To read the gtod we need to grab the xtime lock for read. Reading the gtod
before the TSC can cause a bigger gab if the xtime lock is contended.
This patch simply reverses the order to read the TSC after the gtod.
The locking in the reading of the gtod handles any barriers one might
think is needed.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Cc: Steven Rostedt <srostedt@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: john stultz <johnstul@us.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Reading the CPU clock should try to stay accurate within the CPU.
By reading the CPU clock from another CPU and updating the deltas can
cause unneeded jumps when reading from the local CPU.
This patch changes the code to update the last read TSC only when read
from the local CPU.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Cc: Steven Rostedt <srostedt@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: john stultz <johnstul@us.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The algorithm to calculate the 'now' of another CPU is not correct.
At each scheduler tick, each CPU records the last sched_clock and
gtod (tick_raw and tick_gtod respectively). If the TSC is somewhat the
same in speed between two clocks the algorithm would be:
tick_gtod1 + (now1 - tick_raw1) = tick_gtod2 + (now2 - tick_raw2)
To calculate now2 we would have:
now2 = (tick_gtod1 - tick_gtod2) + (tick_raw2 - tick_raw1) + now1
Currently the algorithm is:
now2 = (tick_gtod1 - tick_gtod2) + (tick_raw1 - tick_raw2) + now1
This solves most of the rest of the issues I've had with timestamps in
ftace.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Working with ftrace I would get large jumps of 11 millisecs or more with
the clock tracer. This killed the latencing timings of ftrace and also
caused the irqoff self tests to fail.
What was happening is with NO_HZ the idle would stop the jiffy counter and
before the jiffy counter was updated the sched_clock would have a bad
delta jiffies to compare with the gtod with the maximum.
The jiffies would stop and the last sched_tick would record the last gtod.
On wakeup, the sched clock update would compare the gtod + delta jiffies
(which would be zero) and compare it to the TSC. The TSC would have
correctly (with a stable TSC) moved forward several jiffies. But because the
jiffies has not been updated yet the clock would be prevented from moving
forward because it would appear that the TSC jumped too far ahead.
The clock would then virtually stop, until the jiffies are updated. Then
the next sched clock update would see that the clock was very much behind
since the delta jiffies is now correct. This would then jump the clock
forward by several jiffies.
This caused ftrace to report several milliseconds of interrupts off
latency at every resume from NO_HZ idle.
This patch adds hooks into the nohz code to disable the checking of the
maximum clock update when nohz is in effect. It resumes the max check
when nohz has updated the jiffies again.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Cc: Steven Rostedt <srostedt@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
With keeping the max and min sched time within one jiffy of the gtod clock
was too tight. Just before a schedule tick the max could easily be hit, as
well as just after a schedule_tick the min could be hit. This caused the
clock to jump around by a jiffy.
This patch widens the minimum to
last gtod + (delta_jiffies ? delta_jiffies - 1 : 0) * TICK_NSECS
and the maximum to
last gtod + (2 + delta_jiffies) * TICK_NSECS
This keeps the minum to gtod or if one jiffy less than delta jiffies
and the maxim 2 jiffies ahead of gtod. This may cause unstable TSCs to be
a bit more sporadic, but it helps keep a clock with a stable TSC working well.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Cc: Steven Rostedt <srostedt@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The sched_clock code tries to keep within the gtod time by one tick (jiffy).
The current code mistakenly keeps track of the delta jiffies between
updates of the clock, where the the delta is used to compare with the
number of jiffies that have past since an update of the gtod. The gtod is
updated at each schedule tick not each sched_clock update. After one
jiffy passes the clock is updated fine. But the delta is taken from the
last update so if the next update happens before the next tick the delta
jiffies used will be incorrect.
This patch changes the code to check the delta of jiffies between ticks
and not updates to match the comparison of the updates with the gtod.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Cc: Steven Rostedt <srostedt@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
PREEMPT_RCU without HOTPLUG_CPU is broken. The rcu_online_cpu is called
to initially populate rcu_cpu_online_map with all online CPUs when the
hotplug event handler is installed, and also to populate the map with
CPUs as they come online. The former case is meant to happen with and
without HOTPLUG_CPU, but without HOTPLUG_CPU, the rcu_offline_cpu
function is no-oped -- while it still gets called, it does not set the
rcu CPU map.
With a blank RCU CPU map, grace periods get to tick by completely
oblivious to active RCU read side critical sections. This results in
free-before-grace bugs.
Fix is obvious once the problem is known. (Also, change __devinit to
__cpuinit so the function gets thrown away on !HOTPLUG_CPU kernels).
Signed-off-by: Nick Piggin <npiggin@suse.de>
Reported-and-tested-by: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
[ Nick is my personal hero of the day - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is the long awaited ftrace.txt. It explains in quite detail how to
use ftrace and the various tracers.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When chainiv postpones requests it never calls their completion functions.
This causes symptoms such as memory leaks when IPsec is in use.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Remove the dubious attempt to prefer 'compute' over 'read'. Not only is it
wrong given commit c337869d (md: do not compute parity unless it is on a failed
drive), but it can trigger a BUG_ON in handle_parity_checks5().
Cc: <stable@kernel.org>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
There is dma_mask in of_device upon of_platform_device_create()
but we don't actually set coherent_dma_mask. This may cause weird
behavior of USB subsystem using of_device USB host drivers.
Signed-off-by: Vitaly Bordug <vitb@kernel.crashing.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The change to iwch_provider.c in commit f4e91eb4 ("IB: convert struct
class_device to struct device") undid the fix done in commit 7f049f2f
("RDMA/cxgb3: Hold rtnl_lock() around ethtool get_drvinfo call"). It
removed the calls to rtnl_lock() that serialized the iw_cxgb3 ethtool
ops calls into the cxgb3 driver. This locking is needed to avoid
messing up the internal state of the cxgb3 driver.
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
On 2.6.26-rc9, the commit 05946bce83
("fsl_diu_fb: fix build with CONFIG_PM=y, plus fix some warnings")
breaks its previous fix f969c5672b
("fsl-diu-db: compile fix")
This patch reverts the broken part.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Acked-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'hotfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6:
SUNRPC: Fix an rpcbind breakage for the case of IPv6 lookups
SUNRPC: Fix a double-free in rpcbind
NFS: Fix readdir cache invalidation
* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
[MIPS] Fix 32bit kernels on R4k with 128 byte cache line size
[MIPS] Atlas, decstation: Fix section mismatches triggered by defconfigs
With the removal of struct file from the xattr code,
reiserfs_file_release() isn't used anymore, so the prealloc isn't
discarded. This causes hangs later down the line.
This patch adds it to reiserfs_delete_inode. In most cases it will be a
no-op due to it already having been called, but will avoid hangs with
xattrs.
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Now that rpcb_next_version has been split into an IPv4 version and an IPv6
version, we Oops when rpcb_call_async attempts to look up the IPv6-specific
RPC procedure in rpcb_next_version.
Fix the Oops simply by having rpcb_getport_async pass the correct RPC
procedure as an argument.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
It is wrong to be freeing up the rpcbind arguments if the call to
rpcb_call_async() fails, since they should already have been freed up by
rpcb_map_release().
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
invalidate_inode_pages2_range() takes page offset arguments, not byte
ranges.
Another thought is that individual pages might perhaps get evicted by VM
pressure, in which case we might perhaps want to re-read not only the
evicted page, but all subsequent pages too (in case the server returns
more/less data per page so that the alignment of the next entry
changes). We should therefore remove the condition that we only do this on
page->index==0.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The generated copy_page for R4k CPU with a 128 byte cache line size used
Create Dirty Exclusive cache line operations even if only part of the
cache line was filled. This change avoids generating cache operations,
if only part of the cache line size is copied in one loop. It also
increases the maxmimum loop size, because the generated code even fits
into the available space for r4k CPUs with 128 byte cache line size.
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Resolve these mismatches by defining affected functions with the __cpuinit
attribute, rather than __init.
Signed-off-by: Shane McDonald <mcdonald.shane@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
The driver uses completely bogus rounding formula for calculating period from
the IDECLK frequency which gives one-off period values (e.g. 11 ns with 100 MHz
IDECLK) which in turn can lead to overclocked IDE transfer timings. Actually,
rounding is just wrong in this case, so use a mere division for a safe result.
While at it, also:
- give 'ide_palm_clk' variable a more suitable name;
- get rid of the useless 'ideclkp' variable;
- drop the LISP stype 'p' postfix from the 'clkp' variable's name. :-)
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Cc: mcherkashin@ru.mvista.com
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Add __ide_default_irq() inline helper and use it instead of
ide_default_irq() in ide-probe.c and ns87415.c (all host drivers
except IDE PCI ones always setup hwif->irq so it is enough to
check only for I/O bases 0x1f0 and 0x170).
This fixes post-2.6.25 regression since ide_default_irq()
define could shadow ide_default_irq() inline.
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
As Andy Whitcroft recently pointed out, the current powerpc version of
huge_ptep_set_wrprotect() has a bug. It just calls ptep_set_wrprotect()
which in turn calls pte_update() then hpte_need_flush() with the 'huge'
argument set to 0. This will cause hpte_need_flush() to flush the wrong
hash entries (of any). Andy's fix for this is already in the powerpc
tree as commit 016b33c495.
I have confirmed this is a real bug, not masked by some other
synchronization, with a new testcase for libhugetlbfs. A process write
a (MAP_PRIVATE) hugepage mapping, fork(), then alter the mapping and
have the child incorrectly see the second write.
Therefore, this should be fixed for 2.6.26, and for the stable tree.
Here is a suitable patch for 2.6.26, which I think will also be suitable
for the stable tree (neither of the headers in question has been changed
much recently).
It is cut down slighlty from Andy's original version, in that it does
not include a 32-bit version of huge_ptep_set_wrprotect(). Currently,
hugepages are not supported on any 32-bit powerpc platform. When they
are, a suitable 32-bit version can be added - the only 32-bit hardware
which supports hugepages does not use the conventional hashtable MMU and
so will have different needs anyway.
Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Stop mprotect's pte_modify from wiping out the s390 pte_special bit, which
caused oops thereafter when vm_normal_page thought X's abnormal was normal.
Debugged-by: Ryan Hope <rmh3093@gmail.com>
Debugged-by: Zan Lynx <zlynx@acm.org>
Acked-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Coverity CID: 2306 & 2307 RESOURCE_LEAK
In the second for loop in test_cipher(), data is allocated space with
kzalloc() and is only ever freed in an error case.
Looking at this loop, data is written to this memory but nothing seems
to read from it.
So here is a patch removing the allocation, I think this is the right
fix.
Only compile tested.
Signed-off-by: Darren Jenkins <darrenrjenkins@gmailcom>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
This reverts commit a167607255. It duplicates
the change from 8d64c781f0 and only one should be
applied, otherwise some of the Dell quirks are lost.
Thanks to Tony Camuso for catching this.
Acked-by: Tony Camuso <tcamuso@redhat.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Patch fixes a race that can result in an oops while adding a
lockres to the dlm lockres tracking list.
Bug introduced by mainline commit 29576f8bb5.
Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Mark Fasheh <mfasheh@suse.com>
There are various constraints on the use of unit-at-a-time:
- i386 uses no-unit-at-a-time for pre-4.0 (not 4.3)
- x86_64 uses unit-at-a-time always
Uli reported a crash on x86_64 with gcc 4.1.2 with unit-at-a-time,
resulting in commit c0a18111e5
Ingo reported a gcc internal error with gcc 4.3 with no-unit-at-a-timem,
resulting in 22eecde2f9
Benny Halevy is seeing extern inlines not resolved with gcc 4.3 with
no-unit-at-a-time
This patch reintroduces unit-at-a-time for gcc >= 4.0, bringing back the
possibility of Uli's crash. If that happens, we'll debug it.
I started seeing both the internal compiler errors and unresolved
inlines on Fedora 9. This patch fixes both problems, without so far
reintroducing the crash reported by Uli.
Signed-off-by: Jeff Dike <jdike@linux.intel.com>
Cc: Benny Halevy <bhalevy@panasas.com>
Cc: Adrian Bunk <bunk@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Ulrich Drepper <drepper@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
A recent patch to legacy_serial.c factored out some code by
using the of_match_node() facility to match a node against
an array of possible matches. However, the patch didn't properly
terminate the array causing potential crashes in cases where no
match is found. In addition, the name of the array was poorly
chosen for a static symbol making debugging harder.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
They print out a pointer in symbolic format, if possible (ie using
symbolic KALLSYMS information). The '%pS' format is for regular direct
pointers (which can point to data or code and that you find on the stack
during backtraces etc), while '%pF' is for C function pointer types.
On most architectures, the two mean exactly the same thing, but some
architectures use an indirect pointer for C function pointers, where the
function pointer points to a function descriptor (which in turn contains
the actual pointer to the code). The '%pF' code automatically does the
appropriate function descriptor dereference on such architectures.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This expands the kernel '%p' handling with an arbitrary alphanumberic
specifier extension string immediately following the '%p'. Right now
it's just being ignored, but the next commit will start adding some
specific pointer type extensions.
NOTE! The reason the extension is appended to the '%p' is to allow
minimal gcc type checking: gcc will still see the '%p' and will check
that the argument passed in is indeed a pointer, and yet will not
complain about the extended information that gcc doesn't understand
about (on the other hand, it also won't actually check that the pointer
type and the extension are compatible).
Alphanumeric characters were chosen because there is no sane existing
use for a string format with a hex pointer representation immediately
followed by alphanumerics (which is what such a format string would have
traditionally resulted in).
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The actual code is the same, just split out into a helper function.
This makes it easier to read, and allows for simple future extension
of %p handling.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The actual code is the same, just split out into a helper function.
This makes it easier to read, and allows for future sharing of the
string code.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The pxa27x DMA controller defaults to 64-bit alignment. This caused
the SCR reads to fail (and, depending on card type, error out) when
card->raw_scr was not aligned on a 8-byte boundary.
For performance reasons all scatter-gather addresses passed to
pxamci_request should be aligned on 8-byte boundaries, but if
this can't be guaranteed, byte aligned DMA transfers in the
have to be enabled in the controller to get correct behaviour.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This reverts commit e872154921.
Andrey Borzenkov reports that it resulted in a totally hung machine for
him when loading the OHCI driver. Extensive netconsole capture with
SysRq output shows that modprobe gets stuck in ohci_hub_status_data()
when probing and enabling the OHCI controller, see for example
http://lkml.org/lkml/2008/7/5/236
for an analysis.
The problem appears to be an interrupt flood triggered by the commit
that gets reverted, and Andrey confirmed that the revert makes things
work for him again.
Reported-and-tested-by: Andrey Borzenkov <arvidjaar@mail.ru>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Acked-by: David Brownell <david-b@pacbell.net>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
