
|
}
|
2014-06-06 15:37:47 -06:00
|
|
|
}
|
2014-06-06 15:37:48 -06:00
|
|
|
return semcnt;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2013-09-11 15:26:24 -06:00
|
|
|
/* Free a semaphore set. freeary() is called with sem_ids.rwsem locked
|
|
|
|
* as a writer and the spinlock for this semaphore set hold. sem_ids.rwsem
|
2007-10-19 00:40:54 -06:00
|
|
|
* remains locked on exit.
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
2008-02-08 05:18:57 -07:00
|
|
|
static void freeary(struct ipc_namespace *ns, struct kern_ipc_perm *ipcp)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2008-07-25 02:48:06 -06:00
|
|
|
struct sem_undo *un, *tu;
|
|
|
|
struct sem_queue *q, *tq;
|
2008-02-08 05:18:57 -07:00
|
|
|
struct sem_array *sma = container_of(ipcp, struct sem_array, sem_perm);
|
2013-04-30 20:15:39 -06:00
|
|
|
int i;
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
DEFINE_WAKE_Q(wake_q);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2008-07-25 02:48:06 -06:00
|
|
|
/* Free the existing undo structures for this semaphore set. */
|
2013-07-08 17:01:11 -06:00
|
|
|
ipc_assert_locked_object(&sma->sem_perm);
|
2008-07-25 02:48:06 -06:00
|
|
|
list_for_each_entry_safe(un, tu, &sma->list_id, list_id) {
|
|
|
|
list_del(&un->list_id);
|
|
|
|
spin_lock(&un->ulp->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
un->semid = -1;
|
2008-07-25 02:48:06 -06:00
|
|
|
list_del_rcu(&un->list_proc);
|
|
|
|
spin_unlock(&un->ulp->lock);
|
2011-03-17 22:09:35 -06:00
|
|
|
kfree_rcu(un, rcu);
|
2008-07-25 02:48:06 -06:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
/* Wake up all pending processes and let them fail with EIDRM. */
|
2013-07-08 17:01:23 -06:00
|
|
|
list_for_each_entry_safe(q, tq, &sma->pending_const, list) {
|
|
|
|
unlink_queue(sma, q);
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
wake_up_sem_queue_prepare(q, -EIDRM, &wake_q);
|
2013-07-08 17:01:23 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
list_for_each_entry_safe(q, tq, &sma->pending_alter, list) {
|
2009-12-15 17:47:32 -07:00
|
|
|
unlink_queue(sma, q);
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
wake_up_sem_queue_prepare(q, -EIDRM, &wake_q);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
2013-04-30 20:15:39 -06:00
|
|
|
for (i = 0; i < sma->sem_nsems; i++) {
|
2017-07-12 15:34:38 -06:00
|
|
|
struct sem *sem = &sma->sems[i];
|
2013-07-08 17:01:23 -06:00
|
|
|
list_for_each_entry_safe(q, tq, &sem->pending_const, list) {
|
|
|
|
unlink_queue(sma, q);
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
wake_up_sem_queue_prepare(q, -EIDRM, &wake_q);
|
2013-07-08 17:01:23 -06:00
|
|
|
}
|
|
|
|
list_for_each_entry_safe(q, tq, &sem->pending_alter, list) {
|
2013-04-30 20:15:39 -06:00
|
|
|
unlink_queue(sma, q);
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
wake_up_sem_queue_prepare(q, -EIDRM, &wake_q);
|
2013-04-30 20:15:39 -06:00
|
|
|
}
|
2018-03-23 00:11:29 -06:00
|
|
|
ipc_update_pid(&sem->sempid, NULL);
|
2013-04-30 20:15:39 -06:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2007-10-19 00:40:48 -06:00
|
|
|
/* Remove the semaphore set from the IDR */
|
|
|
|
sem_rmid(ns, sma);
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
2005-04-16 16:20:36 -06:00
|
|
|
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
wake_up_q(&wake_q);
|
2006-10-02 03:18:22 -06:00
|
|
|
ns->used_sems -= sma->sem_nsems;
|
2017-07-12 15:34:41 -06:00
|
|
|
ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
static unsigned long copy_semid_to_user(void __user *buf, struct semid64_ds *in, int version)
|
|
|
|
{
|
2014-01-27 18:07:04 -07:00
|
|
|
switch (version) {
|
2005-04-16 16:20:36 -06:00
|
|
|
case IPC_64:
|
|
|
|
return copy_to_user(buf, in, sizeof(*in));
|
|
|
|
case IPC_OLD:
|
|
|
|
{
|
|
|
|
struct semid_ds out;
|
|
|
|
|
2010-09-30 16:15:31 -06:00
|
|
|
memset(&out, 0, sizeof(out));
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
ipc64_perm_to_ipc_perm(&in->sem_perm, &out.sem_perm);
|
|
|
|
|
|
|
|
out.sem_otime = in->sem_otime;
|
|
|
|
out.sem_ctime = in->sem_ctime;
|
|
|
|
out.sem_nsems = in->sem_nsems;
|
|
|
|
|
|
|
|
return copy_to_user(buf, &out, sizeof(out));
|
|
|
|
}
|
|
|
|
default:
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2017-08-02 20:51:13 -06:00
|
|
|
static time64_t get_semotime(struct sem_array *sma)
|
2013-07-08 17:01:25 -06:00
|
|
|
{
|
|
|
|
int i;
|
2017-08-02 20:51:13 -06:00
|
|
|
time64_t res;
|
2013-07-08 17:01:25 -06:00
|
|
|
|
2017-07-12 15:34:38 -06:00
|
|
|
res = sma->sems[0].sem_otime;
|
2013-07-08 17:01:25 -06:00
|
|
|
for (i = 1; i < sma->sem_nsems; i++) {
|
2017-08-02 20:51:13 -06:00
|
|
|
time64_t to = sma->sems[i].sem_otime;
|
2013-07-08 17:01:25 -06:00
|
|
|
|
|
|
|
if (to > res)
|
|
|
|
res = to;
|
|
|
|
}
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
2017-07-09 07:11:00 -06:00
|
|
|
static int semctl_stat(struct ipc_namespace *ns, int semid,
|
|
|
|
int cmd, struct semid64_ds *semid64)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct sem_array *sma;
|
2015-04-28 13:39:50 -06:00
|
|
|
time64_t semotime;
|
2017-07-09 07:11:00 -06:00
|
|
|
int err;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2017-07-09 07:11:00 -06:00
|
|
|
memset(semid64, 0, sizeof(*semid64));
|
2014-06-06 15:37:37 -06:00
|
|
|
|
2017-07-09 07:11:00 -06:00
|
|
|
rcu_read_lock();
|
2018-04-10 17:35:26 -06:00
|
|
|
if (cmd == SEM_STAT || cmd == SEM_STAT_ANY) {
|
2017-07-09 07:11:00 -06:00
|
|
|
sma = sem_obtain_object(ns, semid);
|
|
|
|
if (IS_ERR(sma)) {
|
|
|
|
err = PTR_ERR(sma);
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
2018-04-10 17:35:26 -06:00
|
|
|
} else { /* IPC_STAT */
|
2017-07-09 07:11:00 -06:00
|
|
|
sma = sem_obtain_object_check(ns, semid);
|
|
|
|
if (IS_ERR(sma)) {
|
|
|
|
err = PTR_ERR(sma);
|
|
|
|
goto out_unlock;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-04-10 17:35:26 -06:00
|
|
|
/* see comment for SHM_STAT_ANY */
|
|
|
|
if (cmd == SEM_STAT_ANY)
|
|
|
|
audit_ipc_obj(&sma->sem_perm);
|
|
|
|
else {
|
|
|
|
err = -EACCES;
|
|
|
|
if (ipcperms(ns, &sma->sem_perm, S_IRUGO))
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2018-03-22 19:52:43 -06:00
|
|
|
err = security_sem_semctl(&sma->sem_perm, cmd);
|
2017-07-09 07:11:00 -06:00
|
|
|
if (err)
|
|
|
|
goto out_unlock;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2018-02-06 16:40:49 -07:00
|
|
|
ipc_lock_object(&sma->sem_perm);
|
|
|
|
|
|
|
|
if (!ipc_valid_object(&sma->sem_perm)) {
|
|
|
|
ipc_unlock_object(&sma->sem_perm);
|
|
|
|
err = -EIDRM;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
|
|
|
|
2017-07-09 07:11:00 -06:00
|
|
|
kernel_to_ipc64_perm(&sma->sem_perm, &semid64->sem_perm);
|
2015-04-28 13:39:50 -06:00
|
|
|
semotime = get_semotime(sma);
|
|
|
|
semid64->sem_otime = semotime;
|
2017-07-09 07:11:00 -06:00
|
|
|
semid64->sem_ctime = sma->sem_ctime;
|
2015-04-28 13:39:50 -06:00
|
|
|
#ifndef CONFIG_64BIT
|
|
|
|
semid64->sem_otime_high = semotime >> 32;
|
|
|
|
semid64->sem_ctime_high = sma->sem_ctime >> 32;
|
|
|
|
#endif
|
2017-07-09 07:11:00 -06:00
|
|
|
semid64->sem_nsems = sma->sem_nsems;
|
2018-02-06 16:40:49 -07:00
|
|
|
|
2018-08-21 23:01:21 -06:00
|
|
|
if (cmd == IPC_STAT) {
|
|
|
|
/*
|
|
|
|
* As defined in SUS:
|
|
|
|
* Return 0 on success
|
|
|
|
*/
|
|
|
|
err = 0;
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* SEM_STAT and SEM_STAT_ANY (both Linux specific)
|
|
|
|
* Return the full id, including the sequence number
|
|
|
|
*/
|
|
|
|
err = sma->sem_perm.id;
|
|
|
|
}
|
2018-02-06 16:40:49 -07:00
|
|
|
ipc_unlock_object(&sma->sem_perm);
|
2005-04-16 16:20:36 -06:00
|
|
|
out_unlock:
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
rcu_read_unlock();
|
2005-04-16 16:20:36 -06:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2017-07-09 07:11:00 -06:00
|
|
|
static int semctl_info(struct ipc_namespace *ns, int semid,
|
|
|
|
int cmd, void __user *p)
|
|
|
|
{
|
|
|
|
struct seminfo seminfo;
|
2018-08-21 23:02:00 -06:00
|
|
|
int max_idx;
|
2017-07-09 07:11:00 -06:00
|
|
|
int err;
|
|
|
|
|
|
|
|
err = security_sem_semctl(NULL, cmd);
|
|
|
|
if (err)
|
|
|
|
return err;
|
|
|
|
|
|
|
|
memset(&seminfo, 0, sizeof(seminfo));
|
|
|
|
seminfo.semmni = ns->sc_semmni;
|
|
|
|
seminfo.semmns = ns->sc_semmns;
|
|
|
|
seminfo.semmsl = ns->sc_semmsl;
|
|
|
|
seminfo.semopm = ns->sc_semopm;
|
|
|
|
seminfo.semvmx = SEMVMX;
|
|
|
|
seminfo.semmnu = SEMMNU;
|
|
|
|
seminfo.semmap = SEMMAP;
|
|
|
|
seminfo.semume = SEMUME;
|
|
|
|
down_read(&sem_ids(ns).rwsem);
|
|
|
|
if (cmd == SEM_INFO) {
|
|
|
|
seminfo.semusz = sem_ids(ns).in_use;
|
|
|
|
seminfo.semaem = ns->used_sems;
|
|
|
|
} else {
|
|
|
|
seminfo.semusz = SEMUSZ;
|
|
|
|
seminfo.semaem = SEMAEM;
|
|
|
|
}
|
2018-08-21 23:02:00 -06:00
|
|
|
max_idx = ipc_get_maxidx(&sem_ids(ns));
|
2017-07-09 07:11:00 -06:00
|
|
|
up_read(&sem_ids(ns).rwsem);
|
|
|
|
if (copy_to_user(p, &seminfo, sizeof(struct seminfo)))
|
|
|
|
return -EFAULT;
|
2018-08-21 23:02:00 -06:00
|
|
|
return (max_idx < 0) ? 0 : max_idx;
|
2017-07-09 07:11:00 -06:00
|
|
|
}
|
|
|
|
|
2013-03-05 13:04:55 -07:00
|
|
|
static int semctl_setval(struct ipc_namespace *ns, int semid, int semnum,
|
2017-07-09 07:11:00 -06:00
|
|
|
int val)
|
2013-03-05 13:04:55 -07:00
|
|
|
{
|
|
|
|
struct sem_undo *un;
|
|
|
|
struct sem_array *sma;
|
2014-01-27 18:07:04 -07:00
|
|
|
struct sem *curr;
|
2017-07-09 07:11:00 -06:00
|
|
|
int err;
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
DEFINE_WAKE_Q(wake_q);
|
|
|
|
|
2013-04-30 20:15:44 -06:00
|
|
|
if (val > SEMVMX || val < 0)
|
|
|
|
return -ERANGE;
|
2013-03-05 13:04:55 -07:00
|
|
|
|
2013-04-30 20:15:44 -06:00
|
|
|
rcu_read_lock();
|
|
|
|
sma = sem_obtain_object_check(ns, semid);
|
|
|
|
if (IS_ERR(sma)) {
|
|
|
|
rcu_read_unlock();
|
|
|
|
return PTR_ERR(sma);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (semnum < 0 || semnum >= sma->sem_nsems) {
|
|
|
|
rcu_read_unlock();
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
if (ipcperms(ns, &sma->sem_perm, S_IWUGO)) {
|
|
|
|
rcu_read_unlock();
|
|
|
|
return -EACCES;
|
|
|
|
}
|
2013-03-05 13:04:55 -07:00
|
|
|
|
2018-03-22 19:52:43 -06:00
|
|
|
err = security_sem_semctl(&sma->sem_perm, SETVAL);
|
2013-04-30 20:15:44 -06:00
|
|
|
if (err) {
|
|
|
|
rcu_read_unlock();
|
|
|
|
return -EACCES;
|
|
|
|
}
|
2013-03-05 13:04:55 -07:00
|
|
|
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_lock(sma, NULL, -1);
|
2013-03-05 13:04:55 -07:00
|
|
|
|
2014-01-27 18:07:01 -07:00
|
|
|
if (!ipc_valid_object(&sma->sem_perm)) {
|
2013-10-16 14:46:45 -06:00
|
|
|
sem_unlock(sma, -1);
|
|
|
|
rcu_read_unlock();
|
|
|
|
return -EIDRM;
|
|
|
|
}
|
|
|
|
|
2018-06-14 16:27:51 -06:00
|
|
|
semnum = array_index_nospec(semnum, sma->sem_nsems);
|
2017-07-12 15:34:38 -06:00
|
|
|
curr = &sma->sems[semnum];
|
2013-03-05 13:04:55 -07:00
|
|
|
|
2013-07-08 17:01:11 -06:00
|
|
|
ipc_assert_locked_object(&sma->sem_perm);
|
2013-03-05 13:04:55 -07:00
|
|
|
list_for_each_entry(un, &sma->list_id, list_id)
|
|
|
|
un->semadj[semnum] = 0;
|
|
|
|
|
|
|
|
curr->semval = val;
|
2018-03-23 00:11:29 -06:00
|
|
|
ipc_update_pid(&curr->sempid, task_tgid(current));
|
2017-08-02 20:51:13 -06:00
|
|
|
sma->sem_ctime = ktime_get_real_seconds();
|
2013-03-05 13:04:55 -07:00
|
|
|
/* maybe some queued-up processes were waiting for this */
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
do_smart_update(sma, NULL, 0, 0, &wake_q);
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
wake_up_q(&wake_q);
|
2013-04-30 20:15:44 -06:00
|
|
|
return 0;
|
2013-03-05 13:04:55 -07:00
|
|
|
}
|
|
|
|
|
2006-10-02 03:18:22 -06:00
|
|
|
static int semctl_main(struct ipc_namespace *ns, int semid, int semnum,
|
2013-03-05 13:04:55 -07:00
|
|
|
int cmd, void __user *p)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct sem_array *sma;
|
2014-01-27 18:07:04 -07:00
|
|
|
struct sem *curr;
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
int err, nsems;
|
2005-04-16 16:20:36 -06:00
|
|
|
ushort fast_sem_io[SEMMSL_FAST];
|
2014-01-27 18:07:04 -07:00
|
|
|
ushort *sem_io = fast_sem_io;
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
DEFINE_WAKE_Q(wake_q);
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
|
|
|
|
rcu_read_lock();
|
|
|
|
sma = sem_obtain_object_check(ns, semid);
|
|
|
|
if (IS_ERR(sma)) {
|
|
|
|
rcu_read_unlock();
|
2007-10-19 00:40:51 -06:00
|
|
|
return PTR_ERR(sma);
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
nsems = sma->sem_nsems;
|
|
|
|
|
|
|
|
err = -EACCES;
|
2013-05-04 12:04:29 -06:00
|
|
|
if (ipcperms(ns, &sma->sem_perm, cmd == SETALL ? S_IWUGO : S_IRUGO))
|
|
|
|
goto out_rcu_wakeup;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2018-03-22 19:52:43 -06:00
|
|
|
err = security_sem_semctl(&sma->sem_perm, cmd);
|
2013-05-04 12:04:29 -06:00
|
|
|
if (err)
|
|
|
|
goto out_rcu_wakeup;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
err = -EACCES;
|
|
|
|
switch (cmd) {
|
|
|
|
case GETALL:
|
|
|
|
{
|
2013-03-05 13:04:55 -07:00
|
|
|
ushort __user *array = p;
|
2005-04-16 16:20:36 -06:00
|
|
|
int i;
|
|
|
|
|
2013-05-02 17:30:49 -06:00
|
|
|
sem_lock(sma, NULL, -1);
|
2014-01-27 18:07:01 -07:00
|
|
|
if (!ipc_valid_object(&sma->sem_perm)) {
|
2013-10-16 14:46:45 -06:00
|
|
|
err = -EIDRM;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
2014-01-27 18:07:04 -07:00
|
|
|
if (nsems > SEMMSL_FAST) {
|
2017-07-12 15:34:41 -06:00
|
|
|
if (!ipc_rcu_getref(&sma->sem_perm)) {
|
2013-05-02 17:30:49 -06:00
|
|
|
err = -EIDRM;
|
2013-10-16 14:46:45 -06:00
|
|
|
goto out_unlock;
|
2013-05-02 17:30:49 -06:00
|
|
|
}
|
|
|
|
sem_unlock(sma, -1);
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
2017-07-12 15:34:47 -06:00
|
|
|
sem_io = kvmalloc_array(nsems, sizeof(ushort),
|
|
|
|
GFP_KERNEL);
|
2014-01-27 18:07:04 -07:00
|
|
|
if (sem_io == NULL) {
|
2017-07-12 15:34:41 -06:00
|
|
|
ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
|
2005-04-16 16:20:36 -06:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
2013-05-04 11:13:40 -06:00
|
|
|
rcu_read_lock();
|
2008-04-29 02:00:46 -06:00
|
|
|
sem_lock_and_putref(sma);
|
2014-01-27 18:07:01 -07:00
|
|
|
if (!ipc_valid_object(&sma->sem_perm)) {
|
2005-04-16 16:20:36 -06:00
|
|
|
err = -EIDRM;
|
2013-10-16 14:46:45 -06:00
|
|
|
goto out_unlock;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
2013-05-02 17:30:49 -06:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
for (i = 0; i < sma->sem_nsems; i++)
|
2017-07-12 15:34:38 -06:00
|
|
|
sem_io[i] = sma->sems[i].semval;
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
2005-04-16 16:20:36 -06:00
|
|
|
err = 0;
|
2014-01-27 18:07:04 -07:00
|
|
|
if (copy_to_user(array, sem_io, nsems*sizeof(ushort)))
|
2005-04-16 16:20:36 -06:00
|
|
|
err = -EFAULT;
|
|
|
|
goto out_free;
|
|
|
|
}
|
|
|
|
case SETALL:
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
struct sem_undo *un;
|
|
|
|
|
2017-07-12 15:34:41 -06:00
|
|
|
if (!ipc_rcu_getref(&sma->sem_perm)) {
|
2013-10-16 14:46:45 -06:00
|
|
|
err = -EIDRM;
|
|
|
|
goto out_rcu_wakeup;
|
2013-04-30 20:15:44 -06:00
|
|
|
}
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
rcu_read_unlock();
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2014-01-27 18:07:04 -07:00
|
|
|
if (nsems > SEMMSL_FAST) {
|
2017-07-12 15:34:47 -06:00
|
|
|
sem_io = kvmalloc_array(nsems, sizeof(ushort),
|
|
|
|
GFP_KERNEL);
|
2014-01-27 18:07:04 -07:00
|
|
|
if (sem_io == NULL) {
|
2017-07-12 15:34:41 -06:00
|
|
|
ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
|
2005-04-16 16:20:36 -06:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-01-27 18:07:04 -07:00
|
|
|
if (copy_from_user(sem_io, p, nsems*sizeof(ushort))) {
|
2017-07-12 15:34:41 -06:00
|
|
|
ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
|
2005-04-16 16:20:36 -06:00
|
|
|
err = -EFAULT;
|
|
|
|
goto out_free;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < nsems; i++) {
|
|
|
|
if (sem_io[i] > SEMVMX) {
|
2017-07-12 15:34:41 -06:00
|
|
|
ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
|
2005-04-16 16:20:36 -06:00
|
|
|
err = -ERANGE;
|
|
|
|
goto out_free;
|
|
|
|
}
|
|
|
|
}
|
2013-05-04 11:13:40 -06:00
|
|
|
rcu_read_lock();
|
2008-04-29 02:00:46 -06:00
|
|
|
sem_lock_and_putref(sma);
|
2014-01-27 18:07:01 -07:00
|
|
|
if (!ipc_valid_object(&sma->sem_perm)) {
|
2005-04-16 16:20:36 -06:00
|
|
|
err = -EIDRM;
|
2013-10-16 14:46:45 -06:00
|
|
|
goto out_unlock;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2016-03-22 15:27:48 -06:00
|
|
|
for (i = 0; i < nsems; i++) {
|
2017-07-12 15:34:38 -06:00
|
|
|
sma->sems[i].semval = sem_io[i];
|
2018-03-23 00:11:29 -06:00
|
|
|
ipc_update_pid(&sma->sems[i].sempid, task_tgid(current));
|
2016-03-22 15:27:48 -06:00
|
|
|
}
|
2008-07-25 02:48:04 -06:00
|
|
|
|
2013-07-08 17:01:11 -06:00
|
|
|
ipc_assert_locked_object(&sma->sem_perm);
|
2008-07-25 02:48:04 -06:00
|
|
|
list_for_each_entry(un, &sma->list_id, list_id) {
|
2005-04-16 16:20:36 -06:00
|
|
|
for (i = 0; i < nsems; i++)
|
|
|
|
un->semadj[i] = 0;
|
2008-07-25 02:48:04 -06:00
|
|
|
}
|
2017-08-02 20:51:13 -06:00
|
|
|
sma->sem_ctime = ktime_get_real_seconds();
|
2005-04-16 16:20:36 -06:00
|
|
|
/* maybe some queued-up processes were waiting for this */
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
do_smart_update(sma, NULL, 0, 0, &wake_q);
|
2005-04-16 16:20:36 -06:00
|
|
|
err = 0;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
2013-03-05 13:04:55 -07:00
|
|
|
/* GETVAL, GETPID, GETNCTN, GETZCNT: fall-through */
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
err = -EINVAL;
|
2013-05-04 12:04:29 -06:00
|
|
|
if (semnum < 0 || semnum >= nsems)
|
|
|
|
goto out_rcu_wakeup;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_lock(sma, NULL, -1);
|
2014-01-27 18:07:01 -07:00
|
|
|
if (!ipc_valid_object(&sma->sem_perm)) {
|
2013-10-16 14:46:45 -06:00
|
|
|
err = -EIDRM;
|
|
|
|
goto out_unlock;
|
|
|
|
}
|
2018-06-14 16:27:51 -06:00
|
|
|
|
|
|
|
semnum = array_index_nospec(semnum, nsems);
|
2017-07-12 15:34:38 -06:00
|
|
|
curr = &sma->sems[semnum];
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
switch (cmd) {
|
|
|
|
case GETVAL:
|
|
|
|
err = curr->semval;
|
|
|
|
goto out_unlock;
|
|
|
|
case GETPID:
|
2018-03-23 00:11:29 -06:00
|
|
|
err = pid_vnr(curr->sempid);
|
2005-04-16 16:20:36 -06:00
|
|
|
goto out_unlock;
|
|
|
|
case GETNCNT:
|
2014-06-06 15:37:48 -06:00
|
|
|
err = count_semcnt(sma, semnum, 0);
|
2005-04-16 16:20:36 -06:00
|
|
|
goto out_unlock;
|
|
|
|
case GETZCNT:
|
2014-06-06 15:37:48 -06:00
|
|
|
err = count_semcnt(sma, semnum, 1);
|
2005-04-16 16:20:36 -06:00
|
|
|
goto out_unlock;
|
|
|
|
}
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
out_unlock:
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
2013-05-04 12:04:29 -06:00
|
|
|
out_rcu_wakeup:
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
wake_up_q(&wake_q);
|
2005-04-16 16:20:36 -06:00
|
|
|
out_free:
|
2014-01-27 18:07:04 -07:00
|
|
|
if (sem_io != fast_sem_io)
|
2017-07-12 15:34:47 -06:00
|
|
|
kvfree(sem_io);
|
2005-04-16 16:20:36 -06:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2008-04-29 02:00:50 -06:00
|
|
|
static inline unsigned long
|
|
|
|
copy_semid_from_user(struct semid64_ds *out, void __user *buf, int version)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2014-01-27 18:07:04 -07:00
|
|
|
switch (version) {
|
2005-04-16 16:20:36 -06:00
|
|
|
case IPC_64:
|
2008-04-29 02:00:50 -06:00
|
|
|
if (copy_from_user(out, buf, sizeof(*out)))
|
2005-04-16 16:20:36 -06:00
|
|
|
return -EFAULT;
|
|
|
|
return 0;
|
|
|
|
case IPC_OLD:
|
|
|
|
{
|
|
|
|
struct semid_ds tbuf_old;
|
|
|
|
|
2014-01-27 18:07:04 -07:00
|
|
|
if (copy_from_user(&tbuf_old, buf, sizeof(tbuf_old)))
|
2005-04-16 16:20:36 -06:00
|
|
|
return -EFAULT;
|
|
|
|
|
2008-04-29 02:00:50 -06:00
|
|
|
out->sem_perm.uid = tbuf_old.sem_perm.uid;
|
|
|
|
out->sem_perm.gid = tbuf_old.sem_perm.gid;
|
|
|
|
out->sem_perm.mode = tbuf_old.sem_perm.mode;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
default:
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2008-04-29 02:00:49 -06:00
|
|
|
/*
|
2013-09-11 15:26:24 -06:00
|
|
|
* This function handles some semctl commands which require the rwsem
|
2008-04-29 02:00:49 -06:00
|
|
|
* to be held in write mode.
|
2013-09-11 15:26:24 -06:00
|
|
|
* NOTE: no locks must be held, the rwsem is taken inside this function.
|
2008-04-29 02:00:49 -06:00
|
|
|
*/
|
2008-04-29 02:00:49 -06:00
|
|
|
static int semctl_down(struct ipc_namespace *ns, int semid,
|
2017-07-09 07:11:00 -06:00
|
|
|
int cmd, struct semid64_ds *semid64)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct sem_array *sma;
|
|
|
|
int err;
|
|
|
|
struct kern_ipc_perm *ipcp;
|
|
|
|
|
2013-09-11 15:26:24 -06:00
|
|
|
down_write(&sem_ids(ns).rwsem);
|
2013-07-08 17:01:12 -06:00
|
|
|
rcu_read_lock();
|
|
|
|
|
2018-08-21 23:01:34 -06:00
|
|
|
ipcp = ipcctl_obtain_check(ns, &sem_ids(ns), semid, cmd,
|
2017-07-09 07:11:00 -06:00
|
|
|
&semid64->sem_perm, 0);
|
2013-07-08 17:01:12 -06:00
|
|
|
if (IS_ERR(ipcp)) {
|
|
|
|
err = PTR_ERR(ipcp);
|
|
|
|
goto out_unlock1;
|
|
|
|
}
|
2006-04-02 15:07:33 -06:00
|
|
|
|
2008-04-29 02:00:54 -06:00
|
|
|
sma = container_of(ipcp, struct sem_array, sem_perm);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2018-03-22 19:52:43 -06:00
|
|
|
err = security_sem_semctl(&sma->sem_perm, cmd);
|
2013-07-08 17:01:12 -06:00
|
|
|
if (err)
|
|
|
|
goto out_unlock1;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2013-07-08 17:01:12 -06:00
|
|
|
switch (cmd) {
|
2005-04-16 16:20:36 -06:00
|
|
|
case IPC_RMID:
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_lock(sma, NULL, -1);
|
2013-07-08 17:01:12 -06:00
|
|
|
/* freeary unlocks the ipc object and rcu */
|
2008-02-08 05:18:57 -07:00
|
|
|
freeary(ns, ipcp);
|
2008-04-29 02:00:49 -06:00
|
|
|
goto out_up;
|
2005-04-16 16:20:36 -06:00
|
|
|
case IPC_SET:
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_lock(sma, NULL, -1);
|
2017-07-09 07:11:00 -06:00
|
|
|
err = ipc_update_perm(&semid64->sem_perm, ipcp);
|
2012-02-07 17:54:11 -07:00
|
|
|
if (err)
|
2013-07-08 17:01:12 -06:00
|
|
|
goto out_unlock0;
|
2017-08-02 20:51:13 -06:00
|
|
|
sma->sem_ctime = ktime_get_real_seconds();
|
2005-04-16 16:20:36 -06:00
|
|
|
break;
|
|
|
|
default:
|
|
|
|
err = -EINVAL;
|
2013-07-08 17:01:12 -06:00
|
|
|
goto out_unlock1;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2013-07-08 17:01:12 -06:00
|
|
|
out_unlock0:
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
2013-07-08 17:01:12 -06:00
|
|
|
out_unlock1:
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
2008-04-29 02:00:49 -06:00
|
|
|
out_up:
|
2013-09-11 15:26:24 -06:00
|
|
|
up_write(&sem_ids(ns).rwsem);
|
2005-04-16 16:20:36 -06:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
ipc: rename old-style shmctl/semctl/msgctl syscalls
The behavior of these system calls is slightly different between
architectures, as determined by the CONFIG_ARCH_WANT_IPC_PARSE_VERSION
symbol. Most architectures that implement the split IPC syscalls don't set
that symbol and only get the modern version, but alpha, arm, microblaze,
mips-n32, mips-n64 and xtensa expect the caller to pass the IPC_64 flag.
For the architectures that so far only implement sys_ipc(), i.e. m68k,
mips-o32, powerpc, s390, sh, sparc, and x86-32, we want the new behavior
when adding the split syscalls, so we need to distinguish between the
two groups of architectures.
The method I picked for this distinction is to have a separate system call
entry point: sys_old_*ctl() now uses ipc_parse_version, while sys_*ctl()
does not. The system call tables of the five architectures are changed
accordingly.
As an additional benefit, we no longer need the configuration specific
definition for ipc_parse_version(), it always does the same thing now,
but simply won't get called on architectures with the modern interface.
A small downside is that on architectures that do set
ARCH_WANT_IPC_PARSE_VERSION, we now have an extra set of entry points
that are never called. They only add a few bytes of bloat, so it seems
better to keep them compared to adding yet another Kconfig symbol.
I considered adding new syscall numbers for the IPC_64 variants for
consistency, but decided against that for now.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-12-31 14:22:40 -07:00
|
|
|
static long ksys_semctl(int semid, int semnum, int cmd, unsigned long arg, int version)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-10-02 03:18:22 -06:00
|
|
|
struct ipc_namespace *ns;
|
2013-03-05 13:04:55 -07:00
|
|
|
void __user *p = (void __user *)arg;
|
2017-07-09 07:11:00 -06:00
|
|
|
struct semid64_ds semid64;
|
|
|
|
int err;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (semid < 0)
|
|
|
|
return -EINVAL;
|
|
|
|
|
2006-10-02 03:18:22 -06:00
|
|
|
ns = current->nsproxy->ipc_ns;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2014-01-27 18:07:04 -07:00
|
|
|
switch (cmd) {
|
2005-04-16 16:20:36 -06:00
|
|
|
case IPC_INFO:
|
|
|
|
case SEM_INFO:
|
2017-07-09 07:11:00 -06:00
|
|
|
return semctl_info(ns, semid, cmd, p);
|
2008-02-08 05:18:56 -07:00
|
|
|
case IPC_STAT:
|
2005-04-16 16:20:36 -06:00
|
|
|
case SEM_STAT:
|
2018-04-10 17:35:26 -06:00
|
|
|
case SEM_STAT_ANY:
|
2017-07-09 07:11:00 -06:00
|
|
|
err = semctl_stat(ns, semid, cmd, &semid64);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
if (copy_semid_to_user(p, &semid64, version))
|
|
|
|
err = -EFAULT;
|
|
|
|
return err;
|
2005-04-16 16:20:36 -06:00
|
|
|
case GETALL:
|
|
|
|
case GETVAL:
|
|
|
|
case GETPID:
|
|
|
|
case GETNCNT:
|
|
|
|
case GETZCNT:
|
|
|
|
case SETALL:
|
2013-03-05 13:04:55 -07:00
|
|
|
return semctl_main(ns, semid, semnum, cmd, p);
|
2017-07-09 07:11:00 -06:00
|
|
|
case SETVAL: {
|
|
|
|
int val;
|
|
|
|
#if defined(CONFIG_64BIT) && defined(__BIG_ENDIAN)
|
|
|
|
/* big-endian 64bit */
|
|
|
|
val = arg >> 32;
|
|
|
|
#else
|
|
|
|
/* 32bit or little-endian 64bit */
|
|
|
|
val = arg;
|
|
|
|
#endif
|
|
|
|
return semctl_setval(ns, semid, semnum, val);
|
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
case IPC_SET:
|
2017-07-09 07:11:00 -06:00
|
|
|
if (copy_semid_from_user(&semid64, p, version))
|
|
|
|
return -EFAULT;
|
2020-08-23 16:36:59 -06:00
|
|
|
fallthrough;
|
2017-07-09 07:11:00 -06:00
|
|
|
case IPC_RMID:
|
|
|
|
return semctl_down(ns, semid, cmd, &semid64);
|
2005-04-16 16:20:36 -06:00
|
|
|
default:
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-03-20 13:00:39 -06:00
|
|
|
SYSCALL_DEFINE4(semctl, int, semid, int, semnum, int, cmd, unsigned long, arg)
|
|
|
|
{
|
ipc: rename old-style shmctl/semctl/msgctl syscalls
The behavior of these system calls is slightly different between
architectures, as determined by the CONFIG_ARCH_WANT_IPC_PARSE_VERSION
symbol. Most architectures that implement the split IPC syscalls don't set
that symbol and only get the modern version, but alpha, arm, microblaze,
mips-n32, mips-n64 and xtensa expect the caller to pass the IPC_64 flag.
For the architectures that so far only implement sys_ipc(), i.e. m68k,
mips-o32, powerpc, s390, sh, sparc, and x86-32, we want the new behavior
when adding the split syscalls, so we need to distinguish between the
two groups of architectures.
The method I picked for this distinction is to have a separate system call
entry point: sys_old_*ctl() now uses ipc_parse_version, while sys_*ctl()
does not. The system call tables of the five architectures are changed
accordingly.
As an additional benefit, we no longer need the configuration specific
definition for ipc_parse_version(), it always does the same thing now,
but simply won't get called on architectures with the modern interface.
A small downside is that on architectures that do set
ARCH_WANT_IPC_PARSE_VERSION, we now have an extra set of entry points
that are never called. They only add a few bytes of bloat, so it seems
better to keep them compared to adding yet another Kconfig symbol.
I considered adding new syscall numbers for the IPC_64 variants for
consistency, but decided against that for now.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-12-31 14:22:40 -07:00
|
|
|
return ksys_semctl(semid, semnum, cmd, arg, IPC_64);
|
2018-03-20 13:00:39 -06:00
|
|
|
}
|
|
|
|
|
ipc: rename old-style shmctl/semctl/msgctl syscalls
The behavior of these system calls is slightly different between
architectures, as determined by the CONFIG_ARCH_WANT_IPC_PARSE_VERSION
symbol. Most architectures that implement the split IPC syscalls don't set
that symbol and only get the modern version, but alpha, arm, microblaze,
mips-n32, mips-n64 and xtensa expect the caller to pass the IPC_64 flag.
For the architectures that so far only implement sys_ipc(), i.e. m68k,
mips-o32, powerpc, s390, sh, sparc, and x86-32, we want the new behavior
when adding the split syscalls, so we need to distinguish between the
two groups of architectures.
The method I picked for this distinction is to have a separate system call
entry point: sys_old_*ctl() now uses ipc_parse_version, while sys_*ctl()
does not. The system call tables of the five architectures are changed
accordingly.
As an additional benefit, we no longer need the configuration specific
definition for ipc_parse_version(), it always does the same thing now,
but simply won't get called on architectures with the modern interface.
A small downside is that on architectures that do set
ARCH_WANT_IPC_PARSE_VERSION, we now have an extra set of entry points
that are never called. They only add a few bytes of bloat, so it seems
better to keep them compared to adding yet another Kconfig symbol.
I considered adding new syscall numbers for the IPC_64 variants for
consistency, but decided against that for now.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-12-31 14:22:40 -07:00
|
|
|
#ifdef CONFIG_ARCH_WANT_IPC_PARSE_VERSION
|
|
|
|
long ksys_old_semctl(int semid, int semnum, int cmd, unsigned long arg)
|
|
|
|
{
|
|
|
|
int version = ipc_parse_version(&cmd);
|
|
|
|
|
|
|
|
return ksys_semctl(semid, semnum, cmd, arg, version);
|
|
|
|
}
|
|
|
|
|
|
|
|
SYSCALL_DEFINE4(old_semctl, int, semid, int, semnum, int, cmd, unsigned long, arg)
|
|
|
|
{
|
|
|
|
return ksys_old_semctl(semid, semnum, cmd, arg);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2017-07-09 08:03:23 -06:00
|
|
|
#ifdef CONFIG_COMPAT
|
|
|
|
|
|
|
|
struct compat_semid_ds {
|
|
|
|
struct compat_ipc_perm sem_perm;
|
y2038: globally rename compat_time to old_time32
Christoph Hellwig suggested a slightly different path for handling
backwards compatibility with the 32-bit time_t based system calls:
Rather than simply reusing the compat_sys_* entry points on 32-bit
architectures unchanged, we get rid of those entry points and the
compat_time types by renaming them to something that makes more sense
on 32-bit architectures (which don't have a compat mode otherwise),
and then share the entry points under the new name with the 64-bit
architectures that use them for implementing the compatibility.
The following types and interfaces are renamed here, and moved
from linux/compat_time.h to linux/time32.h:
old new
--- ---
compat_time_t old_time32_t
struct compat_timeval struct old_timeval32
struct compat_timespec struct old_timespec32
struct compat_itimerspec struct old_itimerspec32
ns_to_compat_timeval() ns_to_old_timeval32()
get_compat_itimerspec64() get_old_itimerspec32()
put_compat_itimerspec64() put_old_itimerspec32()
compat_get_timespec64() get_old_timespec32()
compat_put_timespec64() put_old_timespec32()
As we already have aliases in place, this patch addresses only the
instances that are relevant to the system call interface in particular,
not those that occur in device drivers and other modules. Those
will get handled separately, while providing the 64-bit version
of the respective interfaces.
I'm not renaming the timex, rusage and itimerval structures, as we are
still debating what the new interface will look like, and whether we
will need a replacement at all.
This also doesn't change the names of the syscall entry points, which can
be done more easily when we actually switch over the 32-bit architectures
to use them, at that point we need to change COMPAT_SYSCALL_DEFINEx to
SYSCALL_DEFINEx with a new name, e.g. with a _time32 suffix.
Suggested-by: Christoph Hellwig <hch@infradead.org>
Link: https://lore.kernel.org/lkml/20180705222110.GA5698@infradead.org/
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-07-13 04:52:28 -06:00
|
|
|
old_time32_t sem_otime;
|
|
|
|
old_time32_t sem_ctime;
|
2017-07-09 08:03:23 -06:00
|
|
|
compat_uptr_t sem_base;
|
|
|
|
compat_uptr_t sem_pending;
|
|
|
|
compat_uptr_t sem_pending_last;
|
|
|
|
compat_uptr_t undo;
|
|
|
|
unsigned short sem_nsems;
|
|
|
|
};
|
|
|
|
|
|
|
|
static int copy_compat_semid_from_user(struct semid64_ds *out, void __user *buf,
|
|
|
|
int version)
|
|
|
|
{
|
|
|
|
memset(out, 0, sizeof(*out));
|
|
|
|
if (version == IPC_64) {
|
2017-09-25 19:37:28 -06:00
|
|
|
struct compat_semid64_ds __user *p = buf;
|
2017-07-09 08:03:23 -06:00
|
|
|
return get_compat_ipc64_perm(&out->sem_perm, &p->sem_perm);
|
|
|
|
} else {
|
2017-09-25 19:37:28 -06:00
|
|
|
struct compat_semid_ds __user *p = buf;
|
2017-07-09 08:03:23 -06:00
|
|
|
return get_compat_ipc_perm(&out->sem_perm, &p->sem_perm);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int copy_compat_semid_to_user(void __user *buf, struct semid64_ds *in,
|
|
|
|
int version)
|
|
|
|
{
|
|
|
|
if (version == IPC_64) {
|
|
|
|
struct compat_semid64_ds v;
|
|
|
|
memset(&v, 0, sizeof(v));
|
|
|
|
to_compat_ipc64_perm(&v.sem_perm, &in->sem_perm);
|
2015-04-28 13:39:50 -06:00
|
|
|
v.sem_otime = lower_32_bits(in->sem_otime);
|
|
|
|
v.sem_otime_high = upper_32_bits(in->sem_otime);
|
|
|
|
v.sem_ctime = lower_32_bits(in->sem_ctime);
|
|
|
|
v.sem_ctime_high = upper_32_bits(in->sem_ctime);
|
2017-07-09 08:03:23 -06:00
|
|
|
v.sem_nsems = in->sem_nsems;
|
|
|
|
return copy_to_user(buf, &v, sizeof(v));
|
|
|
|
} else {
|
|
|
|
struct compat_semid_ds v;
|
|
|
|
memset(&v, 0, sizeof(v));
|
|
|
|
to_compat_ipc_perm(&v.sem_perm, &in->sem_perm);
|
|
|
|
v.sem_otime = in->sem_otime;
|
|
|
|
v.sem_ctime = in->sem_ctime;
|
|
|
|
v.sem_nsems = in->sem_nsems;
|
|
|
|
return copy_to_user(buf, &v, sizeof(v));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
ipc: rename old-style shmctl/semctl/msgctl syscalls
The behavior of these system calls is slightly different between
architectures, as determined by the CONFIG_ARCH_WANT_IPC_PARSE_VERSION
symbol. Most architectures that implement the split IPC syscalls don't set
that symbol and only get the modern version, but alpha, arm, microblaze,
mips-n32, mips-n64 and xtensa expect the caller to pass the IPC_64 flag.
For the architectures that so far only implement sys_ipc(), i.e. m68k,
mips-o32, powerpc, s390, sh, sparc, and x86-32, we want the new behavior
when adding the split syscalls, so we need to distinguish between the
two groups of architectures.
The method I picked for this distinction is to have a separate system call
entry point: sys_old_*ctl() now uses ipc_parse_version, while sys_*ctl()
does not. The system call tables of the five architectures are changed
accordingly.
As an additional benefit, we no longer need the configuration specific
definition for ipc_parse_version(), it always does the same thing now,
but simply won't get called on architectures with the modern interface.
A small downside is that on architectures that do set
ARCH_WANT_IPC_PARSE_VERSION, we now have an extra set of entry points
that are never called. They only add a few bytes of bloat, so it seems
better to keep them compared to adding yet another Kconfig symbol.
I considered adding new syscall numbers for the IPC_64 variants for
consistency, but decided against that for now.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-12-31 14:22:40 -07:00
|
|
|
static long compat_ksys_semctl(int semid, int semnum, int cmd, int arg, int version)
|
2017-07-09 08:03:23 -06:00
|
|
|
{
|
|
|
|
void __user *p = compat_ptr(arg);
|
|
|
|
struct ipc_namespace *ns;
|
|
|
|
struct semid64_ds semid64;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
ns = current->nsproxy->ipc_ns;
|
|
|
|
|
|
|
|
if (semid < 0)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
switch (cmd & (~IPC_64)) {
|
|
|
|
case IPC_INFO:
|
|
|
|
case SEM_INFO:
|
|
|
|
return semctl_info(ns, semid, cmd, p);
|
|
|
|
case IPC_STAT:
|
|
|
|
case SEM_STAT:
|
2018-04-10 17:35:26 -06:00
|
|
|
case SEM_STAT_ANY:
|
2017-07-09 08:03:23 -06:00
|
|
|
err = semctl_stat(ns, semid, cmd, &semid64);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
if (copy_compat_semid_to_user(p, &semid64, version))
|
|
|
|
err = -EFAULT;
|
|
|
|
return err;
|
|
|
|
case GETVAL:
|
|
|
|
case GETPID:
|
|
|
|
case GETNCNT:
|
|
|
|
case GETZCNT:
|
|
|
|
case GETALL:
|
2005-04-16 16:20:36 -06:00
|
|
|
case SETALL:
|
2013-03-05 13:04:55 -07:00
|
|
|
return semctl_main(ns, semid, semnum, cmd, p);
|
|
|
|
case SETVAL:
|
|
|
|
return semctl_setval(ns, semid, semnum, arg);
|
2005-04-16 16:20:36 -06:00
|
|
|
case IPC_SET:
|
2017-07-09 08:03:23 -06:00
|
|
|
if (copy_compat_semid_from_user(&semid64, p, version))
|
|
|
|
return -EFAULT;
|
2020-08-23 16:36:59 -06:00
|
|
|
fallthrough;
|
2017-07-09 08:03:23 -06:00
|
|
|
case IPC_RMID:
|
|
|
|
return semctl_down(ns, semid, cmd, &semid64);
|
2005-04-16 16:20:36 -06:00
|
|
|
default:
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
}
|
2018-03-20 13:00:39 -06:00
|
|
|
|
|
|
|
COMPAT_SYSCALL_DEFINE4(semctl, int, semid, int, semnum, int, cmd, int, arg)
|
|
|
|
{
|
ipc: rename old-style shmctl/semctl/msgctl syscalls
The behavior of these system calls is slightly different between
architectures, as determined by the CONFIG_ARCH_WANT_IPC_PARSE_VERSION
symbol. Most architectures that implement the split IPC syscalls don't set
that symbol and only get the modern version, but alpha, arm, microblaze,
mips-n32, mips-n64 and xtensa expect the caller to pass the IPC_64 flag.
For the architectures that so far only implement sys_ipc(), i.e. m68k,
mips-o32, powerpc, s390, sh, sparc, and x86-32, we want the new behavior
when adding the split syscalls, so we need to distinguish between the
two groups of architectures.
The method I picked for this distinction is to have a separate system call
entry point: sys_old_*ctl() now uses ipc_parse_version, while sys_*ctl()
does not. The system call tables of the five architectures are changed
accordingly.
As an additional benefit, we no longer need the configuration specific
definition for ipc_parse_version(), it always does the same thing now,
but simply won't get called on architectures with the modern interface.
A small downside is that on architectures that do set
ARCH_WANT_IPC_PARSE_VERSION, we now have an extra set of entry points
that are never called. They only add a few bytes of bloat, so it seems
better to keep them compared to adding yet another Kconfig symbol.
I considered adding new syscall numbers for the IPC_64 variants for
consistency, but decided against that for now.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-12-31 14:22:40 -07:00
|
|
|
return compat_ksys_semctl(semid, semnum, cmd, arg, IPC_64);
|
2018-03-20 13:00:39 -06:00
|
|
|
}
|
ipc: rename old-style shmctl/semctl/msgctl syscalls
The behavior of these system calls is slightly different between
architectures, as determined by the CONFIG_ARCH_WANT_IPC_PARSE_VERSION
symbol. Most architectures that implement the split IPC syscalls don't set
that symbol and only get the modern version, but alpha, arm, microblaze,
mips-n32, mips-n64 and xtensa expect the caller to pass the IPC_64 flag.
For the architectures that so far only implement sys_ipc(), i.e. m68k,
mips-o32, powerpc, s390, sh, sparc, and x86-32, we want the new behavior
when adding the split syscalls, so we need to distinguish between the
two groups of architectures.
The method I picked for this distinction is to have a separate system call
entry point: sys_old_*ctl() now uses ipc_parse_version, while sys_*ctl()
does not. The system call tables of the five architectures are changed
accordingly.
As an additional benefit, we no longer need the configuration specific
definition for ipc_parse_version(), it always does the same thing now,
but simply won't get called on architectures with the modern interface.
A small downside is that on architectures that do set
ARCH_WANT_IPC_PARSE_VERSION, we now have an extra set of entry points
that are never called. They only add a few bytes of bloat, so it seems
better to keep them compared to adding yet another Kconfig symbol.
I considered adding new syscall numbers for the IPC_64 variants for
consistency, but decided against that for now.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-12-31 14:22:40 -07:00
|
|
|
|
|
|
|
#ifdef CONFIG_ARCH_WANT_COMPAT_IPC_PARSE_VERSION
|
|
|
|
long compat_ksys_old_semctl(int semid, int semnum, int cmd, int arg)
|
|
|
|
{
|
|
|
|
int version = compat_ipc_parse_version(&cmd);
|
|
|
|
|
|
|
|
return compat_ksys_semctl(semid, semnum, cmd, arg, version);
|
|
|
|
}
|
|
|
|
|
|
|
|
COMPAT_SYSCALL_DEFINE4(old_semctl, int, semid, int, semnum, int, cmd, int, arg)
|
|
|
|
{
|
|
|
|
return compat_ksys_old_semctl(semid, semnum, cmd, arg);
|
|
|
|
}
|
|
|
|
#endif
|
2017-07-09 08:03:23 -06:00
|
|
|
#endif
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
/* If the task doesn't already have a undo_list, then allocate one
|
|
|
|
* here. We guarantee there is only one thread using this undo list,
|
|
|
|
* and current is THE ONE
|
|
|
|
*
|
|
|
|
* If this allocation and assignment succeeds, but later
|
|
|
|
* portions of this code fail, there is no need to free the sem_undo_list.
|
|
|
|
* Just let it stay associated with the task, and it'll be freed later
|
|
|
|
* at exit time.
|
|
|
|
*
|
|
|
|
* This can block, so callers must hold no locks.
|
|
|
|
*/
|
|
|
|
static inline int get_undo_list(struct sem_undo_list **undo_listp)
|
|
|
|
{
|
|
|
|
struct sem_undo_list *undo_list;
|
|
|
|
|
|
|
|
undo_list = current->sysvsem.undo_list;
|
|
|
|
if (!undo_list) {
|
2006-10-02 03:18:25 -06:00
|
|
|
undo_list = kzalloc(sizeof(*undo_list), GFP_KERNEL);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (undo_list == NULL)
|
|
|
|
return -ENOMEM;
|
2005-08-05 15:05:27 -06:00
|
|
|
spin_lock_init(&undo_list->lock);
|
2017-09-08 17:17:42 -06:00
|
|
|
refcount_set(&undo_list->refcnt, 1);
|
2008-07-25 02:48:04 -06:00
|
|
|
INIT_LIST_HEAD(&undo_list->list_proc);
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
current->sysvsem.undo_list = undo_list;
|
|
|
|
}
|
|
|
|
*undo_listp = undo_list;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2009-12-15 17:47:28 -07:00
|
|
|
static struct sem_undo *__lookup_undo(struct sem_undo_list *ulp, int semid)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2009-12-15 17:47:28 -07:00
|
|
|
struct sem_undo *un;
|
2008-07-25 02:48:04 -06:00
|
|
|
|
2019-09-25 17:48:20 -06:00
|
|
|
list_for_each_entry_rcu(un, &ulp->list_proc, list_proc,
|
|
|
|
spin_is_locked(&ulp->lock)) {
|
2009-12-15 17:47:28 -07:00
|
|
|
if (un->semid == semid)
|
|
|
|
return un;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
2008-07-25 02:48:04 -06:00
|
|
|
return NULL;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2009-12-15 17:47:28 -07:00
|
|
|
static struct sem_undo *lookup_undo(struct sem_undo_list *ulp, int semid)
|
|
|
|
{
|
|
|
|
struct sem_undo *un;
|
|
|
|
|
2014-01-27 18:07:04 -07:00
|
|
|
assert_spin_locked(&ulp->lock);
|
2009-12-15 17:47:28 -07:00
|
|
|
|
|
|
|
un = __lookup_undo(ulp, semid);
|
|
|
|
if (un) {
|
|
|
|
list_del_rcu(&un->list_proc);
|
|
|
|
list_add_rcu(&un->list_proc, &ulp->list_proc);
|
|
|
|
}
|
|
|
|
return un;
|
|
|
|
}
|
|
|
|
|
2008-07-25 02:48:04 -06:00
|
|
|
/**
|
2014-01-27 18:07:05 -07:00
|
|
|
* find_alloc_undo - lookup (and if not present create) undo array
|
2008-07-25 02:48:04 -06:00
|
|
|
* @ns: namespace
|
|
|
|
* @semid: semaphore array id
|
|
|
|
*
|
|
|
|
* The function looks up (and if not present creates) the undo structure.
|
|
|
|
* The size of the undo structure depends on the size of the semaphore
|
|
|
|
* array, thus the alloc path is not that straightforward.
|
2008-07-25 02:48:06 -06:00
|
|
|
* Lifetime-rules: sem_undo is rcu-protected, on success, the function
|
|
|
|
* performs a rcu_read_lock().
|
2008-07-25 02:48:04 -06:00
|
|
|
*/
|
|
|
|
static struct sem_undo *find_alloc_undo(struct ipc_namespace *ns, int semid)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct sem_array *sma;
|
|
|
|
struct sem_undo_list *ulp;
|
|
|
|
struct sem_undo *un, *new;
|
2013-04-30 20:15:44 -06:00
|
|
|
int nsems, error;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
error = get_undo_list(&ulp);
|
|
|
|
if (error)
|
|
|
|
return ERR_PTR(error);
|
|
|
|
|
2008-07-25 02:48:06 -06:00
|
|
|
rcu_read_lock();
|
2007-10-19 00:40:55 -06:00
|
|
|
spin_lock(&ulp->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
un = lookup_undo(ulp, semid);
|
2007-10-19 00:40:55 -06:00
|
|
|
spin_unlock(&ulp->lock);
|
2014-01-27 18:07:04 -07:00
|
|
|
if (likely(un != NULL))
|
2005-04-16 16:20:36 -06:00
|
|
|
goto out;
|
|
|
|
|
|
|
|
/* no undo structure around - allocate one. */
|
2008-07-25 02:48:04 -06:00
|
|
|
/* step 1: figure out the size of the semaphore array */
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
sma = sem_obtain_object_check(ns, semid);
|
|
|
|
if (IS_ERR(sma)) {
|
|
|
|
rcu_read_unlock();
|
2010-05-26 15:43:44 -06:00
|
|
|
return ERR_CAST(sma);
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
}
|
2007-10-19 00:40:51 -06:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
nsems = sma->sem_nsems;
|
2017-07-12 15:34:41 -06:00
|
|
|
if (!ipc_rcu_getref(&sma->sem_perm)) {
|
2013-04-30 20:15:44 -06:00
|
|
|
rcu_read_unlock();
|
|
|
|
un = ERR_PTR(-EIDRM);
|
|
|
|
goto out;
|
|
|
|
}
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
rcu_read_unlock();
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2008-07-25 02:48:04 -06:00
|
|
|
/* step 2: allocate new undo structure */
|
2006-12-06 21:38:51 -07:00
|
|
|
new = kzalloc(sizeof(struct sem_undo) + sizeof(short)*nsems, GFP_KERNEL);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (!new) {
|
2017-07-12 15:34:41 -06:00
|
|
|
ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
|
2005-04-16 16:20:36 -06:00
|
|
|
return ERR_PTR(-ENOMEM);
|
|
|
|
}
|
|
|
|
|
2008-07-25 02:48:06 -06:00
|
|
|
/* step 3: Acquire the lock on semaphore array */
|
2013-05-04 11:13:40 -06:00
|
|
|
rcu_read_lock();
|
2008-04-29 02:00:46 -06:00
|
|
|
sem_lock_and_putref(sma);
|
2014-01-27 18:07:01 -07:00
|
|
|
if (!ipc_valid_object(&sma->sem_perm)) {
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
2005-04-16 16:20:36 -06:00
|
|
|
kfree(new);
|
|
|
|
un = ERR_PTR(-EIDRM);
|
|
|
|
goto out;
|
|
|
|
}
|
2008-07-25 02:48:06 -06:00
|
|
|
spin_lock(&ulp->lock);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* step 4: check for races: did someone else allocate the undo struct?
|
|
|
|
*/
|
|
|
|
un = lookup_undo(ulp, semid);
|
|
|
|
if (un) {
|
|
|
|
kfree(new);
|
|
|
|
goto success;
|
|
|
|
}
|
2008-07-25 02:48:04 -06:00
|
|
|
/* step 5: initialize & link new undo structure */
|
|
|
|
new->semadj = (short *) &new[1];
|
2008-07-25 02:48:06 -06:00
|
|
|
new->ulp = ulp;
|
2008-07-25 02:48:04 -06:00
|
|
|
new->semid = semid;
|
|
|
|
assert_spin_locked(&ulp->lock);
|
2008-07-25 02:48:06 -06:00
|
|
|
list_add_rcu(&new->list_proc, &ulp->list_proc);
|
2013-07-08 17:01:11 -06:00
|
|
|
ipc_assert_locked_object(&sma->sem_perm);
|
2008-07-25 02:48:04 -06:00
|
|
|
list_add(&new->list_id, &sma->list_id);
|
2008-07-25 02:48:06 -06:00
|
|
|
un = new;
|
2008-07-25 02:48:04 -06:00
|
|
|
|
2008-07-25 02:48:06 -06:00
|
|
|
success:
|
2007-10-19 00:40:55 -06:00
|
|
|
spin_unlock(&ulp->lock);
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
2005-04-16 16:20:36 -06:00
|
|
|
out:
|
|
|
|
return un;
|
|
|
|
}
|
|
|
|
|
2017-07-09 08:50:14 -06:00
|
|
|
static long do_semtimedop(int semid, struct sembuf __user *tsops,
|
2017-08-02 20:51:10 -06:00
|
|
|
unsigned nsops, const struct timespec64 *timeout)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
int error = -EINVAL;
|
|
|
|
struct sem_array *sma;
|
|
|
|
struct sembuf fast_sops[SEMOPM_FAST];
|
2014-01-27 18:07:04 -07:00
|
|
|
struct sembuf *sops = fast_sops, *sop;
|
2005-04-16 16:20:36 -06:00
|
|
|
struct sem_undo *un;
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
int max, locknum;
|
|
|
|
bool undos = false, alter = false, dupsop = false;
|
2005-04-16 16:20:36 -06:00
|
|
|
struct sem_queue queue;
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
unsigned long dup = 0, jiffies_left = 0;
|
2006-10-02 03:18:22 -06:00
|
|
|
struct ipc_namespace *ns;
|
|
|
|
|
|
|
|
ns = current->nsproxy->ipc_ns;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (nsops < 1 || semid < 0)
|
|
|
|
return -EINVAL;
|
2006-10-02 03:18:22 -06:00
|
|
|
if (nsops > ns->sc_semopm)
|
2005-04-16 16:20:36 -06:00
|
|
|
return -E2BIG;
|
2014-01-27 18:07:04 -07:00
|
|
|
if (nsops > SEMOPM_FAST) {
|
treewide: kvmalloc() -> kvmalloc_array()
The kvmalloc() function has a 2-factor argument form, kvmalloc_array(). This
patch replaces cases of:
kvmalloc(a * b, gfp)
with:
kvmalloc_array(a * b, gfp)
as well as handling cases of:
kvmalloc(a * b * c, gfp)
with:
kvmalloc(array3_size(a, b, c), gfp)
as it's slightly less ugly than:
kvmalloc_array(array_size(a, b), c, gfp)
This does, however, attempt to ignore constant size factors like:
kvmalloc(4 * 1024, gfp)
though any constants defined via macros get caught up in the conversion.
Any factors with a sizeof() of "unsigned char", "char", and "u8" were
dropped, since they're redundant.
The Coccinelle script used for this was:
// Fix redundant parens around sizeof().
@@
type TYPE;
expression THING, E;
@@
(
kvmalloc(
- (sizeof(TYPE)) * E
+ sizeof(TYPE) * E
, ...)
|
kvmalloc(
- (sizeof(THING)) * E
+ sizeof(THING) * E
, ...)
)
// Drop single-byte sizes and redundant parens.
@@
expression COUNT;
typedef u8;
typedef __u8;
@@
(
kvmalloc(
- sizeof(u8) * (COUNT)
+ COUNT
, ...)
|
kvmalloc(
- sizeof(__u8) * (COUNT)
+ COUNT
, ...)
|
kvmalloc(
- sizeof(char) * (COUNT)
+ COUNT
, ...)
|
kvmalloc(
- sizeof(unsigned char) * (COUNT)
+ COUNT
, ...)
|
kvmalloc(
- sizeof(u8) * COUNT
+ COUNT
, ...)
|
kvmalloc(
- sizeof(__u8) * COUNT
+ COUNT
, ...)
|
kvmalloc(
- sizeof(char) * COUNT
+ COUNT
, ...)
|
kvmalloc(
- sizeof(unsigned char) * COUNT
+ COUNT
, ...)
)
// 2-factor product with sizeof(type/expression) and identifier or constant.
@@
type TYPE;
expression THING;
identifier COUNT_ID;
constant COUNT_CONST;
@@
(
- kvmalloc
+ kvmalloc_array
(
- sizeof(TYPE) * (COUNT_ID)
+ COUNT_ID, sizeof(TYPE)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(TYPE) * COUNT_ID
+ COUNT_ID, sizeof(TYPE)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(TYPE) * (COUNT_CONST)
+ COUNT_CONST, sizeof(TYPE)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(TYPE) * COUNT_CONST
+ COUNT_CONST, sizeof(TYPE)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(THING) * (COUNT_ID)
+ COUNT_ID, sizeof(THING)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(THING) * COUNT_ID
+ COUNT_ID, sizeof(THING)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(THING) * (COUNT_CONST)
+ COUNT_CONST, sizeof(THING)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(THING) * COUNT_CONST
+ COUNT_CONST, sizeof(THING)
, ...)
)
// 2-factor product, only identifiers.
@@
identifier SIZE, COUNT;
@@
- kvmalloc
+ kvmalloc_array
(
- SIZE * COUNT
+ COUNT, SIZE
, ...)
// 3-factor product with 1 sizeof(type) or sizeof(expression), with
// redundant parens removed.
@@
expression THING;
identifier STRIDE, COUNT;
type TYPE;
@@
(
kvmalloc(
- sizeof(TYPE) * (COUNT) * (STRIDE)
+ array3_size(COUNT, STRIDE, sizeof(TYPE))
, ...)
|
kvmalloc(
- sizeof(TYPE) * (COUNT) * STRIDE
+ array3_size(COUNT, STRIDE, sizeof(TYPE))
, ...)
|
kvmalloc(
- sizeof(TYPE) * COUNT * (STRIDE)
+ array3_size(COUNT, STRIDE, sizeof(TYPE))
, ...)
|
kvmalloc(
- sizeof(TYPE) * COUNT * STRIDE
+ array3_size(COUNT, STRIDE, sizeof(TYPE))
, ...)
|
kvmalloc(
- sizeof(THING) * (COUNT) * (STRIDE)
+ array3_size(COUNT, STRIDE, sizeof(THING))
, ...)
|
kvmalloc(
- sizeof(THING) * (COUNT) * STRIDE
+ array3_size(COUNT, STRIDE, sizeof(THING))
, ...)
|
kvmalloc(
- sizeof(THING) * COUNT * (STRIDE)
+ array3_size(COUNT, STRIDE, sizeof(THING))
, ...)
|
kvmalloc(
- sizeof(THING) * COUNT * STRIDE
+ array3_size(COUNT, STRIDE, sizeof(THING))
, ...)
)
// 3-factor product with 2 sizeof(variable), with redundant parens removed.
@@
expression THING1, THING2;
identifier COUNT;
type TYPE1, TYPE2;
@@
(
kvmalloc(
- sizeof(TYPE1) * sizeof(TYPE2) * COUNT
+ array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2))
, ...)
|
kvmalloc(
- sizeof(TYPE1) * sizeof(THING2) * (COUNT)
+ array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2))
, ...)
|
kvmalloc(
- sizeof(THING1) * sizeof(THING2) * COUNT
+ array3_size(COUNT, sizeof(THING1), sizeof(THING2))
, ...)
|
kvmalloc(
- sizeof(THING1) * sizeof(THING2) * (COUNT)
+ array3_size(COUNT, sizeof(THING1), sizeof(THING2))
, ...)
|
kvmalloc(
- sizeof(TYPE1) * sizeof(THING2) * COUNT
+ array3_size(COUNT, sizeof(TYPE1), sizeof(THING2))
, ...)
|
kvmalloc(
- sizeof(TYPE1) * sizeof(THING2) * (COUNT)
+ array3_size(COUNT, sizeof(TYPE1), sizeof(THING2))
, ...)
)
// 3-factor product, only identifiers, with redundant parens removed.
@@
identifier STRIDE, SIZE, COUNT;
@@
(
kvmalloc(
- (COUNT) * STRIDE * SIZE
+ array3_size(COUNT, STRIDE, SIZE)
, ...)
|
kvmalloc(
- COUNT * (STRIDE) * SIZE
+ array3_size(COUNT, STRIDE, SIZE)
, ...)
|
kvmalloc(
- COUNT * STRIDE * (SIZE)
+ array3_size(COUNT, STRIDE, SIZE)
, ...)
|
kvmalloc(
- (COUNT) * (STRIDE) * SIZE
+ array3_size(COUNT, STRIDE, SIZE)
, ...)
|
kvmalloc(
- COUNT * (STRIDE) * (SIZE)
+ array3_size(COUNT, STRIDE, SIZE)
, ...)
|
kvmalloc(
- (COUNT) * STRIDE * (SIZE)
+ array3_size(COUNT, STRIDE, SIZE)
, ...)
|
kvmalloc(
- (COUNT) * (STRIDE) * (SIZE)
+ array3_size(COUNT, STRIDE, SIZE)
, ...)
|
kvmalloc(
- COUNT * STRIDE * SIZE
+ array3_size(COUNT, STRIDE, SIZE)
, ...)
)
// Any remaining multi-factor products, first at least 3-factor products,
// when they're not all constants...
@@
expression E1, E2, E3;
constant C1, C2, C3;
@@
(
kvmalloc(C1 * C2 * C3, ...)
|
kvmalloc(
- (E1) * E2 * E3
+ array3_size(E1, E2, E3)
, ...)
|
kvmalloc(
- (E1) * (E2) * E3
+ array3_size(E1, E2, E3)
, ...)
|
kvmalloc(
- (E1) * (E2) * (E3)
+ array3_size(E1, E2, E3)
, ...)
|
kvmalloc(
- E1 * E2 * E3
+ array3_size(E1, E2, E3)
, ...)
)
// And then all remaining 2 factors products when they're not all constants,
// keeping sizeof() as the second factor argument.
@@
expression THING, E1, E2;
type TYPE;
constant C1, C2, C3;
@@
(
kvmalloc(sizeof(THING) * C2, ...)
|
kvmalloc(sizeof(TYPE) * C2, ...)
|
kvmalloc(C1 * C2 * C3, ...)
|
kvmalloc(C1 * C2, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(TYPE) * (E2)
+ E2, sizeof(TYPE)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(TYPE) * E2
+ E2, sizeof(TYPE)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(THING) * (E2)
+ E2, sizeof(THING)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- sizeof(THING) * E2
+ E2, sizeof(THING)
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- (E1) * E2
+ E1, E2
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- (E1) * (E2)
+ E1, E2
, ...)
|
- kvmalloc
+ kvmalloc_array
(
- E1 * E2
+ E1, E2
, ...)
)
Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-12 15:04:32 -06:00
|
|
|
sops = kvmalloc_array(nsops, sizeof(*sops), GFP_KERNEL);
|
2014-01-27 18:07:04 -07:00
|
|
|
if (sops == NULL)
|
2005-04-16 16:20:36 -06:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
|
2014-01-27 18:07:04 -07:00
|
|
|
if (copy_from_user(sops, tsops, nsops * sizeof(*tsops))) {
|
|
|
|
error = -EFAULT;
|
2005-04-16 16:20:36 -06:00
|
|
|
goto out_free;
|
|
|
|
}
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
if (timeout) {
|
2017-07-09 08:50:14 -06:00
|
|
|
if (timeout->tv_sec < 0 || timeout->tv_nsec < 0 ||
|
|
|
|
timeout->tv_nsec >= 1000000000L) {
|
2005-04-16 16:20:36 -06:00
|
|
|
error = -EINVAL;
|
|
|
|
goto out_free;
|
|
|
|
}
|
2017-08-02 20:51:10 -06:00
|
|
|
jiffies_left = timespec64_to_jiffies(timeout);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
max = 0;
|
|
|
|
for (sop = sops; sop < sops + nsops; sop++) {
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
unsigned long mask = 1ULL << ((sop->sem_num) % BITS_PER_LONG);
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
if (sop->sem_num >= max)
|
|
|
|
max = sop->sem_num;
|
|
|
|
if (sop->sem_flg & SEM_UNDO)
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
undos = true;
|
|
|
|
if (dup & mask) {
|
|
|
|
/*
|
|
|
|
* There was a previous alter access that appears
|
|
|
|
* to have accessed the same semaphore, thus use
|
|
|
|
* the dupsop logic. "appears", because the detection
|
|
|
|
* can only check % BITS_PER_LONG.
|
|
|
|
*/
|
|
|
|
dupsop = true;
|
|
|
|
}
|
|
|
|
if (sop->sem_op != 0) {
|
|
|
|
alter = true;
|
|
|
|
dup |= mask;
|
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
if (undos) {
|
2013-04-30 20:15:44 -06:00
|
|
|
/* On success, find_alloc_undo takes the rcu_read_lock */
|
2008-07-25 02:48:04 -06:00
|
|
|
un = find_alloc_undo(ns, semid);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (IS_ERR(un)) {
|
|
|
|
error = PTR_ERR(un);
|
|
|
|
goto out_free;
|
|
|
|
}
|
2013-04-30 20:15:44 -06:00
|
|
|
} else {
|
2005-04-16 16:20:36 -06:00
|
|
|
un = NULL;
|
2013-04-30 20:15:44 -06:00
|
|
|
rcu_read_lock();
|
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
sma = sem_obtain_object_check(ns, semid);
|
2007-10-19 00:40:51 -06:00
|
|
|
if (IS_ERR(sma)) {
|
2013-04-30 20:15:44 -06:00
|
|
|
rcu_read_unlock();
|
2007-10-19 00:40:51 -06:00
|
|
|
error = PTR_ERR(sma);
|
2005-04-16 16:20:36 -06:00
|
|
|
goto out_free;
|
2007-10-19 00:40:51 -06:00
|
|
|
}
|
|
|
|
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
error = -EFBIG;
|
2016-12-14 16:06:31 -07:00
|
|
|
if (max >= sma->sem_nsems) {
|
|
|
|
rcu_read_unlock();
|
|
|
|
goto out_free;
|
|
|
|
}
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
|
|
|
|
error = -EACCES;
|
2016-12-14 16:06:31 -07:00
|
|
|
if (ipcperms(ns, &sma->sem_perm, alter ? S_IWUGO : S_IRUGO)) {
|
|
|
|
rcu_read_unlock();
|
|
|
|
goto out_free;
|
|
|
|
}
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
|
2018-03-22 19:52:43 -06:00
|
|
|
error = security_sem_semop(&sma->sem_perm, sops, nsops, alter);
|
2016-12-14 16:06:31 -07:00
|
|
|
if (error) {
|
|
|
|
rcu_read_unlock();
|
|
|
|
goto out_free;
|
|
|
|
}
|
ipc,sem: do not hold ipc lock more than necessary
Instead of holding the ipc lock for permissions and security checks, among
others, only acquire it when necessary.
Some numbers....
1) With Rik's semop-multi.c microbenchmark we can see the following
results:
Baseline (3.9-rc1):
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 151452270, ops/sec 5048409
+ 59.40% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 6.14% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 3.84% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 3.64% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 2.06% a.out [kernel.kallsyms] [k] copy_user_enhanced_fast_string
+ 1.86% a.out [kernel.kallsyms] [k] ipc_lock
With this patchset:
cpus 4, threads: 256, semaphores: 128, test duration: 30 secs
total operations: 273156400, ops/sec 9105213
+ 18.54% a.out [kernel.kallsyms] [k] _raw_spin_lock
+ 11.72% a.out [kernel.kallsyms] [k] sys_semtimedop
+ 7.70% a.out [kernel.kallsyms] [k] ipc_has_perm.isra.21
+ 6.58% a.out [kernel.kallsyms] [k] avc_has_perm_flags
+ 6.54% a.out [kernel.kallsyms] [k] __audit_syscall_exit
+ 4.71% a.out [kernel.kallsyms] [k] ipc_obtain_object_check
2) While on an Oracle swingbench DSS (data mining) workload the
improvements are not as exciting as with Rik's benchmark, we can see
some positive numbers. For an 8 socket machine the following are the
percentages of %sys time incurred in the ipc lock:
Baseline (3.9-rc1):
100 swingbench users: 8,74%
400 swingbench users: 21,86%
800 swingbench users: 84,35%
With this patchset:
100 swingbench users: 8,11%
400 swingbench users: 19,93%
800 swingbench users: 77,69%
[riel@redhat.com: fix two locking bugs]
[sasha.levin@oracle.com: prevent releasing RCU read lock twice in semctl_main]
[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Chegu Vinod <chegu_vinod@hp.com>
Acked-by: Michel Lespinasse <walken@google.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Jason Low <jason.low2@hp.com>
Cc: Emmanuel Benisty <benisty.e@gmail.com>
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Stanislav Kinsbursky <skinsbursky@parallels.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-04-30 20:15:29 -06:00
|
|
|
|
2013-10-16 14:46:45 -06:00
|
|
|
error = -EIDRM;
|
|
|
|
locknum = sem_lock(sma, sops, nsops);
|
2014-01-27 18:07:01 -07:00
|
|
|
/*
|
|
|
|
* We eventually might perform the following check in a lockless
|
|
|
|
* fashion, considering ipc_valid_object() locking constraints.
|
|
|
|
* If nsops == 1 and there is no contention for sem_perm.lock, then
|
|
|
|
* only a per-semaphore lock is held and it's OK to proceed with the
|
|
|
|
* check below. More details on the fine grained locking scheme
|
|
|
|
* entangled here and why it's RMID race safe on comments at sem_lock()
|
|
|
|
*/
|
|
|
|
if (!ipc_valid_object(&sma->sem_perm))
|
2013-10-16 14:46:45 -06:00
|
|
|
goto out_unlock_free;
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
2008-07-25 02:48:04 -06:00
|
|
|
* semid identifiers are not unique - find_alloc_undo may have
|
2005-04-16 16:20:36 -06:00
|
|
|
* allocated an undo structure, it was invalidated by an RMID
|
2008-07-25 02:48:04 -06:00
|
|
|
* and now a new array with received the same id. Check and fail.
|
2011-03-30 19:57:33 -06:00
|
|
|
* This case can be detected checking un->semid. The existence of
|
2008-07-25 02:48:06 -06:00
|
|
|
* "un" itself is guaranteed by rcu.
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
2013-04-30 20:15:44 -06:00
|
|
|
if (un && un->semid == -1)
|
|
|
|
goto out_unlock_free;
|
2008-07-25 02:48:04 -06:00
|
|
|
|
2014-06-06 15:37:49 -06:00
|
|
|
queue.sops = sops;
|
|
|
|
queue.nsops = nsops;
|
|
|
|
queue.undo = un;
|
2018-03-23 00:11:29 -06:00
|
|
|
queue.pid = task_tgid(current);
|
2014-06-06 15:37:49 -06:00
|
|
|
queue.alter = alter;
|
ipc/sem: optimize perform_atomic_semop()
This is the main workhorse that deals with semop user calls such that
the waitforzero or semval update operations, on the set, can complete on
not as the sma currently stands. Currently, the set is iterated twice
(setting semval, then backwards for the sempid value). Slowpaths, and
particularly SEM_UNDO calls, must undo any altered sem when it is
detected that the caller must block or has errored-out.
With larger sets, there can occur situations where this involves a lot
of cycles and can obviously be a suboptimal use of cached resources in
shared memory. Ie, discarding CPU caches that are also calling semop
and have the sembuf cached (and can complete), while the current lock
holder doing the semop will block, error, or does a waitforzero
operation.
This patch proposes still iterating the set twice, but the first scan is
read-only, and we perform the actual updates afterward, once we know
that the call will succeed. In order to not suffer from the overhead of
dealing with sops that act on the same sem_num, such (rare) cases use
perform_atomic_semop_slow(), which is exactly what we have now.
Duplicates are detected before grabbing sem_lock, and uses simple a
32/64-bit hash array variable to based on the sem_num we are working on.
In addition add some comments to when we expect to the caller to block.
[akpm@linux-foundation.org: coding-style fixes]
[colin.king@canonical.com: ensure we left shift a ULL rather than a 32 bit integer]
Link: http://lkml.kernel.org/r/20161028181129.7311-1-colin.king@canonical.com
Link: http://lkml.kernel.org/r/20160921194603.GB21438@linux-80c1.suse
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:37 -07:00
|
|
|
queue.dupsop = dupsop;
|
2014-06-06 15:37:49 -06:00
|
|
|
|
|
|
|
error = perform_atomic_semop(sma, &queue);
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
if (error == 0) { /* non-blocking succesfull path */
|
|
|
|
DEFINE_WAKE_Q(wake_q);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the operation was successful, then do
|
2013-09-30 14:45:25 -06:00
|
|
|
* the required updates.
|
|
|
|
*/
|
|
|
|
if (alter)
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
do_smart_update(sma, sops, nsops, 1, &wake_q);
|
2013-09-30 14:45:25 -06:00
|
|
|
else
|
|
|
|
set_semotime(sma, sops);
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
|
|
|
|
sem_unlock(sma, locknum);
|
|
|
|
rcu_read_unlock();
|
|
|
|
wake_up_q(&wake_q);
|
|
|
|
|
|
|
|
goto out_free;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
if (error < 0) /* non-blocking error path */
|
2013-09-30 14:45:25 -06:00
|
|
|
goto out_unlock_free;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
/*
|
|
|
|
* We need to sleep on this operation, so we put the current
|
2005-04-16 16:20:36 -06:00
|
|
|
* task into the pending queue and go to sleep.
|
|
|
|
*/
|
2009-12-15 17:47:32 -07:00
|
|
|
if (nsops == 1) {
|
|
|
|
struct sem *curr;
|
2018-06-14 16:27:51 -06:00
|
|
|
int idx = array_index_nospec(sops->sem_num, sma->sem_nsems);
|
|
|
|
curr = &sma->sems[idx];
|
2009-12-15 17:47:32 -07:00
|
|
|
|
2013-07-08 17:01:24 -06:00
|
|
|
if (alter) {
|
|
|
|
if (sma->complex_count) {
|
|
|
|
list_add_tail(&queue.list,
|
|
|
|
&sma->pending_alter);
|
|
|
|
} else {
|
|
|
|
|
|
|
|
list_add_tail(&queue.list,
|
|
|
|
&curr->pending_alter);
|
|
|
|
}
|
|
|
|
} else {
|
2013-07-08 17:01:23 -06:00
|
|
|
list_add_tail(&queue.list, &curr->pending_const);
|
2013-07-08 17:01:24 -06:00
|
|
|
}
|
2009-12-15 17:47:32 -07:00
|
|
|
} else {
|
2013-07-08 17:01:24 -06:00
|
|
|
if (!sma->complex_count)
|
|
|
|
merge_queues(sma);
|
|
|
|
|
2013-04-30 20:15:39 -06:00
|
|
|
if (alter)
|
2013-07-08 17:01:23 -06:00
|
|
|
list_add_tail(&queue.list, &sma->pending_alter);
|
2013-04-30 20:15:39 -06:00
|
|
|
else
|
2013-07-08 17:01:23 -06:00
|
|
|
list_add_tail(&queue.list, &sma->pending_const);
|
|
|
|
|
2009-12-15 17:47:32 -07:00
|
|
|
sma->complex_count++;
|
|
|
|
}
|
|
|
|
|
2016-12-14 16:06:46 -07:00
|
|
|
do {
|
2020-02-03 18:34:42 -07:00
|
|
|
/* memory ordering ensured by the lock in sem_lock() */
|
2018-07-26 17:37:19 -06:00
|
|
|
WRITE_ONCE(queue.status, -EINTR);
|
2016-12-14 16:06:46 -07:00
|
|
|
queue.sleeper = current;
|
2011-11-02 14:38:52 -06:00
|
|
|
|
2020-02-03 18:34:42 -07:00
|
|
|
/* memory ordering is ensured by the lock in sem_lock() */
|
2016-12-14 16:06:46 -07:00
|
|
|
__set_current_state(TASK_INTERRUPTIBLE);
|
|
|
|
sem_unlock(sma, locknum);
|
|
|
|
rcu_read_unlock();
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2016-12-14 16:06:46 -07:00
|
|
|
if (timeout)
|
|
|
|
jiffies_left = schedule_timeout(jiffies_left);
|
|
|
|
else
|
|
|
|
schedule();
|
2005-04-16 16:20:36 -06:00
|
|
|
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
/*
|
2016-12-14 16:06:46 -07:00
|
|
|
* fastpath: the semop has completed, either successfully or
|
|
|
|
* not, from the syscall pov, is quite irrelevant to us at this
|
|
|
|
* point; we're done.
|
|
|
|
*
|
|
|
|
* We _do_ care, nonetheless, about being awoken by a signal or
|
|
|
|
* spuriously. The queue.status is checked again in the
|
|
|
|
* slowpath (aka after taking sem_lock), such that we can detect
|
|
|
|
* scenarios where we were awakened externally, during the
|
|
|
|
* window between wake_q_add() and wake_up_q().
|
2010-07-20 14:24:23 -06:00
|
|
|
*/
|
2016-12-14 16:06:46 -07:00
|
|
|
error = READ_ONCE(queue.status);
|
|
|
|
if (error != -EINTR) {
|
2020-02-03 18:34:42 -07:00
|
|
|
/* see SEM_BARRIER_2 for purpose/pairing */
|
|
|
|
smp_acquire__after_ctrl_dep();
|
2016-12-14 16:06:46 -07:00
|
|
|
goto out_free;
|
|
|
|
}
|
2011-07-25 18:11:47 -06:00
|
|
|
|
2016-12-14 16:06:46 -07:00
|
|
|
rcu_read_lock();
|
2017-01-10 17:57:48 -07:00
|
|
|
locknum = sem_lock(sma, sops, nsops);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2016-12-14 16:06:49 -07:00
|
|
|
if (!ipc_valid_object(&sma->sem_perm))
|
|
|
|
goto out_unlock_free;
|
|
|
|
|
2020-02-03 18:34:42 -07:00
|
|
|
/*
|
|
|
|
* No necessity for any barrier: We are protect by sem_lock()
|
|
|
|
*/
|
2016-12-14 16:06:49 -07:00
|
|
|
error = READ_ONCE(queue.status);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2016-12-14 16:06:46 -07:00
|
|
|
/*
|
|
|
|
* If queue.status != -EINTR we are woken up by another process.
|
|
|
|
* Leave without unlink_queue(), but with sem_unlock().
|
|
|
|
*/
|
|
|
|
if (error != -EINTR)
|
|
|
|
goto out_unlock_free;
|
2011-11-02 14:38:52 -06:00
|
|
|
|
2016-12-14 16:06:46 -07:00
|
|
|
/*
|
|
|
|
* If an interrupt occurred we have to clean up the queue.
|
|
|
|
*/
|
|
|
|
if (timeout && jiffies_left == 0)
|
|
|
|
error = -EAGAIN;
|
|
|
|
} while (error == -EINTR && !signal_pending(current)); /* spurious */
|
2011-11-02 14:38:52 -06:00
|
|
|
|
2009-12-15 17:47:32 -07:00
|
|
|
unlink_queue(sma, &queue);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
out_unlock_free:
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, locknum);
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
2005-04-16 16:20:36 -06:00
|
|
|
out_free:
|
2014-01-27 18:07:04 -07:00
|
|
|
if (sops != fast_sops)
|
2017-09-08 17:17:52 -06:00
|
|
|
kvfree(sops);
|
2005-04-16 16:20:36 -06:00
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
2018-03-20 12:48:14 -06:00
|
|
|
long ksys_semtimedop(int semid, struct sembuf __user *tsops,
|
2018-04-13 05:58:00 -06:00
|
|
|
unsigned int nsops, const struct __kernel_timespec __user *timeout)
|
2017-07-09 08:50:14 -06:00
|
|
|
{
|
|
|
|
if (timeout) {
|
2017-08-02 20:51:10 -06:00
|
|
|
struct timespec64 ts;
|
|
|
|
if (get_timespec64(&ts, timeout))
|
2017-07-09 08:50:14 -06:00
|
|
|
return -EFAULT;
|
|
|
|
return do_semtimedop(semid, tsops, nsops, &ts);
|
|
|
|
}
|
|
|
|
return do_semtimedop(semid, tsops, nsops, NULL);
|
|
|
|
}
|
|
|
|
|
2018-03-20 12:48:14 -06:00
|
|
|
SYSCALL_DEFINE4(semtimedop, int, semid, struct sembuf __user *, tsops,
|
2018-04-13 05:58:00 -06:00
|
|
|
unsigned int, nsops, const struct __kernel_timespec __user *, timeout)
|
2018-03-20 12:48:14 -06:00
|
|
|
{
|
|
|
|
return ksys_semtimedop(semid, tsops, nsops, timeout);
|
|
|
|
}
|
|
|
|
|
2018-04-13 05:58:23 -06:00
|
|
|
#ifdef CONFIG_COMPAT_32BIT_TIME
|
2018-03-20 12:48:14 -06:00
|
|
|
long compat_ksys_semtimedop(int semid, struct sembuf __user *tsems,
|
|
|
|
unsigned int nsops,
|
y2038: globally rename compat_time to old_time32
Christoph Hellwig suggested a slightly different path for handling
backwards compatibility with the 32-bit time_t based system calls:
Rather than simply reusing the compat_sys_* entry points on 32-bit
architectures unchanged, we get rid of those entry points and the
compat_time types by renaming them to something that makes more sense
on 32-bit architectures (which don't have a compat mode otherwise),
and then share the entry points under the new name with the 64-bit
architectures that use them for implementing the compatibility.
The following types and interfaces are renamed here, and moved
from linux/compat_time.h to linux/time32.h:
old new
--- ---
compat_time_t old_time32_t
struct compat_timeval struct old_timeval32
struct compat_timespec struct old_timespec32
struct compat_itimerspec struct old_itimerspec32
ns_to_compat_timeval() ns_to_old_timeval32()
get_compat_itimerspec64() get_old_itimerspec32()
put_compat_itimerspec64() put_old_itimerspec32()
compat_get_timespec64() get_old_timespec32()
compat_put_timespec64() put_old_timespec32()
As we already have aliases in place, this patch addresses only the
instances that are relevant to the system call interface in particular,
not those that occur in device drivers and other modules. Those
will get handled separately, while providing the 64-bit version
of the respective interfaces.
I'm not renaming the timex, rusage and itimerval structures, as we are
still debating what the new interface will look like, and whether we
will need a replacement at all.
This also doesn't change the names of the syscall entry points, which can
be done more easily when we actually switch over the 32-bit architectures
to use them, at that point we need to change COMPAT_SYSCALL_DEFINEx to
SYSCALL_DEFINEx with a new name, e.g. with a _time32 suffix.
Suggested-by: Christoph Hellwig <hch@infradead.org>
Link: https://lore.kernel.org/lkml/20180705222110.GA5698@infradead.org/
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-07-13 04:52:28 -06:00
|
|
|
const struct old_timespec32 __user *timeout)
|
2017-07-09 08:50:14 -06:00
|
|
|
{
|
|
|
|
if (timeout) {
|
2017-08-02 20:51:10 -06:00
|
|
|
struct timespec64 ts;
|
y2038: globally rename compat_time to old_time32
Christoph Hellwig suggested a slightly different path for handling
backwards compatibility with the 32-bit time_t based system calls:
Rather than simply reusing the compat_sys_* entry points on 32-bit
architectures unchanged, we get rid of those entry points and the
compat_time types by renaming them to something that makes more sense
on 32-bit architectures (which don't have a compat mode otherwise),
and then share the entry points under the new name with the 64-bit
architectures that use them for implementing the compatibility.
The following types and interfaces are renamed here, and moved
from linux/compat_time.h to linux/time32.h:
old new
--- ---
compat_time_t old_time32_t
struct compat_timeval struct old_timeval32
struct compat_timespec struct old_timespec32
struct compat_itimerspec struct old_itimerspec32
ns_to_compat_timeval() ns_to_old_timeval32()
get_compat_itimerspec64() get_old_itimerspec32()
put_compat_itimerspec64() put_old_itimerspec32()
compat_get_timespec64() get_old_timespec32()
compat_put_timespec64() put_old_timespec32()
As we already have aliases in place, this patch addresses only the
instances that are relevant to the system call interface in particular,
not those that occur in device drivers and other modules. Those
will get handled separately, while providing the 64-bit version
of the respective interfaces.
I'm not renaming the timex, rusage and itimerval structures, as we are
still debating what the new interface will look like, and whether we
will need a replacement at all.
This also doesn't change the names of the syscall entry points, which can
be done more easily when we actually switch over the 32-bit architectures
to use them, at that point we need to change COMPAT_SYSCALL_DEFINEx to
SYSCALL_DEFINEx with a new name, e.g. with a _time32 suffix.
Suggested-by: Christoph Hellwig <hch@infradead.org>
Link: https://lore.kernel.org/lkml/20180705222110.GA5698@infradead.org/
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-07-13 04:52:28 -06:00
|
|
|
if (get_old_timespec32(&ts, timeout))
|
2017-07-09 08:50:14 -06:00
|
|
|
return -EFAULT;
|
|
|
|
return do_semtimedop(semid, tsems, nsops, &ts);
|
|
|
|
}
|
|
|
|
return do_semtimedop(semid, tsems, nsops, NULL);
|
|
|
|
}
|
2018-03-20 12:48:14 -06:00
|
|
|
|
2019-01-06 16:33:08 -07:00
|
|
|
SYSCALL_DEFINE4(semtimedop_time32, int, semid, struct sembuf __user *, tsems,
|
2018-03-20 12:48:14 -06:00
|
|
|
unsigned int, nsops,
|
y2038: globally rename compat_time to old_time32
Christoph Hellwig suggested a slightly different path for handling
backwards compatibility with the 32-bit time_t based system calls:
Rather than simply reusing the compat_sys_* entry points on 32-bit
architectures unchanged, we get rid of those entry points and the
compat_time types by renaming them to something that makes more sense
on 32-bit architectures (which don't have a compat mode otherwise),
and then share the entry points under the new name with the 64-bit
architectures that use them for implementing the compatibility.
The following types and interfaces are renamed here, and moved
from linux/compat_time.h to linux/time32.h:
old new
--- ---
compat_time_t old_time32_t
struct compat_timeval struct old_timeval32
struct compat_timespec struct old_timespec32
struct compat_itimerspec struct old_itimerspec32
ns_to_compat_timeval() ns_to_old_timeval32()
get_compat_itimerspec64() get_old_itimerspec32()
put_compat_itimerspec64() put_old_itimerspec32()
compat_get_timespec64() get_old_timespec32()
compat_put_timespec64() put_old_timespec32()
As we already have aliases in place, this patch addresses only the
instances that are relevant to the system call interface in particular,
not those that occur in device drivers and other modules. Those
will get handled separately, while providing the 64-bit version
of the respective interfaces.
I'm not renaming the timex, rusage and itimerval structures, as we are
still debating what the new interface will look like, and whether we
will need a replacement at all.
This also doesn't change the names of the syscall entry points, which can
be done more easily when we actually switch over the 32-bit architectures
to use them, at that point we need to change COMPAT_SYSCALL_DEFINEx to
SYSCALL_DEFINEx with a new name, e.g. with a _time32 suffix.
Suggested-by: Christoph Hellwig <hch@infradead.org>
Link: https://lore.kernel.org/lkml/20180705222110.GA5698@infradead.org/
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2018-07-13 04:52:28 -06:00
|
|
|
const struct old_timespec32 __user *, timeout)
|
2018-03-20 12:48:14 -06:00
|
|
|
{
|
|
|
|
return compat_ksys_semtimedop(semid, tsems, nsops, timeout);
|
|
|
|
}
|
2017-07-09 08:50:14 -06:00
|
|
|
#endif
|
|
|
|
|
2009-01-14 06:14:27 -07:00
|
|
|
SYSCALL_DEFINE3(semop, int, semid, struct sembuf __user *, tsops,
|
|
|
|
unsigned, nsops)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2017-07-09 08:50:14 -06:00
|
|
|
return do_semtimedop(semid, tsops, nsops, NULL);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/* If CLONE_SYSVSEM is set, establish sharing of SEM_UNDO state between
|
|
|
|
* parent and child tasks.
|
|
|
|
*/
|
|
|
|
|
|
|
|
int copy_semundo(unsigned long clone_flags, struct task_struct *tsk)
|
|
|
|
{
|
|
|
|
struct sem_undo_list *undo_list;
|
|
|
|
int error;
|
|
|
|
|
|
|
|
if (clone_flags & CLONE_SYSVSEM) {
|
|
|
|
error = get_undo_list(&undo_list);
|
|
|
|
if (error)
|
|
|
|
return error;
|
2017-09-08 17:17:42 -06:00
|
|
|
refcount_inc(&undo_list->refcnt);
|
2005-04-16 16:20:36 -06:00
|
|
|
tsk->sysvsem.undo_list = undo_list;
|
2014-06-06 15:37:37 -06:00
|
|
|
} else
|
2005-04-16 16:20:36 -06:00
|
|
|
tsk->sysvsem.undo_list = NULL;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* add semadj values to semaphores, free undo structures.
|
|
|
|
* undo structures are not freed when semaphore arrays are destroyed
|
|
|
|
* so some of them may be out of date.
|
|
|
|
* IMPLEMENTATION NOTE: There is some confusion over whether the
|
|
|
|
* set of adjustments that needs to be done should be done in an atomic
|
|
|
|
* manner or not. That is, if we are attempting to decrement the semval
|
|
|
|
* should we queue up and wait until we can do so legally?
|
|
|
|
* The original implementation attempted to do this (queue and wait).
|
|
|
|
* The current implementation does not do so. The POSIX standard
|
|
|
|
* and SVID should be consulted to determine what behavior is mandated.
|
|
|
|
*/
|
|
|
|
void exit_sem(struct task_struct *tsk)
|
|
|
|
{
|
2008-07-25 02:48:04 -06:00
|
|
|
struct sem_undo_list *ulp;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2008-07-25 02:48:04 -06:00
|
|
|
ulp = tsk->sysvsem.undo_list;
|
|
|
|
if (!ulp)
|
2005-04-16 16:20:36 -06:00
|
|
|
return;
|
2008-04-29 02:00:57 -06:00
|
|
|
tsk->sysvsem.undo_list = NULL;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2017-09-08 17:17:42 -06:00
|
|
|
if (!refcount_dec_and_test(&ulp->refcnt))
|
2005-04-16 16:20:36 -06:00
|
|
|
return;
|
|
|
|
|
2008-07-25 02:48:06 -06:00
|
|
|
for (;;) {
|
2005-04-16 16:20:36 -06:00
|
|
|
struct sem_array *sma;
|
2008-07-25 02:48:06 -06:00
|
|
|
struct sem_undo *un;
|
2013-04-30 20:15:44 -06:00
|
|
|
int semid, i;
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
DEFINE_WAKE_Q(wake_q);
|
2008-07-25 02:48:04 -06:00
|
|
|
|
2016-10-11 14:55:05 -06:00
|
|
|
cond_resched();
|
|
|
|
|
2008-07-25 02:48:06 -06:00
|
|
|
rcu_read_lock();
|
2009-04-14 12:17:16 -06:00
|
|
|
un = list_entry_rcu(ulp->list_proc.next,
|
|
|
|
struct sem_undo, list_proc);
|
ipc,sem: fix use after free on IPC_RMID after a task using same semaphore set exits
The current semaphore code allows a potential use after free: in
exit_sem we may free the task's sem_undo_list while there is still
another task looping through the same semaphore set and cleaning the
sem_undo list at freeary function (the task called IPC_RMID for the same
semaphore set).
For example, with a test program [1] running which keeps forking a lot
of processes (which then do a semop call with SEM_UNDO flag), and with
the parent right after removing the semaphore set with IPC_RMID, and a
kernel built with CONFIG_SLAB, CONFIG_SLAB_DEBUG and
CONFIG_DEBUG_SPINLOCK, you can easily see something like the following
in the kernel log:
Slab corruption (Not tainted): kmalloc-64 start=ffff88003b45c1c0, len=64
000: 6b 6b 6b 6b 6b 6b 6b 6b 00 6b 6b 6b 6b 6b 6b 6b kkkkkkkk.kkkkkkk
010: ff ff ff ff 6b 6b 6b 6b ff ff ff ff ff ff ff ff ....kkkk........
Prev obj: start=ffff88003b45c180, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff c0 fb 01 37 00 88 ff ff ...........7....
Next obj: start=ffff88003b45c200, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff 68 29 a7 3c 00 88 ff ff ........h).<....
BUG: spinlock wrong CPU on CPU#2, test/18028
general protection fault: 0000 [#1] SMP
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 2 PID: 18028 Comm: test Not tainted 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: spin_dump+0x53/0xc0
Call Trace:
spin_bug+0x30/0x40
do_raw_spin_unlock+0x71/0xa0
_raw_spin_unlock+0xe/0x10
freeary+0x82/0x2a0
? _raw_spin_lock+0xe/0x10
semctl_down.clone.0+0xce/0x160
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
SyS_semctl+0x236/0x2c0
? syscall_trace_leave+0xde/0x130
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 8b 80 88 03 00 00 48 8d 88 60 05 00 00 48 c7 c7 a0 2c a4 81 31 c0 65 8b 15 eb 40 f3 7e e8 08 31 68 00 4d 85 e4 44 8b 4b 08 74 5e <45> 8b 84 24 88 03 00 00 49 8d 8c 24 60 05 00 00 8b 53 04 48 89
RIP [<ffffffff810d6053>] spin_dump+0x53/0xc0
RSP <ffff88003750fd68>
---[ end trace 783ebb76612867a0 ]---
NMI watchdog: BUG: soft lockup - CPU#3 stuck for 22s! [test:18053]
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 3 PID: 18053 Comm: test Tainted: G D 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: native_read_tsc+0x0/0x20
Call Trace:
? delay_tsc+0x40/0x70
__delay+0xf/0x20
do_raw_spin_lock+0x96/0x140
_raw_spin_lock+0xe/0x10
sem_lock_and_putref+0x11/0x70
SYSC_semtimedop+0x7bf/0x960
? handle_mm_fault+0xbf6/0x1880
? dequeue_task_fair+0x79/0x4a0
? __do_page_fault+0x19a/0x430
? kfree_debugcheck+0x16/0x40
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
? do_audit_syscall_entry+0x66/0x70
? syscall_trace_enter_phase1+0x139/0x160
SyS_semtimedop+0xe/0x10
SyS_semop+0x10/0x20
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 47 10 83 e8 01 85 c0 89 47 10 75 08 65 48 89 3d 1f 74 ff 7e c9 c3 0f 1f 44 00 00 55 48 89 e5 e8 87 17 04 00 66 90 c9 c3 0f 1f 00 <55> 48 89 e5 0f 31 89 c1 48 89 d0 48 c1 e0 20 89 c9 48 09 c8 c9
Kernel panic - not syncing: softlockup: hung tasks
I wasn't able to trigger any badness on a recent kernel without the
proper config debugs enabled, however I have softlockup reports on some
kernel versions, in the semaphore code, which are similar as above (the
scenario is seen on some servers running IBM DB2 which uses semaphore
syscalls).
The patch here fixes the race against freeary, by acquiring or waiting
on the sem_undo_list lock as necessary (exit_sem can race with freeary,
while freeary sets un->semid to -1 and removes the same sem_undo from
list_proc or when it removes the last sem_undo).
After the patch I'm unable to reproduce the problem using the test case
[1].
[1] Test case used below:
#include <stdio.h>
#include <sys/types.h>
#include <sys/ipc.h>
#include <sys/sem.h>
#include <sys/wait.h>
#include <stdlib.h>
#include <time.h>
#include <unistd.h>
#include <errno.h>
#define NSEM 1
#define NSET 5
int sid[NSET];
void thread()
{
struct sembuf op;
int s;
uid_t pid = getuid();
s = rand() % NSET;
op.sem_num = pid % NSEM;
op.sem_op = 1;
op.sem_flg = SEM_UNDO;
semop(sid[s], &op, 1);
exit(EXIT_SUCCESS);
}
void create_set()
{
int i, j;
pid_t p;
union {
int val;
struct semid_ds *buf;
unsigned short int *array;
struct seminfo *__buf;
} un;
/* Create and initialize semaphore set */
for (i = 0; i < NSET; i++) {
sid[i] = semget(IPC_PRIVATE , NSEM, 0644 | IPC_CREAT);
if (sid[i] < 0) {
perror("semget");
exit(EXIT_FAILURE);
}
}
un.val = 0;
for (i = 0; i < NSET; i++) {
for (j = 0; j < NSEM; j++) {
if (semctl(sid[i], j, SETVAL, un) < 0)
perror("semctl");
}
}
/* Launch threads that operate on semaphore set */
for (i = 0; i < NSEM * NSET * NSET; i++) {
p = fork();
if (p < 0)
perror("fork");
if (p == 0)
thread();
}
/* Free semaphore set */
for (i = 0; i < NSET; i++) {
if (semctl(sid[i], NSEM, IPC_RMID))
perror("IPC_RMID");
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
int main(int argc, char **argv)
{
pid_t p;
srand(time(NULL));
while (1) {
p = fork();
if (p < 0) {
perror("fork");
exit(EXIT_FAILURE);
}
if (p == 0) {
create_set();
goto end;
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
end:
return 0;
}
[akpm@linux-foundation.org: use normal comment layout]
Signed-off-by: Herton R. Krzesinski <herton@redhat.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Rafael Aquini <aquini@redhat.com>
CC: Aristeu Rozanski <aris@redhat.com>
Cc: David Jeffery <djeffery@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-08-14 16:35:02 -06:00
|
|
|
if (&un->list_proc == &ulp->list_proc) {
|
|
|
|
/*
|
|
|
|
* We must wait for freeary() before freeing this ulp,
|
|
|
|
* in case we raced with last sem_undo. There is a small
|
|
|
|
* possibility where we exit while freeary() didn't
|
|
|
|
* finish unlocking sem_undo_list.
|
|
|
|
*/
|
2017-06-29 13:59:32 -06:00
|
|
|
spin_lock(&ulp->lock);
|
|
|
|
spin_unlock(&ulp->lock);
|
ipc,sem: fix use after free on IPC_RMID after a task using same semaphore set exits
The current semaphore code allows a potential use after free: in
exit_sem we may free the task's sem_undo_list while there is still
another task looping through the same semaphore set and cleaning the
sem_undo list at freeary function (the task called IPC_RMID for the same
semaphore set).
For example, with a test program [1] running which keeps forking a lot
of processes (which then do a semop call with SEM_UNDO flag), and with
the parent right after removing the semaphore set with IPC_RMID, and a
kernel built with CONFIG_SLAB, CONFIG_SLAB_DEBUG and
CONFIG_DEBUG_SPINLOCK, you can easily see something like the following
in the kernel log:
Slab corruption (Not tainted): kmalloc-64 start=ffff88003b45c1c0, len=64
000: 6b 6b 6b 6b 6b 6b 6b 6b 00 6b 6b 6b 6b 6b 6b 6b kkkkkkkk.kkkkkkk
010: ff ff ff ff 6b 6b 6b 6b ff ff ff ff ff ff ff ff ....kkkk........
Prev obj: start=ffff88003b45c180, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff c0 fb 01 37 00 88 ff ff ...........7....
Next obj: start=ffff88003b45c200, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff 68 29 a7 3c 00 88 ff ff ........h).<....
BUG: spinlock wrong CPU on CPU#2, test/18028
general protection fault: 0000 [#1] SMP
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 2 PID: 18028 Comm: test Not tainted 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: spin_dump+0x53/0xc0
Call Trace:
spin_bug+0x30/0x40
do_raw_spin_unlock+0x71/0xa0
_raw_spin_unlock+0xe/0x10
freeary+0x82/0x2a0
? _raw_spin_lock+0xe/0x10
semctl_down.clone.0+0xce/0x160
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
SyS_semctl+0x236/0x2c0
? syscall_trace_leave+0xde/0x130
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 8b 80 88 03 00 00 48 8d 88 60 05 00 00 48 c7 c7 a0 2c a4 81 31 c0 65 8b 15 eb 40 f3 7e e8 08 31 68 00 4d 85 e4 44 8b 4b 08 74 5e <45> 8b 84 24 88 03 00 00 49 8d 8c 24 60 05 00 00 8b 53 04 48 89
RIP [<ffffffff810d6053>] spin_dump+0x53/0xc0
RSP <ffff88003750fd68>
---[ end trace 783ebb76612867a0 ]---
NMI watchdog: BUG: soft lockup - CPU#3 stuck for 22s! [test:18053]
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 3 PID: 18053 Comm: test Tainted: G D 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: native_read_tsc+0x0/0x20
Call Trace:
? delay_tsc+0x40/0x70
__delay+0xf/0x20
do_raw_spin_lock+0x96/0x140
_raw_spin_lock+0xe/0x10
sem_lock_and_putref+0x11/0x70
SYSC_semtimedop+0x7bf/0x960
? handle_mm_fault+0xbf6/0x1880
? dequeue_task_fair+0x79/0x4a0
? __do_page_fault+0x19a/0x430
? kfree_debugcheck+0x16/0x40
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
? do_audit_syscall_entry+0x66/0x70
? syscall_trace_enter_phase1+0x139/0x160
SyS_semtimedop+0xe/0x10
SyS_semop+0x10/0x20
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 47 10 83 e8 01 85 c0 89 47 10 75 08 65 48 89 3d 1f 74 ff 7e c9 c3 0f 1f 44 00 00 55 48 89 e5 e8 87 17 04 00 66 90 c9 c3 0f 1f 00 <55> 48 89 e5 0f 31 89 c1 48 89 d0 48 c1 e0 20 89 c9 48 09 c8 c9
Kernel panic - not syncing: softlockup: hung tasks
I wasn't able to trigger any badness on a recent kernel without the
proper config debugs enabled, however I have softlockup reports on some
kernel versions, in the semaphore code, which are similar as above (the
scenario is seen on some servers running IBM DB2 which uses semaphore
syscalls).
The patch here fixes the race against freeary, by acquiring or waiting
on the sem_undo_list lock as necessary (exit_sem can race with freeary,
while freeary sets un->semid to -1 and removes the same sem_undo from
list_proc or when it removes the last sem_undo).
After the patch I'm unable to reproduce the problem using the test case
[1].
[1] Test case used below:
#include <stdio.h>
#include <sys/types.h>
#include <sys/ipc.h>
#include <sys/sem.h>
#include <sys/wait.h>
#include <stdlib.h>
#include <time.h>
#include <unistd.h>
#include <errno.h>
#define NSEM 1
#define NSET 5
int sid[NSET];
void thread()
{
struct sembuf op;
int s;
uid_t pid = getuid();
s = rand() % NSET;
op.sem_num = pid % NSEM;
op.sem_op = 1;
op.sem_flg = SEM_UNDO;
semop(sid[s], &op, 1);
exit(EXIT_SUCCESS);
}
void create_set()
{
int i, j;
pid_t p;
union {
int val;
struct semid_ds *buf;
unsigned short int *array;
struct seminfo *__buf;
} un;
/* Create and initialize semaphore set */
for (i = 0; i < NSET; i++) {
sid[i] = semget(IPC_PRIVATE , NSEM, 0644 | IPC_CREAT);
if (sid[i] < 0) {
perror("semget");
exit(EXIT_FAILURE);
}
}
un.val = 0;
for (i = 0; i < NSET; i++) {
for (j = 0; j < NSEM; j++) {
if (semctl(sid[i], j, SETVAL, un) < 0)
perror("semctl");
}
}
/* Launch threads that operate on semaphore set */
for (i = 0; i < NSEM * NSET * NSET; i++) {
p = fork();
if (p < 0)
perror("fork");
if (p == 0)
thread();
}
/* Free semaphore set */
for (i = 0; i < NSET; i++) {
if (semctl(sid[i], NSEM, IPC_RMID))
perror("IPC_RMID");
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
int main(int argc, char **argv)
{
pid_t p;
srand(time(NULL));
while (1) {
p = fork();
if (p < 0) {
perror("fork");
exit(EXIT_FAILURE);
}
if (p == 0) {
create_set();
goto end;
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
end:
return 0;
}
[akpm@linux-foundation.org: use normal comment layout]
Signed-off-by: Herton R. Krzesinski <herton@redhat.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Rafael Aquini <aquini@redhat.com>
CC: Aristeu Rozanski <aris@redhat.com>
Cc: David Jeffery <djeffery@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-08-14 16:35:02 -06:00
|
|
|
rcu_read_unlock();
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
spin_lock(&ulp->lock);
|
|
|
|
semid = un->semid;
|
|
|
|
spin_unlock(&ulp->lock);
|
2008-07-25 02:48:04 -06:00
|
|
|
|
ipc,sem: fix use after free on IPC_RMID after a task using same semaphore set exits
The current semaphore code allows a potential use after free: in
exit_sem we may free the task's sem_undo_list while there is still
another task looping through the same semaphore set and cleaning the
sem_undo list at freeary function (the task called IPC_RMID for the same
semaphore set).
For example, with a test program [1] running which keeps forking a lot
of processes (which then do a semop call with SEM_UNDO flag), and with
the parent right after removing the semaphore set with IPC_RMID, and a
kernel built with CONFIG_SLAB, CONFIG_SLAB_DEBUG and
CONFIG_DEBUG_SPINLOCK, you can easily see something like the following
in the kernel log:
Slab corruption (Not tainted): kmalloc-64 start=ffff88003b45c1c0, len=64
000: 6b 6b 6b 6b 6b 6b 6b 6b 00 6b 6b 6b 6b 6b 6b 6b kkkkkkkk.kkkkkkk
010: ff ff ff ff 6b 6b 6b 6b ff ff ff ff ff ff ff ff ....kkkk........
Prev obj: start=ffff88003b45c180, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff c0 fb 01 37 00 88 ff ff ...........7....
Next obj: start=ffff88003b45c200, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff 68 29 a7 3c 00 88 ff ff ........h).<....
BUG: spinlock wrong CPU on CPU#2, test/18028
general protection fault: 0000 [#1] SMP
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 2 PID: 18028 Comm: test Not tainted 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: spin_dump+0x53/0xc0
Call Trace:
spin_bug+0x30/0x40
do_raw_spin_unlock+0x71/0xa0
_raw_spin_unlock+0xe/0x10
freeary+0x82/0x2a0
? _raw_spin_lock+0xe/0x10
semctl_down.clone.0+0xce/0x160
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
SyS_semctl+0x236/0x2c0
? syscall_trace_leave+0xde/0x130
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 8b 80 88 03 00 00 48 8d 88 60 05 00 00 48 c7 c7 a0 2c a4 81 31 c0 65 8b 15 eb 40 f3 7e e8 08 31 68 00 4d 85 e4 44 8b 4b 08 74 5e <45> 8b 84 24 88 03 00 00 49 8d 8c 24 60 05 00 00 8b 53 04 48 89
RIP [<ffffffff810d6053>] spin_dump+0x53/0xc0
RSP <ffff88003750fd68>
---[ end trace 783ebb76612867a0 ]---
NMI watchdog: BUG: soft lockup - CPU#3 stuck for 22s! [test:18053]
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 3 PID: 18053 Comm: test Tainted: G D 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: native_read_tsc+0x0/0x20
Call Trace:
? delay_tsc+0x40/0x70
__delay+0xf/0x20
do_raw_spin_lock+0x96/0x140
_raw_spin_lock+0xe/0x10
sem_lock_and_putref+0x11/0x70
SYSC_semtimedop+0x7bf/0x960
? handle_mm_fault+0xbf6/0x1880
? dequeue_task_fair+0x79/0x4a0
? __do_page_fault+0x19a/0x430
? kfree_debugcheck+0x16/0x40
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
? do_audit_syscall_entry+0x66/0x70
? syscall_trace_enter_phase1+0x139/0x160
SyS_semtimedop+0xe/0x10
SyS_semop+0x10/0x20
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 47 10 83 e8 01 85 c0 89 47 10 75 08 65 48 89 3d 1f 74 ff 7e c9 c3 0f 1f 44 00 00 55 48 89 e5 e8 87 17 04 00 66 90 c9 c3 0f 1f 00 <55> 48 89 e5 0f 31 89 c1 48 89 d0 48 c1 e0 20 89 c9 48 09 c8 c9
Kernel panic - not syncing: softlockup: hung tasks
I wasn't able to trigger any badness on a recent kernel without the
proper config debugs enabled, however I have softlockup reports on some
kernel versions, in the semaphore code, which are similar as above (the
scenario is seen on some servers running IBM DB2 which uses semaphore
syscalls).
The patch here fixes the race against freeary, by acquiring or waiting
on the sem_undo_list lock as necessary (exit_sem can race with freeary,
while freeary sets un->semid to -1 and removes the same sem_undo from
list_proc or when it removes the last sem_undo).
After the patch I'm unable to reproduce the problem using the test case
[1].
[1] Test case used below:
#include <stdio.h>
#include <sys/types.h>
#include <sys/ipc.h>
#include <sys/sem.h>
#include <sys/wait.h>
#include <stdlib.h>
#include <time.h>
#include <unistd.h>
#include <errno.h>
#define NSEM 1
#define NSET 5
int sid[NSET];
void thread()
{
struct sembuf op;
int s;
uid_t pid = getuid();
s = rand() % NSET;
op.sem_num = pid % NSEM;
op.sem_op = 1;
op.sem_flg = SEM_UNDO;
semop(sid[s], &op, 1);
exit(EXIT_SUCCESS);
}
void create_set()
{
int i, j;
pid_t p;
union {
int val;
struct semid_ds *buf;
unsigned short int *array;
struct seminfo *__buf;
} un;
/* Create and initialize semaphore set */
for (i = 0; i < NSET; i++) {
sid[i] = semget(IPC_PRIVATE , NSEM, 0644 | IPC_CREAT);
if (sid[i] < 0) {
perror("semget");
exit(EXIT_FAILURE);
}
}
un.val = 0;
for (i = 0; i < NSET; i++) {
for (j = 0; j < NSEM; j++) {
if (semctl(sid[i], j, SETVAL, un) < 0)
perror("semctl");
}
}
/* Launch threads that operate on semaphore set */
for (i = 0; i < NSEM * NSET * NSET; i++) {
p = fork();
if (p < 0)
perror("fork");
if (p == 0)
thread();
}
/* Free semaphore set */
for (i = 0; i < NSET; i++) {
if (semctl(sid[i], NSEM, IPC_RMID))
perror("IPC_RMID");
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
int main(int argc, char **argv)
{
pid_t p;
srand(time(NULL));
while (1) {
p = fork();
if (p < 0) {
perror("fork");
exit(EXIT_FAILURE);
}
if (p == 0) {
create_set();
goto end;
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
end:
return 0;
}
[akpm@linux-foundation.org: use normal comment layout]
Signed-off-by: Herton R. Krzesinski <herton@redhat.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Rafael Aquini <aquini@redhat.com>
CC: Aristeu Rozanski <aris@redhat.com>
Cc: David Jeffery <djeffery@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-08-14 16:35:02 -06:00
|
|
|
/* exit_sem raced with IPC_RMID, nothing to do */
|
2013-04-30 20:15:44 -06:00
|
|
|
if (semid == -1) {
|
|
|
|
rcu_read_unlock();
|
ipc,sem: fix use after free on IPC_RMID after a task using same semaphore set exits
The current semaphore code allows a potential use after free: in
exit_sem we may free the task's sem_undo_list while there is still
another task looping through the same semaphore set and cleaning the
sem_undo list at freeary function (the task called IPC_RMID for the same
semaphore set).
For example, with a test program [1] running which keeps forking a lot
of processes (which then do a semop call with SEM_UNDO flag), and with
the parent right after removing the semaphore set with IPC_RMID, and a
kernel built with CONFIG_SLAB, CONFIG_SLAB_DEBUG and
CONFIG_DEBUG_SPINLOCK, you can easily see something like the following
in the kernel log:
Slab corruption (Not tainted): kmalloc-64 start=ffff88003b45c1c0, len=64
000: 6b 6b 6b 6b 6b 6b 6b 6b 00 6b 6b 6b 6b 6b 6b 6b kkkkkkkk.kkkkkkk
010: ff ff ff ff 6b 6b 6b 6b ff ff ff ff ff ff ff ff ....kkkk........
Prev obj: start=ffff88003b45c180, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff c0 fb 01 37 00 88 ff ff ...........7....
Next obj: start=ffff88003b45c200, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff 68 29 a7 3c 00 88 ff ff ........h).<....
BUG: spinlock wrong CPU on CPU#2, test/18028
general protection fault: 0000 [#1] SMP
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 2 PID: 18028 Comm: test Not tainted 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: spin_dump+0x53/0xc0
Call Trace:
spin_bug+0x30/0x40
do_raw_spin_unlock+0x71/0xa0
_raw_spin_unlock+0xe/0x10
freeary+0x82/0x2a0
? _raw_spin_lock+0xe/0x10
semctl_down.clone.0+0xce/0x160
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
SyS_semctl+0x236/0x2c0
? syscall_trace_leave+0xde/0x130
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 8b 80 88 03 00 00 48 8d 88 60 05 00 00 48 c7 c7 a0 2c a4 81 31 c0 65 8b 15 eb 40 f3 7e e8 08 31 68 00 4d 85 e4 44 8b 4b 08 74 5e <45> 8b 84 24 88 03 00 00 49 8d 8c 24 60 05 00 00 8b 53 04 48 89
RIP [<ffffffff810d6053>] spin_dump+0x53/0xc0
RSP <ffff88003750fd68>
---[ end trace 783ebb76612867a0 ]---
NMI watchdog: BUG: soft lockup - CPU#3 stuck for 22s! [test:18053]
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 3 PID: 18053 Comm: test Tainted: G D 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: native_read_tsc+0x0/0x20
Call Trace:
? delay_tsc+0x40/0x70
__delay+0xf/0x20
do_raw_spin_lock+0x96/0x140
_raw_spin_lock+0xe/0x10
sem_lock_and_putref+0x11/0x70
SYSC_semtimedop+0x7bf/0x960
? handle_mm_fault+0xbf6/0x1880
? dequeue_task_fair+0x79/0x4a0
? __do_page_fault+0x19a/0x430
? kfree_debugcheck+0x16/0x40
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
? do_audit_syscall_entry+0x66/0x70
? syscall_trace_enter_phase1+0x139/0x160
SyS_semtimedop+0xe/0x10
SyS_semop+0x10/0x20
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 47 10 83 e8 01 85 c0 89 47 10 75 08 65 48 89 3d 1f 74 ff 7e c9 c3 0f 1f 44 00 00 55 48 89 e5 e8 87 17 04 00 66 90 c9 c3 0f 1f 00 <55> 48 89 e5 0f 31 89 c1 48 89 d0 48 c1 e0 20 89 c9 48 09 c8 c9
Kernel panic - not syncing: softlockup: hung tasks
I wasn't able to trigger any badness on a recent kernel without the
proper config debugs enabled, however I have softlockup reports on some
kernel versions, in the semaphore code, which are similar as above (the
scenario is seen on some servers running IBM DB2 which uses semaphore
syscalls).
The patch here fixes the race against freeary, by acquiring or waiting
on the sem_undo_list lock as necessary (exit_sem can race with freeary,
while freeary sets un->semid to -1 and removes the same sem_undo from
list_proc or when it removes the last sem_undo).
After the patch I'm unable to reproduce the problem using the test case
[1].
[1] Test case used below:
#include <stdio.h>
#include <sys/types.h>
#include <sys/ipc.h>
#include <sys/sem.h>
#include <sys/wait.h>
#include <stdlib.h>
#include <time.h>
#include <unistd.h>
#include <errno.h>
#define NSEM 1
#define NSET 5
int sid[NSET];
void thread()
{
struct sembuf op;
int s;
uid_t pid = getuid();
s = rand() % NSET;
op.sem_num = pid % NSEM;
op.sem_op = 1;
op.sem_flg = SEM_UNDO;
semop(sid[s], &op, 1);
exit(EXIT_SUCCESS);
}
void create_set()
{
int i, j;
pid_t p;
union {
int val;
struct semid_ds *buf;
unsigned short int *array;
struct seminfo *__buf;
} un;
/* Create and initialize semaphore set */
for (i = 0; i < NSET; i++) {
sid[i] = semget(IPC_PRIVATE , NSEM, 0644 | IPC_CREAT);
if (sid[i] < 0) {
perror("semget");
exit(EXIT_FAILURE);
}
}
un.val = 0;
for (i = 0; i < NSET; i++) {
for (j = 0; j < NSEM; j++) {
if (semctl(sid[i], j, SETVAL, un) < 0)
perror("semctl");
}
}
/* Launch threads that operate on semaphore set */
for (i = 0; i < NSEM * NSET * NSET; i++) {
p = fork();
if (p < 0)
perror("fork");
if (p == 0)
thread();
}
/* Free semaphore set */
for (i = 0; i < NSET; i++) {
if (semctl(sid[i], NSEM, IPC_RMID))
perror("IPC_RMID");
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
int main(int argc, char **argv)
{
pid_t p;
srand(time(NULL));
while (1) {
p = fork();
if (p < 0) {
perror("fork");
exit(EXIT_FAILURE);
}
if (p == 0) {
create_set();
goto end;
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
end:
return 0;
}
[akpm@linux-foundation.org: use normal comment layout]
Signed-off-by: Herton R. Krzesinski <herton@redhat.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Rafael Aquini <aquini@redhat.com>
CC: Aristeu Rozanski <aris@redhat.com>
Cc: David Jeffery <djeffery@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-08-14 16:35:02 -06:00
|
|
|
continue;
|
2013-04-30 20:15:44 -06:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
ipc,sem: fix use after free on IPC_RMID after a task using same semaphore set exits
The current semaphore code allows a potential use after free: in
exit_sem we may free the task's sem_undo_list while there is still
another task looping through the same semaphore set and cleaning the
sem_undo list at freeary function (the task called IPC_RMID for the same
semaphore set).
For example, with a test program [1] running which keeps forking a lot
of processes (which then do a semop call with SEM_UNDO flag), and with
the parent right after removing the semaphore set with IPC_RMID, and a
kernel built with CONFIG_SLAB, CONFIG_SLAB_DEBUG and
CONFIG_DEBUG_SPINLOCK, you can easily see something like the following
in the kernel log:
Slab corruption (Not tainted): kmalloc-64 start=ffff88003b45c1c0, len=64
000: 6b 6b 6b 6b 6b 6b 6b 6b 00 6b 6b 6b 6b 6b 6b 6b kkkkkkkk.kkkkkkk
010: ff ff ff ff 6b 6b 6b 6b ff ff ff ff ff ff ff ff ....kkkk........
Prev obj: start=ffff88003b45c180, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff c0 fb 01 37 00 88 ff ff ...........7....
Next obj: start=ffff88003b45c200, len=64
000: 00 00 00 00 ad 4e ad de ff ff ff ff 5a 5a 5a 5a .....N......ZZZZ
010: ff ff ff ff ff ff ff ff 68 29 a7 3c 00 88 ff ff ........h).<....
BUG: spinlock wrong CPU on CPU#2, test/18028
general protection fault: 0000 [#1] SMP
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 2 PID: 18028 Comm: test Not tainted 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: spin_dump+0x53/0xc0
Call Trace:
spin_bug+0x30/0x40
do_raw_spin_unlock+0x71/0xa0
_raw_spin_unlock+0xe/0x10
freeary+0x82/0x2a0
? _raw_spin_lock+0xe/0x10
semctl_down.clone.0+0xce/0x160
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
SyS_semctl+0x236/0x2c0
? syscall_trace_leave+0xde/0x130
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 8b 80 88 03 00 00 48 8d 88 60 05 00 00 48 c7 c7 a0 2c a4 81 31 c0 65 8b 15 eb 40 f3 7e e8 08 31 68 00 4d 85 e4 44 8b 4b 08 74 5e <45> 8b 84 24 88 03 00 00 49 8d 8c 24 60 05 00 00 8b 53 04 48 89
RIP [<ffffffff810d6053>] spin_dump+0x53/0xc0
RSP <ffff88003750fd68>
---[ end trace 783ebb76612867a0 ]---
NMI watchdog: BUG: soft lockup - CPU#3 stuck for 22s! [test:18053]
Modules linked in: 8021q mrp garp stp llc nf_conntrack_ipv4 nf_defrag_ipv4 ip6t_REJECT nf_reject_ipv6 nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables binfmt_misc ppdev input_leds joydev parport_pc parport floppy serio_raw virtio_balloon virtio_rng virtio_console virtio_net iosf_mbi crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcspkr qxl ttm drm_kms_helper drm snd_hda_codec_generic i2c_piix4 snd_hda_intel snd_hda_codec snd_hda_core snd_hwdep snd_seq snd_seq_device snd_pcm snd_timer snd soundcore crc32c_intel virtio_pci virtio_ring virtio pata_acpi ata_generic [last unloaded: speedstep_lib]
CPU: 3 PID: 18053 Comm: test Tainted: G D 4.2.0-rc5+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.8.1-20150318_183358- 04/01/2014
RIP: native_read_tsc+0x0/0x20
Call Trace:
? delay_tsc+0x40/0x70
__delay+0xf/0x20
do_raw_spin_lock+0x96/0x140
_raw_spin_lock+0xe/0x10
sem_lock_and_putref+0x11/0x70
SYSC_semtimedop+0x7bf/0x960
? handle_mm_fault+0xbf6/0x1880
? dequeue_task_fair+0x79/0x4a0
? __do_page_fault+0x19a/0x430
? kfree_debugcheck+0x16/0x40
? __do_page_fault+0x19a/0x430
? __audit_syscall_entry+0xa8/0x100
? do_audit_syscall_entry+0x66/0x70
? syscall_trace_enter_phase1+0x139/0x160
SyS_semtimedop+0xe/0x10
SyS_semop+0x10/0x20
entry_SYSCALL_64_fastpath+0x12/0x71
Code: 47 10 83 e8 01 85 c0 89 47 10 75 08 65 48 89 3d 1f 74 ff 7e c9 c3 0f 1f 44 00 00 55 48 89 e5 e8 87 17 04 00 66 90 c9 c3 0f 1f 00 <55> 48 89 e5 0f 31 89 c1 48 89 d0 48 c1 e0 20 89 c9 48 09 c8 c9
Kernel panic - not syncing: softlockup: hung tasks
I wasn't able to trigger any badness on a recent kernel without the
proper config debugs enabled, however I have softlockup reports on some
kernel versions, in the semaphore code, which are similar as above (the
scenario is seen on some servers running IBM DB2 which uses semaphore
syscalls).
The patch here fixes the race against freeary, by acquiring or waiting
on the sem_undo_list lock as necessary (exit_sem can race with freeary,
while freeary sets un->semid to -1 and removes the same sem_undo from
list_proc or when it removes the last sem_undo).
After the patch I'm unable to reproduce the problem using the test case
[1].
[1] Test case used below:
#include <stdio.h>
#include <sys/types.h>
#include <sys/ipc.h>
#include <sys/sem.h>
#include <sys/wait.h>
#include <stdlib.h>
#include <time.h>
#include <unistd.h>
#include <errno.h>
#define NSEM 1
#define NSET 5
int sid[NSET];
void thread()
{
struct sembuf op;
int s;
uid_t pid = getuid();
s = rand() % NSET;
op.sem_num = pid % NSEM;
op.sem_op = 1;
op.sem_flg = SEM_UNDO;
semop(sid[s], &op, 1);
exit(EXIT_SUCCESS);
}
void create_set()
{
int i, j;
pid_t p;
union {
int val;
struct semid_ds *buf;
unsigned short int *array;
struct seminfo *__buf;
} un;
/* Create and initialize semaphore set */
for (i = 0; i < NSET; i++) {
sid[i] = semget(IPC_PRIVATE , NSEM, 0644 | IPC_CREAT);
if (sid[i] < 0) {
perror("semget");
exit(EXIT_FAILURE);
}
}
un.val = 0;
for (i = 0; i < NSET; i++) {
for (j = 0; j < NSEM; j++) {
if (semctl(sid[i], j, SETVAL, un) < 0)
perror("semctl");
}
}
/* Launch threads that operate on semaphore set */
for (i = 0; i < NSEM * NSET * NSET; i++) {
p = fork();
if (p < 0)
perror("fork");
if (p == 0)
thread();
}
/* Free semaphore set */
for (i = 0; i < NSET; i++) {
if (semctl(sid[i], NSEM, IPC_RMID))
perror("IPC_RMID");
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
int main(int argc, char **argv)
{
pid_t p;
srand(time(NULL));
while (1) {
p = fork();
if (p < 0) {
perror("fork");
exit(EXIT_FAILURE);
}
if (p == 0) {
create_set();
goto end;
}
/* Wait for forked processes to exit */
while (wait(NULL)) {
if (errno == ECHILD)
break;
};
}
end:
return 0;
}
[akpm@linux-foundation.org: use normal comment layout]
Signed-off-by: Herton R. Krzesinski <herton@redhat.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Rafael Aquini <aquini@redhat.com>
CC: Aristeu Rozanski <aris@redhat.com>
Cc: David Jeffery <djeffery@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-08-14 16:35:02 -06:00
|
|
|
sma = sem_obtain_object_check(tsk->nsproxy->ipc_ns, semid);
|
2008-07-25 02:48:06 -06:00
|
|
|
/* exit_sem raced with IPC_RMID, nothing to do */
|
2013-04-30 20:15:44 -06:00
|
|
|
if (IS_ERR(sma)) {
|
|
|
|
rcu_read_unlock();
|
2008-07-25 02:48:06 -06:00
|
|
|
continue;
|
2013-04-30 20:15:44 -06:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_lock(sma, NULL, -1);
|
2013-10-16 14:46:45 -06:00
|
|
|
/* exit_sem raced with IPC_RMID, nothing to do */
|
2014-01-27 18:07:01 -07:00
|
|
|
if (!ipc_valid_object(&sma->sem_perm)) {
|
2013-10-16 14:46:45 -06:00
|
|
|
sem_unlock(sma, -1);
|
|
|
|
rcu_read_unlock();
|
|
|
|
continue;
|
|
|
|
}
|
2009-12-15 17:47:28 -07:00
|
|
|
un = __lookup_undo(ulp, semid);
|
2008-07-25 02:48:06 -06:00
|
|
|
if (un == NULL) {
|
|
|
|
/* exit_sem raced with IPC_RMID+semget() that created
|
|
|
|
* exactly the same semid. Nothing to do.
|
|
|
|
*/
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
2008-07-25 02:48:06 -06:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* remove un from the linked lists */
|
2013-07-08 17:01:11 -06:00
|
|
|
ipc_assert_locked_object(&sma->sem_perm);
|
2008-07-25 02:48:04 -06:00
|
|
|
list_del(&un->list_id);
|
|
|
|
|
Revert "ipc,sem: remove uneeded sem_undo_list lock usage in exit_sem()"
This reverts commit a97955844807e327df11aa33869009d14d6b7de0.
Commit a97955844807 ("ipc,sem: remove uneeded sem_undo_list lock usage
in exit_sem()") removes a lock that is needed. This leads to a process
looping infinitely in exit_sem() and can also lead to a crash. There is
a reproducer available in [1] and with the commit reverted the issue
does not reproduce anymore.
Using the reproducer found in [1] is fairly easy to reach a point where
one of the child processes is looping infinitely in exit_sem between
for(;;) and if (semid == -1) block, while it's trying to free its last
sem_undo structure which has already been freed by freeary().
Each sem_undo struct is on two lists: one per semaphore set (list_id)
and one per process (list_proc). The list_id list tracks undos by
semaphore set, and the list_proc by process.
Undo structures are removed either by freeary() or by exit_sem(). The
freeary function is invoked when the user invokes a syscall to remove a
semaphore set. During this operation freeary() traverses the list_id
associated with the semaphore set and removes the undo structures from
both the list_id and list_proc lists.
For this case, exit_sem() is called at process exit. Each process
contains a struct sem_undo_list (referred to as "ulp") which contains
the head for the list_proc list. When the process exits, exit_sem()
traverses this list to remove each sem_undo struct. As in freeary(),
whenever a sem_undo struct is removed from list_proc, it is also removed
from the list_id list.
Removing elements from list_id is safe for both exit_sem() and freeary()
due to sem_lock(). Removing elements from list_proc is not safe;
freeary() locks &un->ulp->lock when it performs
list_del_rcu(&un->list_proc) but exit_sem() does not (locking was
removed by commit a97955844807 ("ipc,sem: remove uneeded sem_undo_list
lock usage in exit_sem()").
This can result in the following situation while executing the
reproducer [1] : Consider a child process in exit_sem() and the parent
in freeary() (because of semctl(sid[i], NSEM, IPC_RMID)).
- The list_proc for the child contains the last two undo structs A and
B (the rest have been removed either by exit_sem() or freeary()).
- The semid for A is 1 and semid for B is 2.
- exit_sem() removes A and at the same time freeary() removes B.
- Since A and B have different semid sem_lock() will acquire different
locks for each process and both can proceed.
The bug is that they remove A and B from the same list_proc at the same
time because only freeary() acquires the ulp lock. When exit_sem()
removes A it makes ulp->list_proc.next to point at B and at the same
time freeary() removes B setting B->semid=-1.
At the next iteration of for(;;) loop exit_sem() will try to remove B.
The only way to break from for(;;) is for (&un->list_proc ==
&ulp->list_proc) to be true which is not. Then exit_sem() will check if
B->semid=-1 which is and will continue looping in for(;;) until the
memory for B is reallocated and the value at B->semid is changed.
At that point, exit_sem() will crash attempting to unlink B from the
lists (this can be easily triggered by running the reproducer [1] a
second time).
To prove this scenario instrumentation was added to keep information
about each sem_undo (un) struct that is removed per process and per
semaphore set (sma).
CPU0 CPU1
[caller holds sem_lock(sma for A)] ...
freeary() exit_sem()
... ...
... sem_lock(sma for B)
spin_lock(A->ulp->lock) ...
list_del_rcu(un_A->list_proc) list_del_rcu(un_B->list_proc)
Undo structures A and B have different semid and sem_lock() operations
proceed. However they belong to the same list_proc list and they are
removed at the same time. This results into ulp->list_proc.next
pointing to the address of B which is already removed.
After reverting commit a97955844807 ("ipc,sem: remove uneeded
sem_undo_list lock usage in exit_sem()") the issue was no longer
reproducible.
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1694779
Link: http://lkml.kernel.org/r/20191211191318.11860-1-ioanna-maria.alifieraki@canonical.com
Fixes: a97955844807 ("ipc,sem: remove uneeded sem_undo_list lock usage in exit_sem()")
Signed-off-by: Ioanna Alifieraki <ioanna-maria.alifieraki@canonical.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Acked-by: Herton R. Krzesinski <herton@redhat.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: <malat@debian.org>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Jay Vosburgh <jay.vosburgh@canonical.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-20 21:04:00 -07:00
|
|
|
spin_lock(&ulp->lock);
|
2008-07-25 02:48:06 -06:00
|
|
|
list_del_rcu(&un->list_proc);
|
Revert "ipc,sem: remove uneeded sem_undo_list lock usage in exit_sem()"
This reverts commit a97955844807e327df11aa33869009d14d6b7de0.
Commit a97955844807 ("ipc,sem: remove uneeded sem_undo_list lock usage
in exit_sem()") removes a lock that is needed. This leads to a process
looping infinitely in exit_sem() and can also lead to a crash. There is
a reproducer available in [1] and with the commit reverted the issue
does not reproduce anymore.
Using the reproducer found in [1] is fairly easy to reach a point where
one of the child processes is looping infinitely in exit_sem between
for(;;) and if (semid == -1) block, while it's trying to free its last
sem_undo structure which has already been freed by freeary().
Each sem_undo struct is on two lists: one per semaphore set (list_id)
and one per process (list_proc). The list_id list tracks undos by
semaphore set, and the list_proc by process.
Undo structures are removed either by freeary() or by exit_sem(). The
freeary function is invoked when the user invokes a syscall to remove a
semaphore set. During this operation freeary() traverses the list_id
associated with the semaphore set and removes the undo structures from
both the list_id and list_proc lists.
For this case, exit_sem() is called at process exit. Each process
contains a struct sem_undo_list (referred to as "ulp") which contains
the head for the list_proc list. When the process exits, exit_sem()
traverses this list to remove each sem_undo struct. As in freeary(),
whenever a sem_undo struct is removed from list_proc, it is also removed
from the list_id list.
Removing elements from list_id is safe for both exit_sem() and freeary()
due to sem_lock(). Removing elements from list_proc is not safe;
freeary() locks &un->ulp->lock when it performs
list_del_rcu(&un->list_proc) but exit_sem() does not (locking was
removed by commit a97955844807 ("ipc,sem: remove uneeded sem_undo_list
lock usage in exit_sem()").
This can result in the following situation while executing the
reproducer [1] : Consider a child process in exit_sem() and the parent
in freeary() (because of semctl(sid[i], NSEM, IPC_RMID)).
- The list_proc for the child contains the last two undo structs A and
B (the rest have been removed either by exit_sem() or freeary()).
- The semid for A is 1 and semid for B is 2.
- exit_sem() removes A and at the same time freeary() removes B.
- Since A and B have different semid sem_lock() will acquire different
locks for each process and both can proceed.
The bug is that they remove A and B from the same list_proc at the same
time because only freeary() acquires the ulp lock. When exit_sem()
removes A it makes ulp->list_proc.next to point at B and at the same
time freeary() removes B setting B->semid=-1.
At the next iteration of for(;;) loop exit_sem() will try to remove B.
The only way to break from for(;;) is for (&un->list_proc ==
&ulp->list_proc) to be true which is not. Then exit_sem() will check if
B->semid=-1 which is and will continue looping in for(;;) until the
memory for B is reallocated and the value at B->semid is changed.
At that point, exit_sem() will crash attempting to unlink B from the
lists (this can be easily triggered by running the reproducer [1] a
second time).
To prove this scenario instrumentation was added to keep information
about each sem_undo (un) struct that is removed per process and per
semaphore set (sma).
CPU0 CPU1
[caller holds sem_lock(sma for A)] ...
freeary() exit_sem()
... ...
... sem_lock(sma for B)
spin_lock(A->ulp->lock) ...
list_del_rcu(un_A->list_proc) list_del_rcu(un_B->list_proc)
Undo structures A and B have different semid and sem_lock() operations
proceed. However they belong to the same list_proc list and they are
removed at the same time. This results into ulp->list_proc.next
pointing to the address of B which is already removed.
After reverting commit a97955844807 ("ipc,sem: remove uneeded
sem_undo_list lock usage in exit_sem()") the issue was no longer
reproducible.
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1694779
Link: http://lkml.kernel.org/r/20191211191318.11860-1-ioanna-maria.alifieraki@canonical.com
Fixes: a97955844807 ("ipc,sem: remove uneeded sem_undo_list lock usage in exit_sem()")
Signed-off-by: Ioanna Alifieraki <ioanna-maria.alifieraki@canonical.com>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Acked-by: Herton R. Krzesinski <herton@redhat.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: <malat@debian.org>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Jay Vosburgh <jay.vosburgh@canonical.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-20 21:04:00 -07:00
|
|
|
spin_unlock(&ulp->lock);
|
2008-07-25 02:48:06 -06:00
|
|
|
|
2008-07-25 02:48:04 -06:00
|
|
|
/* perform adjustments registered in un */
|
|
|
|
for (i = 0; i < sma->sem_nsems; i++) {
|
2017-07-12 15:34:38 -06:00
|
|
|
struct sem *semaphore = &sma->sems[i];
|
2008-07-25 02:48:04 -06:00
|
|
|
if (un->semadj[i]) {
|
|
|
|
semaphore->semval += un->semadj[i];
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* Range checks of the new semaphore value,
|
|
|
|
* not defined by sus:
|
|
|
|
* - Some unices ignore the undo entirely
|
|
|
|
* (e.g. HP UX 11i 11.22, Tru64 V5.1)
|
|
|
|
* - some cap the value (e.g. FreeBSD caps
|
|
|
|
* at 0, but doesn't enforce SEMVMX)
|
|
|
|
*
|
|
|
|
* Linux caps the semaphore value, both at 0
|
|
|
|
* and at SEMVMX.
|
|
|
|
*
|
2014-01-27 18:07:04 -07:00
|
|
|
* Manfred <manfred@colorfullife.com>
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
2006-03-26 02:37:17 -07:00
|
|
|
if (semaphore->semval < 0)
|
|
|
|
semaphore->semval = 0;
|
|
|
|
if (semaphore->semval > SEMVMX)
|
|
|
|
semaphore->semval = SEMVMX;
|
2018-03-23 00:11:29 -06:00
|
|
|
ipc_update_pid(&semaphore->sempid, task_tgid(current));
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
/* maybe some queued-up processes were waiting for this */
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
do_smart_update(sma, NULL, 0, 1, &wake_q);
|
2013-04-30 20:15:44 -06:00
|
|
|
sem_unlock(sma, -1);
|
ipc: move rcu_read_unlock() out of sem_unlock() and into callers
The IPC locking is a mess, and sem_unlock() unlocks not only the
semaphore spinlock, it also drops the rcu read lock. Unlike sem_lock(),
which just gets the spin-lock, and expects the caller to get the rcu
read lock.
This all makes things very hard to follow, and it's very confusing when
you take the rcu read lock in one function, and then release it in
another. And it has caused actual bugs: the sem_obtain_lock() function
ended up dropping the RCU read lock twice in one error path, because it
first did the sem_unlock(), and then did a rcu_read_unlock() to match
the rcu_read_lock() it had done.
This is just a totally mindless "remove rcu_read_unlock() from
sem_unlock() and add it immediately after each caller" (except for the
aforementioned bug where we did too many rcu_read_unlock(), and in
find_alloc_undo() where we just got the rcu_read_lock() to correct for
the fact that sem_unlock would immediately drop it again).
We can (and should) clean things up further, but this fixes the bug with
the minimal amount of subtlety.
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-05-03 16:04:40 -06:00
|
|
|
rcu_read_unlock();
|
ipc/sem: rework task wakeups
Our sysv sems have been using the notion of lockless wakeups for a
while, ever since commit 0a2b9d4c7967 ("ipc/sem.c: move wake_up_process
out of the spinlock section"), in order to reduce the sem_lock hold
times. This in-house pending queue can be replaced by wake_q (just like
all the rest of ipc now), in that it provides the following advantages:
o Simplifies and gets rid of unnecessary code.
o We get rid of the IN_WAKEUP complexities. Given that wake_q_add()
grabs reference to the task, if awoken due to an unrelated event,
between the wake_q_add() and wake_up_q() window, we cannot race with
sys_exit and the imminent call to wake_up_process().
o By not spinning IN_WAKEUP, we no longer need to disable preemption.
In consequence, the wakeup paths (after schedule(), that is) must
acknowledge an external signal/event, as well spurious wakeup occurring
during the pending wakeup window. Obviously no changes in semantics
that could be visible to the user. The fastpath is _only_ for when we
know for sure that we were awoken due to a the waker's successful semop
call (queue.status is not -EINTR).
On a 48-core Haswell, running the ipcscale 'waitforzero' test, the
following is seen with increasing thread counts:
v4.8-rc5 v4.8-rc5
semopv2
Hmean sembench-sem-2 574733.00 ( 0.00%) 578322.00 ( 0.62%)
Hmean sembench-sem-8 811708.00 ( 0.00%) 824689.00 ( 1.59%)
Hmean sembench-sem-12 842448.00 ( 0.00%) 845409.00 ( 0.35%)
Hmean sembench-sem-21 933003.00 ( 0.00%) 977748.00 ( 4.80%)
Hmean sembench-sem-48 935910.00 ( 0.00%) 1004759.00 ( 7.36%)
Hmean sembench-sem-79 937186.00 ( 0.00%) 983976.00 ( 4.99%)
Hmean sembench-sem-234 974256.00 ( 0.00%) 1060294.00 ( 8.83%)
Hmean sembench-sem-265 975468.00 ( 0.00%) 1016243.00 ( 4.18%)
Hmean sembench-sem-296 991280.00 ( 0.00%) 1042659.00 ( 5.18%)
Hmean sembench-sem-327 975415.00 ( 0.00%) 1029977.00 ( 5.59%)
Hmean sembench-sem-358 1014286.00 ( 0.00%) 1049624.00 ( 3.48%)
Hmean sembench-sem-389 972939.00 ( 0.00%) 1043127.00 ( 7.21%)
Hmean sembench-sem-420 981909.00 ( 0.00%) 1056747.00 ( 7.62%)
Hmean sembench-sem-451 990139.00 ( 0.00%) 1051609.00 ( 6.21%)
Hmean sembench-sem-482 965735.00 ( 0.00%) 1040313.00 ( 7.72%)
[akpm@linux-foundation.org: coding-style fixes]
[sfr@canb.auug.org.au: merge fix for WAKE_Q to DEFINE_WAKE_Q rename]
Link: http://lkml.kernel.org/r/20161122210410.5eca9fc2@canb.auug.org.au
Link: http://lkml.kernel.org/r/1474225896-10066-3-git-send-email-dave@stgolabs.net
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Acked-by: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-14 16:06:34 -07:00
|
|
|
wake_up_q(&wake_q);
|
2008-07-25 02:48:06 -06:00
|
|
|
|
2011-03-17 22:09:35 -06:00
|
|
|
kfree_rcu(un, rcu);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
2008-07-25 02:48:04 -06:00
|
|
|
kfree(ulp);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef CONFIG_PROC_FS
|
2005-09-06 16:17:10 -06:00
|
|
|
static int sysvipc_sem_proc_show(struct seq_file *s, void *it)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2012-02-07 17:54:11 -07:00
|
|
|
struct user_namespace *user_ns = seq_user_ns(s);
|
2017-08-02 14:32:21 -06:00
|
|
|
struct kern_ipc_perm *ipcp = it;
|
|
|
|
struct sem_array *sma = container_of(ipcp, struct sem_array, sem_perm);
|
2017-08-02 20:51:13 -06:00
|
|
|
time64_t sem_otime;
|
2013-07-08 17:01:25 -06:00
|
|
|
|
2013-09-30 14:45:07 -06:00
|
|
|
/*
|
|
|
|
* The proc interface isn't aware of sem_lock(), it calls
|
|
|
|
* ipc_lock_object() directly (in sysvipc_find_ipc).
|
2016-10-11 14:54:50 -06:00
|
|
|
* In order to stay compatible with sem_lock(), we must
|
|
|
|
* enter / leave complex_mode.
|
2013-09-30 14:45:07 -06:00
|
|
|
*/
|
2016-10-11 14:54:50 -06:00
|
|
|
complexmode_enter(sma);
|
2013-09-30 14:45:07 -06:00
|
|
|
|
2013-07-08 17:01:25 -06:00
|
|
|
sem_otime = get_semotime(sma);
|
2005-09-06 16:17:10 -06:00
|
|
|
|
2015-04-15 17:17:54 -06:00
|
|
|
seq_printf(s,
|
2017-08-02 20:51:13 -06:00
|
|
|
"%10d %10d %4o %10u %5u %5u %5u %5u %10llu %10llu\n",
|
2015-04-15 17:17:54 -06:00
|
|
|
sma->sem_perm.key,
|
|
|
|
sma->sem_perm.id,
|
|
|
|
sma->sem_perm.mode,
|
|
|
|
sma->sem_nsems,
|
|
|
|
from_kuid_munged(user_ns, sma->sem_perm.uid),
|
|
|
|
from_kgid_munged(user_ns, sma->sem_perm.gid),
|
|
|
|
from_kuid_munged(user_ns, sma->sem_perm.cuid),
|
|
|
|
from_kgid_munged(user_ns, sma->sem_perm.cgid),
|
|
|
|
sem_otime,
|
|
|
|
sma->sem_ctime);
|
|
|
|
|
2016-10-11 14:54:50 -06:00
|
|
|
complexmode_tryleave(sma);
|
|
|
|
|
2015-04-15 17:17:54 -06:00
|
|
|
return 0;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
#endif
|