Part of long forgotten patch
http://groups.google.com/group/fa.linux.kernel/msg/e98e941ce1cf29f6?dmode=source
Since then, m32r grabbed two copies.
Leave s390 copy because of important absence of CONFIG_VT, but remove
references to non-existent timerlist_lock. ia64 also loses timerlist_lock.
Signed-off-by: Alexey Dobriyan <adobriyan@openvz.org>
Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Andi Kleen <ak@muc.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Hirokazu Takata <takata@linux-m32r.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fix-rmmod-read-write-races-in-proc-entries.patch doesn't want dynamically
allocated ->proc_fops, because it will set it to NULL at module unload time.
Regardless of module status, switch to statically allocated ->proc_fops which
leads to simpler code without wrappers.
AFAICS, also fix the following bug: "sn_force_interrupt" proc entry set
->write for itself, but was created with 0444 permissions. Change to 0644.
Signed-off-by: Alexey Dobriyan <adobriyan@openvz.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
I noticed that almost all architectures implemented exactly the same
sys32_sysinfo... except parisc, where a bug was to be found in handling of
the uptime. So let's remove a whole whack of code for fun and profit.
Cribbed compat_sys_sysinfo from x86_64's implementation, since I figured it
would be the best tested.
This patch incorporates Arnd's suggestion of not using set_fs/get_fs, but
instead extracting out the common code from sys_sysinfo.
Cc: Christoph Hellwig <hch@infradead.org>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Replace appropriate pairs of "kmem_cache_alloc()" + "memset(0)" with the
corresponding "kmem_cache_zalloc()" call.
Signed-off-by: Robert P. J. Day <rpjday@mindspring.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Roland McGrath <roland@redhat.com>
Cc: James Bottomley <James.Bottomley@steeleye.com>
Cc: Greg KH <greg@kroah.com>
Acked-by: Joel Becker <Joel.Becker@oracle.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Jan Kara <jack@ucw.cz>
Cc: Michael Halcrow <mhalcrow@us.ibm.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Stephen Smalley <sds@tycho.nsa.gov>
Cc: James Morris <jmorris@namei.org>
Cc: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Update all arch/*/kernel/vmlinux.lds.S to not include space for initramfs
when CONFIG_BLK_DEV_INITRAMFS is not selected. This saves another 4 kbytes
on most platfoms (some reserve PAGE_SIZE for initramfs).
Signed-off-by: Jean-Paul Saman <jean-paul.saman@nxp.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
ZONE_DMA less operation for IA64 SGI platform
Disable ZONE_DMA for SGI SN2. All memory is addressable by all devices and we
do not need any special memory pool.
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch simply defines CONFIG_ZONE_DMA for all arches. We later do special
things with CONFIG_ZONE_DMA after the VM and an arch are prepared to work
without ZONE_DMA.
CONFIG_ZONE_DMA can be defined in two ways depending on how an architecture
handles ISA DMA.
First if CONFIG_GENERIC_ISA_DMA is set by the arch then we know that the arch
needs ZONE_DMA because ISA DMA devices are supported. We can catch this in
mm/Kconfig and do not need to modify arch code.
Second, arches may use ZONE_DMA in an unknown way. We set CONFIG_ZONE_DMA for
all arches that do not set CONFIG_GENERIC_ISA_DMA in order to insure backwards
compatibility. The arches may later undefine ZONE_DMA if their arch code has
been verified to not depend on ZONE_DMA.
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Cc: Andi Kleen <ak@suse.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Matthew Wilcox <willy@debian.org>
Cc: James Bottomley <James.Bottomley@steeleye.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Function is unnecessary now. We can use the summing features of the ZVCs to
get the values we need.
Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The arch hooks arch_setup_msi_irq and arch_teardown_msi_irq are now
responsible for allocating and freeing the linux irq in addition to
setting up the the linux irq to work with the interrupt.
arch_setup_msi_irq now takes a pci_device and a msi_desc and returns
an irq.
With this change in place this code should be useable by all platforms
except those that won't let the OS touch the hardware like ppc RTAS.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
We need to be able to get from an irq number to a struct msi_desc.
The msi_desc array in msi.c had several short comings the big one was
that it could not be used outside of msi.c. Using irq_data in struct
irq_desc almost worked except on some architectures irq_data needs to
be used for something else.
So this patch adds a msi_desc pointer to irq_desc, adds the appropriate
wrappers and changes all of the msi code to use them.
The dynamic_irq_init/cleanup code was tweaked to ensure the new
field is left in a well defined state.
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (140 commits)
ACPICA: reduce table header messages to fit within 80 columns
asus-laptop: merge with ACPICA table update
ACPI: bay: Convert ACPI Bay driver to be compatible with sysfs update.
ACPI: bay: new driver is EXPERIMENTAL
ACPI: bay: make drive_bays static
ACPI: bay: make bay a platform driver
ACPI: bay: remove prototype procfs code
ACPI: bay: delete unused variable
ACPI: bay: new driver adding removable drive bay support
ACPI: dock: check if parent is on dock
ACPICA: fix gcc build warnings
Altix: Add ACPI SSDT PCI device support (hotplug)
Altix: ACPI SSDT PCI device support
ACPICA: reduce conflicts with Altix patch series
ACPI_NUMA: fix HP IA64 simulator issue with extended memory domain
ACPI: fix HP RX2600 IA64 boot
ACPI: build fix for IBM x440 - CONFIG_X86_SUMMIT
ACPICA: Update version to 20070126
ACPICA: Fix for incorrect parameter passed to AcpiTbDeleteTable during table load.
ACPICA: Update copyright to 2007.
...
This patch fixes
- marking I-cache clean of pages DMAed to now only done for IA64
- broken multiple inclusion in include/asm-x86_64/swiotlb.h
- missing call to mark_clean in swiotlb_sync_sg()
- a (perhaps only theoretical) issue in swiotlb_dma_supported() when
io_tlb_end is exactly at the end of memory
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
getcpu system call returns cpu# and node# on which this system call and
its caller are running. This patch hooks up its implementation on IA64.
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Eliminate arch specific memory_present call ia64 NUMA by utilizing
sparse_memory_present_with_active_regions.
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
On the ia64 architecture only this patch upgrades show_mem() for sparse
memory to be the same as it was for discontig memory. It has been shown to
work on NUMA and flatmem architectures.
Signed-off-by: George Beshers <gbeshers@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Add missing exports to allow several drivers to be built as module with
CONFIG_IA64_HP_ZX1_SWIOTLB.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Occasionally the FSYS_RETURN patch list can have an odd length, causing other
data structures to get out of alignment. In OpenVZ it is odd and we get
misaligned kernel image, which does not boot.
Signed-off-by: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Signed-off-by: Kirill Korotaev <dev@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
While pursuing and unrelated issue with 64Mb granules I noticed a problem
related to inconsistent use of add_active_range. There doesn't appear any
reason to me why FLATMEM versus DISCONTIG_MEM should register memory to
add_active_range with different code. So I've changed the code into a
common implementation.
The other subtle issue fixed by this patch was calling add_active_range in
count_node_pages before granule aligning is performed. We were lucky with
16MB granules but not so with 64MB granules. count_node_pages has reserved
regions filtered out and as a consequence linked kernel text and data
aren't covered by calls to count_node_pages. So linked kernel regions
wasn't reported to add_active_regions. This resulted in free_initmem
causing numerous bad_page reports. This won't occur with this patch
because now all known memory regions are reported by
register_active_ranges.
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Bob Picco <bob.picco@hp.com>
Acked-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Don't force CONFIG_SWIOTLB on when not actually needed (i.e. HP_ZX1 and
SGI_SN2).
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
When we offline a CPU, migrate_irqs() tries to determine whether the
affinity bits of the IRQ descriptor match any of the remaining online
CPUs. If not, it fixes up the interrupt to point somewhere else.
Unfortunately, if an IRQ is unregistered the IRQ descriptor may still
have affinity to the CPU being offlined, but the no_irq_chip handler
doesn't provide a set_affinity function. This causes us to hit the
WARN_ON in migrate_irqs().
The easiest solution seems to be setting all the bits in the affinity
mask when the last interrupt is removed from the vector. I hit this on
an older kernel with Xen/ia64 using driver domains (so it probably needs
more testing on upstream). Xen essentially uses the bind/unbind
interface in sysfs to unregister a device from a driver and thus
unregister the interrupt.
Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
All IA64 systems except IA64_HP_SIM include ACPI and PCI.
So prevent IA64 Kconfigs that try to do irritating things like building
PCI without building ACPI.
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
This patch fixes a NULL-pointer dereference in ia64_machine_kexec().
The variable ia64_kimage is set in machine_kexec_prepare() which is
called from sys_kexec_load(). If kdump wasn't configured before,
ia64_kimage is NULL. machine_kdump_on_init() passes ia64_kimage() to
machine_kexec() which assumes a valid value.
The patch also adds a few sanity checks for the image to simplify
debugging of similar problems in future.
Signed-off-by: Bernhard Walle <bwalle@suse.de>
Signed-off-by: Tony Luck <tony.luck@intel.com>
I encountered one problem when running ptrace test case the situation
is this: traced process's syscall parameter needs to be accessed, but
for sys_clone system call with clone_flag (CLONE_VFORK | CLONE_VM |
SIGCHLD) parameter. This syscall's parameter accessing result is wrong.
The reason is that vforked child process mm point is the same, but
tgid is different. Without this patch find_thread_for_addr will return
vforked process if vforked process is also stopped, but not the thread
which calls vfork syscall.
Signed-off-by: Tony Luck <tony.luck@intel.com>
Some patches have turned up on xen-devel recently to convert strcpy()
to safer alternatives and so forth. While reviewing those patches
I noticed that the features string building could be cleaned up.
This patch uses snprintf() instead of strcpy() and direct character
pointer manipulation. It makes the features string building safe and
gets rid of the special case for features output in show_cpuinfo()
Additionally I removed the (int) cast of ARRAY_SIZE, which seems to
serve no purpose.
Signed-off-by: Aron Griffis <aron@hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
As is pointed out in
http://www.gelato.org/community/view_linear.php?id=1_1036&from=authors&value=Ian%20Wienand#1_1039,
if single step on break instruction, the break fault has higher
priority than the single-step trap. When the break fault handler
is entered, it advances the IP by 1 instruction so break instruction
single-stepping is skipped, actually it is next instruction which
is single stepped.
This patch modifies this, it adds TIF_SINGLESTEP bit for thread
flags, and generate a fake sigtrap when single stepping break
instruction. Test case in attachment can verify this. Any comments
is welcome.
Signed-off-by: bibo, mao <bibo.mao@intel.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
This moves the ia64 implementation of machine_shutdown() from
machine_kexec.c to process.c, which is in keeping with the implelmentation
on other architectures, and seems like a much more appropriate home for it.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Remove the Remove inline declaration of efi_get_pal_addr() as it is
declared in linux/efi.h.
Signed-Off-By: Simon Horman <horms@verge.net.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
linux/uaccess.h was being included, but it seems that
really the following includes are needed.
asm/page.h: for __va() and PAGE_SHIFT
asm/uaccess.h: for copy_to_user()
I guess that linux/uaccess.h pulls in both asm/page.h and asm/uaccess.h.
I notices this while backporting the code to xen's linux-2.6.16.33,
which does not have linux/uaccess.h. I'm posting it as I think it is a
correct, though somewhat cosmetic fix.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Fix a typo in the saved_max_pfn description in contig.c
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Set saved_max_pfn when discontig memory is in use.
This sets up saved_max_pfn when disctontig memory is in use.
This mirrors the code for contig memory.
This patch does not entirely solve the problem of making vmcore work,
however it does appear to be neccessary. Please consider applying.
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Kexec support for 2.6.20 on ia64 does not build properly using a config
made up by CONFIG_SMP=n and CONFIG_HOTPLUG_CPU=n:
Signed-off-by: Magnus Damm <magnus@valinux.co.jp>
Acked-by: Simon Horman <horms@verge.net.au>
Acked-by: Jay Lan <jlan@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
The SN Altix platform does not conform to the IOSAPIC IRQ routing model.
Add code in acpi_unregister_gsi() to check if (acpi_irq_model ==
ACPI_IRQ_MODEL_PLATFORM) and return.
Due to an oversight, this code was not added previously when
similar code was added to acpi_register_gsi().
http://marc.theaimsgroup.com/?l=linux-acpi&m=116680983430121&w=2
Signed-off-by: John Keller <jpk@sgi.com>
Acked-by: Len Brown <lenb@kernel.org>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch fixes up ia64 kexec support for HP rx2620 hardware. It does
this by skipping migration of already disabled irqs. This is most likely a
problem on other ia64 platforms as well, but I've only been able to
reproduce it on one machine so far.
The full story is that handle_bad_irq() gets invoked before starting the
new kernel without this patch. This seems to happen when fixup_irqs()
calls generic_handle_irq() on already migrated (and disabled) irqs. So by
avoiding migration of disabled irqs we stay away of handle_bad_irq().
The code has been tested on three different ia64 machines, all with good
results. It is possible to trigger the same bug by offlining a processor
using echo 0 > /sys/devices/system/cpu/cpuX/online.
More detailed information is available in the following mail thread:
http://lists.osdl.org/pipermail/fastboot/2007-January/thread.html#5774
Signed-off-by: Magnus Damm <magnus@valinux.co.jp>
Acked-by: Simon Horman <horms@verge.net.au>
Acked-by: Zou, Nanhai <nanhai.zou@intel.com>
Acked-by: Jay Lan <jlan@sgi.com>
Acked-by: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add SN platform support for running with an ACPI
capable PROM that defines PCI devices in SSDT
tables. There is a SSDT table for every occupied
slot on a root bus, containing info for every
PPB and/or device on the bus. The SSDTs will be
dynamically loaded/unloaded at hotplug enable/disable.
Platform specific information that is currently
passed via a SAL call, will now be passed via the
Vendor resource in the ACPI Device object(s) defined
in each SSDT.
Signed-off-by: John Keller <jpk@sgi.com>
Cc: Greg KH <greg@kroah.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Len Brown <len.brown@intel.com>
ACPI 3.0 incorporated the SRAT spec, upping the table version to 2,
and extending the size of the proximity domain from 1-byte to 4-bytes.
This extension was into a reserved field that firmware should
set to 0, but the HP simulator had non-zero values there
resulting in unexpected huge numbers.
So mask the domain down to 8-bits for now.
A more general fix will be to check the table version
supplied by firmware and get paranoid about reserved fields.
Signed-off-by: Alexey Starikovskiy <alexey.y.starikovskiy@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
Revert "ACPI: ibm-acpi: make non-generic bay support optional"
ACPI: update MAINTAINERS
ACPI: schedule obsolete features for deletion
ACPI: delete two spurious ACPI messages
ACPI: rename cstate_entry_s to cstate_entry
ACPI: ec: enable printk on cmdline use
ACPI: Altix: ACPI _PRT support
Fix an oops experienced on the Cell architecture when init-time functions,
early_*(), are called at runtime. It alters the call paths to make sure
that the callers explicitly say whether the call is being made on behalf of
a hotplug even, or happening at boot-time.
It has been compile tested on ppc64, ia64, s390, i386 and x86_64.
Acked-by: Arnd Bergmann <arndb@de.ibm.com>
Signed-off-by: Dave Hansen <haveblue@us.ibm.com>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Acked-by: Andy Whitcroft <apw@shadowen.org>
Cc: Christoph Lameter <clameter@engr.sgi.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Provide ACPI _PRT support for SN Altix systems.
The SN Altix platform does not conform to the
IOSAPIC IRQ routing model, so a new acpi_irq_model
(ACPI_IRQ_MODEL_PLATFORM) has been defined. The SN
platform specific code sets acpi_irq_model to
this new value, and keys off of it in acpi_register_gsi()
to avoid the iosapic code path.
Signed-off-by: John Keller <jpk@sgi.com>
Signed-off-by: Len Brown <len.brown@intel.com>
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (68 commits)
ACPI: replace kmalloc+memset with kzalloc
ACPI: Add support for acpi_load_table/acpi_unload_table_id
fbdev: update after backlight argument change
ACPI: video: Add dev argument for backlight_device_register
ACPI: Implement acpi_video_get_next_level()
ACPI: Kconfig - depend on PM rather than selecting it
ACPI: fix NULL check in drivers/acpi/osl.c
ACPI: make drivers/acpi/ec.c:ec_ecdt static
ACPI: prevent processor module from loading on failures
ACPI: fix single linked list manipulation
ACPI: ibm_acpi: allow clean removal
ACPI: fix git automerge failure
ACPI: ibm_acpi: respond to workqueue update
ACPI: dock: add uevent to indicate change in device status
ACPI: ec: Lindent once again
ACPI: ec: Change #define to enums there possible.
ACPI: ec: Style changes.
ACPI: ec: Acquire Global Lock under EC mutex.
ACPI: ec: Drop udelay() from poll mode. Loop by reading status field instead.
ACPI: ec: Rename gpe_bit to gpe
...
Fernando Lopez-Lezcano reported frequent scheduling latencies and audio
xruns starting at the 2.6.18-rt kernel, and those problems persisted all
until current -rt kernels. The latencies were serious and unjustified by
system load, often in the milliseconds range.
After a patient and heroic multi-month effort of Fernando, where he
tested dozens of kernels, tried various configs, boot options,
test-patches of mine and provided latency traces of those incidents, the
following 'smoking gun' trace was captured by him:
_------=> CPU#
/ _-----=> irqs-off
| / _----=> need-resched
|| / _---=> hardirq/softirq
||| / _--=> preempt-depth
|||| /
||||| delay
cmd pid ||||| time | caller
\ / ||||| \ | /
IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup (try_to_wake_up)
IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup <<...>-5856> (37 0)
IRQ_19-1479 1D..1 0us : __trace_start_sched_wakeup (c01262ba 0 0)
IRQ_19-1479 1D..1 0us : resched_task (try_to_wake_up)
IRQ_19-1479 1D..1 0us : __spin_unlock_irqrestore (try_to_wake_up)
...
<idle>-0 1...1 11us!: default_idle (cpu_idle)
...
<idle>-0 0Dn.1 602us : smp_apic_timer_interrupt (c0103baf 1 0)
...
<...>-5856 0D..2 618us : __switch_to (__schedule)
<...>-5856 0D..2 618us : __schedule <<idle>-0> (20 162)
<...>-5856 0D..2 619us : __spin_unlock_irq (__schedule)
<...>-5856 0...1 619us : trace_stop_sched_switched (__schedule)
<...>-5856 0D..1 619us : trace_stop_sched_switched <<...>-5856> (37 0)
what is visible in this trace is that CPU#1 ran try_to_wake_up() for
PID:5856, it placed PID:5856 on CPU#0's runqueue and ran resched_task()
for CPU#0. But it decided to not send an IPI that no CPU - due to
TS_POLLING. But CPU#0 never woke up after its NEED_RESCHED bit was set,
and only rescheduled to PID:5856 upon the next lapic timer IRQ. The
result was a 600+ usecs latency and a missed wakeup!
the bug turned out to be an idle-wakeup bug introduced into the mainline
kernel this summer via an optimization in the x86_64 tree:
commit 495ab9c045
Author: Andi Kleen <ak@suse.de>
Date: Mon Jun 26 13:59:11 2006 +0200
[PATCH] i386/x86-64/ia64: Move polling flag into thread_info_status
During some profiling I noticed that default_idle causes a lot of
memory traffic. I think that is caused by the atomic operations
to clear/set the polling flag in thread_info. There is actually
no reason to make this atomic - only the idle thread does it
to itself, other CPUs only read it. So I moved it into ti->status.
the problem is this type of change:
if (!hlt_counter && boot_cpu_data.hlt_works_ok) {
- clear_thread_flag(TIF_POLLING_NRFLAG);
+ current_thread_info()->status &= ~TS_POLLING;
smp_mb__after_clear_bit();
while (!need_resched()) {
local_irq_disable();
this changes clear_thread_flag() to an explicit clearing of TS_POLLING.
clear_thread_flag() is defined as:
clear_bit(flag, &ti->flags);
and clear_bit() is a LOCK-ed atomic instruction on all x86 platforms:
static inline void clear_bit(int nr, volatile unsigned long * addr)
{
__asm__ __volatile__( LOCK_PREFIX
"btrl %1,%0"
hence smp_mb__after_clear_bit() is defined as a simple compile barrier:
#define smp_mb__after_clear_bit() barrier()
but the explicit TS_POLLING clearing introduced by the patch:
+ current_thread_info()->status &= ~TS_POLLING;
is not an atomic op! So the clearing of the TS_POLLING bit is freely
reorderable with the reading of the NEED_RESCHED bit - and both now
reside in different memory addresses.
CPU idle wakeup very much depends on ordered memory ops, the clearing of
the TS_POLLING flag must always be done before we test need_resched()
and hit the idle instruction(s). [Symmetrically, the wakeup code needs
to set NEED_RESCHED before it tests the TS_POLLING flag, so memory
ordering is paramount.]
Fernando's dual-core Athlon64 system has a sufficiently advanced memory
ordering model so that it triggered this scenario very often.
( And it also turned out that the reason why these latencies never
triggered on my testsystems is that i routinely use idle=poll, which
was the only idle variant not affected by this bug. )
The fix is to change the smp_mb__after_clear_bit() to an smp_mb(), to
act as an absolute barrier between the TS_POLLING write and the
NEED_RESCHED read. This affects almost all idling methods (default,
ACPI, APM), on all 3 x86 architectures: i386, x86_64, ia64.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Tested-by: Fernando Lopez-Lezcano <nando@ccrma.Stanford.EDU>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
