+ ignore-stolen-time-in-the-softlockup-watchdog-fix.patch added to -mm tree

+ ignore-stolen-time-in-the-softlockup-watchdog-fix.patch added to -mm tree

[akpm]

The patch titled
     ignore-stolen-time-in-the-softlockup-watchdog-fix
has been added to the -mm tree.  Its filename is
     ignore-stolen-time-in-the-softlockup-watchdog-fix.patch

*** Remember to use Documentation/SubmitChecklist when testing your code ***

See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
out what to do about this

------------------------------------------------------
Subject: ignore-stolen-time-in-the-softlockup-watchdog-fix
From: Andrew Morton <akpm@linux-foundation.org>

Avoid reevaluating smp_processor_id()

Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Zachary Amsden <zach@vmware.com>
Cc: James Morris <jmorris@namei.org>
Cc: Dan Hecht <dhecht@vmware.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Chris Lalancette <clalance@redhat.com>
Cc: Rick Lindsley <ricklind@us.ibm.com>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 kernel/softlockup.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff -puN kernel/softlockup.c~ignore-stolen-time-in-the-softlockup-watchdog-fix kernel/softlockup.c
--- a/kernel/softlockup.c~ignore-stolen-time-in-the-softlockup-watchdog-fix
+++ a/kernel/softlockup.c
@@ -57,7 +57,7 @@ EXPORT_SYMBOL(touch_softlockup_watchdog)
 void softlockup_tick(void)
 {
 	int this_cpu = smp_processor_id();
-	unsigned long touch_timestamp = __get_cpu_var(touch_timestamp);
+	unsigned long touch_timestamp = per_cpu(touch_timestamp, this_cpu);
 	unsigned long print_timestamp;
 	unsigned long now;
 
@@ -65,7 +65,7 @@ void softlockup_tick(void)
 	if (touch_timestamp == 0)
 		return;
 
-	print_timestamp = __get_cpu_var(print_timestamp);
+	print_timestamp = per_cpu(print_timestamp, this_cpu);
 
 	/* report at most once a second */
 	if (print_timestamp < (touch_timestamp + 1) ||
_

Patches currently in -mm which might be from akpm@linux-foundation.org are

origin.patch
acpi-thermal-fix-mod_timer-interval.patch
packet-fix-error-handling.patch
fix-x86-fix-potential-overflow-in-perfctr-reservation.patch
sctp-fix-sctp_getsockopt_local_addrs_old-to-use-local-storage-fix.patch
slab-introduce-krealloc-fix.patch
git-acpi.patch
git-alsa.patch
git-alsa-fixup.patch
git-agpgart.patch
git-powerpc.patch
ppc4xx_sgdma-needs-dma_mappingh.patch
revert-gregkh-driver-remove-struct-subsystem-as-it-is-no-longer-needed.patch
more-fix-gregkh-driver-sysfs-kill-unnecessary-attribute-owner.patch
even-more-fix-gregkh-driver-sysfs-kill-unnecessary-attribute-owner.patch
even-even-more-fix-gregkh-driver-sysfs-kill-unnecessary-attribute-owner.patch
device_schedule_callback-needs-a-module-reference-fix.patch
define-platform-wakeup-hook-use-in-pci_enable_wake-fix.patch
dev_dbg-check-dev_dbg-arguments-fix.patch
dev_dbg-check-dev_dbg-arguments-fix-2.patch
sysfs-binc-printk-fix.patch
git-drm.patch
git-dvb.patch
git-dvb-vs-gregkh-driver-sysfs-kill-unnecessary-attribute-owner.patch
applesmc-fix-crash-when-activating-a-led-trigger-on-the-keyboard-backlight-use-a-workqueue-fix.patch
git-gfs2-nmw.patch
git-ieee1394.patch
sbp2-include-fixes.patch
git-infiniband-make-it-build-hack.patch
git-input.patch
git-input-fixup.patch
git-kvm.patch
git-libata-all.patch
libata-acpi-add-infrastructure-for-drivers-to-use-fix.patch
pata_acpi-restore-driver-fix.patch
pata_acpi-restore-driver-fix-2.patch
drivers-ata-pata_cmd640c-fix-build-with-config_pm=n.patch
revert-rm-pointless-dmaengine-exports.patch
git-md-accel-fix.patch
git-mmc-build-fix.patch
git-mmc-versus-uevent-use-add_uevent_var-instead-of-open-coding-it.patch
git-mtd.patch
git-ubi.patch
git-ubi-fixup.patch
revert-gitpowerpc-ehea-changes.patch
git-netdev-all.patch
git-netdev-all-fixup.patch
reapply-gitpowerpc-ehea-changes.patch
git-netdev-all-baycom_ser_fdx-fix.patch
vioc-warning-fix.patch
vioc-cast-warning-fix.patch
git-e1000.patch
git-e1000-fixup-2.patch
git-net.patch
git-net-fixup.patch
git-net-fix-yamc.patch
git-net-vs-git-netdev-all.patch
input-rfkill-add-support-for-input-key-to-control-wireless-radio-fixes-fix.patch
input-rfkill-add-support-for-input-key-to-control-wireless-radio-fixes-2.patch
input-rfkill-add-support-for-input-key-to-control-wireless-radio-fixes-3.patch
irda_device_dongle_init-fix-kzallocgfp_kernel-in-spinlock.patch
rfcomm_worker-fix-wakeup-race.patch
git-ocfs2.patch
git-parisc.patch
rm9000-serial-driver.patch
fix-gregkh-pci-pci-remove-the-broken-pci_multithread_probe-option.patch
git-pciseg.patch
git-s390.patch
git-s390-fixup.patch
git-s390-vs-gregkh-driver-sysfs-kill-unnecessary-attribute-owner.patch
git-scsi-misc.patch
scsi-fix-config_scsi_wait_scan=m-fix.patch
scsi-fix-config_scsi_wait_scan=m-fix-fix.patch
pci-error-recovery-symbios-scsi-base-support-fixes.patch
pci-error-recovery-symbios-scsi-first-failure-fix.patch
cxacru-add-documentation-file-fix.patch
fix-x86_64-mm-fam10-mwait-idle.patch
more-fix-x86_64-mm-fam10-mwait-idle.patch
fix-x86_64-mm-sched-clock-share.patch
revert-x86_64-mm-account-for-module-percpu-space-separately-from-kernel-percpu.patch
rename-the-parainstructions-symbols-to-be-consistent-with-the-others-fix.patch
x86_64-inhibit-machine-from-asserting-an-nmi-when-doing-alt-sysrq-m-operation-tidy.patch
resurrect-the-vmi-lazy-mode-fixes-fix.patch
i386-map-enough-initial-memory-to-create-lowmem-mappings-fix.patch
x86_64-unexport-cpu_llc_id.patch
xfs-clean-up-shrinker-games.patch
add-__gfp_movable-for-callers-to-flag-allocations-from-high-memory-that-may-be-migrated-fix.patch
mm-merge-populate-and-nopage-into-fault-fixes-nonlinear-tidy.patch
mm-merge-nopfn-into-fault-fix.patch
i386-use-pte_update_defer-in-ptep_test_and_clear_dirtyyoung-fix.patch
smaps-add-clear_refs-file-to-clear-reference-fix.patch
smaps-add-clear_refs-file-to-clear-reference-fix-fix-2.patch
maps2-move-the-page-walker-code-to-lib-fix.patch
maps2-add-proc-pid-pagemap-interface-fix-fix.patch
bias-the-location-of-pages-freed-for-min_free_kbytes-in-the-same-max_order_nr_pages-blocks-tidy.patch
mm-move-common-segment-checks-to-separate-helper-function-v7-tidy.patch
slab-mark-set_up_list3s-__init.patch
extend-print_symbol-capability-fix-fix.patch
slub-core-tidy.patch
slub-core-tidy-2.patch
slub-core-tidy-3.patch
slub-core-tidy-4.patch
slub-core-tidy-5.patch
slub-core-tidy-6.patch
slub-core-tidy-7.patch
slub-core-tidy-8.patch
slub-core-tidy-9.patch
slub-core-add-explanation-for-locking-fix.patch
slub-core-explain-sizing-of-slabs-in-detail-fix.patch
slub-core-more-statics.patch
slub-enable-tracking-of-full-slabs-fix.patch
slub-add-ability-to-list-alloc--free-callers-per-slab-tidy.patch
slub-user-documentation-fix.patch
mm-optimize-kill_bdev-fix.patch
lazy-freeing-of-memory-through-madv_free-fix.patch
lazy-freeing-of-memory-through-madv_free-sparc-fix.patch
lazy-freeing-of-memory-through-madv_free-vs-mm-madvise-avoid-exclusive-mmap_sem.patch
driver_bfin_serial_core-update.patch
srmcons-fix-kmallocgfp_kernel-inside-spinlock.patch
uml-driver-formatting-fixes-fix.patch
reduce-size-of-task_struct-on-64-bit-machines.patch
mm-shrink-parent-dentries-when-shrinking-slab.patch
merge-sys_clone-sys_unshare-nsproxy-and-namespace-fix-fix-fix.patch
virtual_eisa_root_init-should-be-__init.patch
proc-maps-protection.patch
fix-cycladesh-for-x86_64-and-probably-others-fix.patch
rtc-add-rtc-rs5c313-driver-tidy.patch
rtc-add-rtc-rs5c313-driver-is-busted.patch
enlarge-console-name.patch
move-die-notifier-handling-to-common-code-fixes-2.patch
move-die-notifier-handling-to-common-code-fix-vmalloc_sync_all.patch
fix-sscanf-%n-match-at-end-of-input-string-tidy.patch
parport-dev-driver-model-support-powerpc-fix.patch
cache-pipe-buf-page-address-for-non-highmem-arch-fix.patch
cache-pipe-buf-page-address-for-non-highmem-arch-fix-tidy.patch
add-support-for-deferrable-timers-respun-tidy.patch
linux-sysdevh-needs-to-include-linux-moduleh.patch
time-smp-friendly-alignment-of-struct-clocksource.patch
move-timekeeping-code-to-timekeepingc-fix.patch
ignore-stolen-time-in-the-softlockup-watchdog-fix.patch
fix-kevents-childs-priority-greediness-fix.patch
display-all-possible-partitions-when-the-root-filesystem-failed-to-mount-fix.patch
enhance-initcall_debug-measure-latency-fix.patch
expose-range-checking-functions-from-arch-specific-update-fix.patch
pad-irq_desc-to-internode-cacheline-size-fix.patch
dtlk-fix-error-checks-in-module_init-fix.patch
document-spin_lock_unlocked-rw_lock_unlocked-deprecation-fix.patch
upper-32-bits.patch
define-and-use-new-eventscpu_lock_acquire-and-cpu_lock_release.patch
call-cpu_chain-with-cpu_down_failed-if-cpu_down_prepare-failed-vs-reduce-size-of-task_struct-on-64-bit-machines.patch
kthread-dont-depend-on-work-queues-take-2-fix.patch
speedup-divides-by-cpu_power-in-scheduler.patch
revert-sched-redundant-reschedule-when-set_user_nice-boosts-a-prio-of-a-task-from-the-expired-array-update.patch
revert-sched-redundant-reschedule-when-set_user_nice-boosts-a-prio-of-a-task-from-the-expired-array.patch
sched-consolidate-sched_clock-drift-adjustments-fix.patch
lutimesat-compat-syscall-and-wire-up-on-x86_64.patch
revert-rtc-add-rtc_merge_alarm.patch
declare-struct-ktime.patch
make-futex_wait-use-an-hrtimer-for-timeout-fix.patch
sys_futex64-allows-64bit-futexes-workaround.patch
kprobes-the-on-off-knob-thru-debugfs-updated-fix.patch
kprobes-the-on-off-knob-thru-debugfs-updated-fix-fix-fix.patch
atomich-add-atomic64-cmpxchg-xchg-and-add_unless-to-powerpc.patch
local_t-powerpc-extension.patch
linux-kernel-markers-i386-optimization-fix.patch
signal-timer-event-fds-v9-signalfd-core-fix.patch
signal-timer-event-fds-v9-signalfd-core-fix-fix.patch
signal-timer-event-fds-v9-timerfd-core-fix.patch
signal-timer-event-fds-v9-eventfd-core-fix.patch
signal-timer-event-fds-v9-eventfd-core-fix-fix.patch
revoke-core-code-fix-shared-mapping-revoke.patch
revoke-wire-up-i386-system-calls-x86_64-fix.patch
x86-serial-convert-legacy-com-ports-to-platform-devices-fix.patch
lguest-the-host-code-vs-sys_futex64-allows-64bit-futexes-get_futex_key-must-check-proper-alignement-for-64bit-futexes.patch
lguest-the-host-code-vs-x86_64-mm-i386-separate-hardware-defined-tss-from-linux-additions.patch
lguest-the-host-code-vs-futex-new-private-futexes.patch
fs-convert-core-functions-to-zero_user_page-pass-kmap-type.patch
fs-convert-core-functions-to-zero_user_page-fix-2.patch
ntfs-use-zero_user_page-fix.patch
unprivileged-mounts-account-user-mounts-fix.patch
unprivileged-mounts-propagate-error-values-from-clone_mnt-fix.patch
unprivileged-mounts-allow-unprivileged-bind-mounts-fix.patch
reiser4-slab-allocators-remove-slab_debug_initial-flag.patch
fbdev-hecuba-framebuffer-driver.patch
vt-add-color-support-to-the-underline-and-italic-attributes-fix.patch
sm501fb-printk-warning-fixes.patch
integrity-new-hooks-fix.patch
integrity-evm-as-an-integrity-service-provider-tidy.patch
integrity-evm-as-an-integrity-service-provider-tidy-fix.patch
integrity-evm-as-an-integrity-service-provider-tidy-fix-2.patch
integrity-ima-integrity_measure-support-tidy.patch
integrity-ima-integrity_measure-support-fix.patch
integrity-ima-integrity_measure-support-fix-2.patch
integrity-tpm-internal-kernel-interface-tidy.patch

Re: + ignore-stolen-time-in-the-softlockup-watchdog-fix.patch added to -mm tree

[Andrew Morton]

On Wed, 25 Apr 2007 11:48:45 +1000 Paul Mackerras <paulus@samba.org> wrote:

> akpm@linux-foundation.org writes:
> 
> > Avoid reevaluating smp_processor_id()
> 
> >  	int this_cpu = smp_processor_id();
> > -	unsigned long touch_timestamp = __get_cpu_var(touch_timestamp);
> > +	unsigned long touch_timestamp = per_cpu(touch_timestamp, this_cpu);
> 
> This is actually worse on those cpus that have a fast way to get to
> the current cpu's per-cpu data.
> 

The patch adds 4 bytes (strange number) to powerpc text and removes 32 from
i386's.  A net win, I guess.

i386 now has a considerably better smp_processor_id(), but no special-case
handling of per-cpu data.  Maybe Jeremy has plans to move i386's per-cpu
data into the pda also?

Re: + ignore-stolen-time-in-the-softlockup-watchdog-fix.patch added to -mm tree

[Andrew Morton]

On Tue, 24 Apr 2007 19:09:13 -0700 Jeremy Fitzhardinge <jeremy@goop.org> wrote:

> akpm@linux-foundation.org wrote:
> > The patch titled
> >      ignore-stolen-time-in-the-softlockup-watchdog-fix
> > has been added to the -mm tree.  Its filename is
> >      ignore-stolen-time-in-the-softlockup-watchdog-fix.patch
> >
> > *** Remember to use Documentation/SubmitChecklist when testing your code ***
> >
> > See http://www.zip.com.au/~akpm/linux/patches/stuff/added-to-mm.txt to find
> > out what to do about this
> >
> > ------------------------------------------------------
> > Subject: ignore-stolen-time-in-the-softlockup-watchdog-fix
> > From: Andrew Morton <akpm@linux-foundation.org>
> >
> > Avoid reevaluating smp_processor_id()
> >   
> 
> I don't think this is a useful patch.  Are there any architectures which
> have an expensive smp_processor_id() and use it in per_cpu?

Any which enable CONFIG_DEBUG_PREEMPT...

>  On x86
> smp_processor_id is cheap, and once the pda->percpu patches go in, won't
> be used for get_cpu_var anyway.
> 
>     J

Re: + ignore-stolen-time-in-the-softlockup-watchdog-fix.patch added to -mm tree

[Andrew Morton]

On Tue, 24 Apr 2007 21:37:47 -0700 Jeremy Fitzhardinge <jeremy@goop.org> wrote:

> Andrew Morton wrote:
> > Any which enable CONFIG_DEBUG_PREEMPT...
> >   
> 
> Hm.  Normally smp_processor_id() CSEs nicely, so it isn't generally
> necessary to do it manually.

For all architectures?

>  Not sure what CONFIG_DEBUG_PREEMPT will do
> to it.

#ifdef CONFIG_DEBUG_PREEMPT
  extern unsigned int debug_smp_processor_id(void);
# define smp_processor_id() debug_smp_processor_id()
#else
# define smp_processor_id() raw_smp_processor_id()
#endif

We could perhaps mark debug_smp_processor_id() as pure or const or whatever
the trick is to say it doesn't need to be reevaluated.  That would usually
be OK, because as long as preempt is disabled that _is_ how it works,
however code might do:

	preempt_disable();
	smp_processor_id();
	preempt_enable();

	<preempt or schedule()>

	preempt_disable();
	smp_processor_id();
	preempt_enable();

and caching the value would be wrong.

hm.  If "smp_processor_id() CSEs nicely", what prevents the compiler from
making the same mistake?  IOW, what are the barriers for CSE?  Function
calls?