pandora-kernel.git
9 years agoLinux 3.2.26 v3.2.26
Ben Hutchings [Sat, 4 Aug 2012 15:31:19 +0000 (16:31 +0100)]
Linux 3.2.26

9 years agox86: Simplify code by removing a !SMP #ifdefs from 'struct cpuinfo_x86'
Kevin Winchester [Wed, 21 Dec 2011 00:52:22 +0000 (20:52 -0400)]
x86: Simplify code by removing a !SMP #ifdefs from 'struct cpuinfo_x86'

commit 141168c36cdee3ff23d9c7700b0edc47cb65479f and
commit 3f806e50981825fa56a7f1938f24c0680816be45 upstream.

Several fields in struct cpuinfo_x86 were not defined for the
!SMP case, likely to save space.  However, those fields still
have some meaning for UP, and keeping them allows some #ifdef
removal from other files.  The additional size of the UP kernel
from this change is not significant enough to worry about
keeping up the distinction:

   text    data     bss     dec     hex filename
4737168  506459  972040 6215667  5ed7f3 vmlinux.o.before
4737444  506459  972040 6215943  5ed907 vmlinux.o.after

for a difference of 276 bytes for an example UP config.

If someone wants those 276 bytes back badly then it should
be implemented in a cleaner way.

Signed-off-by: Kevin Winchester <kjwinchester@gmail.com>
Cc: Steffen Persvold <sp@numascale.com>
Link: http://lkml.kernel.org/r/1324428742-12498-1-git-send-email-kjwinchester@gmail.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoLinux 3.2.25 v3.2.25
Ben Hutchings [Thu, 2 Aug 2012 13:38:04 +0000 (14:38 +0100)]
Linux 3.2.25

9 years agomm: fix wrong argument of migrate_huge_pages() in soft_offline_huge_page()
Joonsoo Kim [Mon, 30 Jul 2012 21:39:04 +0000 (14:39 -0700)]
mm: fix wrong argument of migrate_huge_pages() in soft_offline_huge_page()

commit dc32f63453f56d07a1073a697dcd843dd3098c09 upstream.

Commit a6bc32b89922 ("mm: compaction: introduce sync-light migration for
use by compaction") changed the declaration of migrate_pages() and
migrate_huge_pages().

But it missed changing the argument of migrate_huge_pages() in
soft_offline_huge_page().  In this case, we should call
migrate_huge_pages() with MIGRATE_SYNC.

Additionally, there is a mismatch between type the of argument and the
function declaration for migrate_pages().

Signed-off-by: Joonsoo Kim <js1304@gmail.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Mel Gorman <mgorman@suse.de>
Acked-by: David Rientjes <rientjes@google.com>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agonouveau: Fix alignment requirements on src and dst addresses
Maarten Lankhorst [Mon, 4 Jun 2012 10:00:31 +0000 (12:00 +0200)]
nouveau: Fix alignment requirements on src and dst addresses

commit ce806a30470bcd846d148bf39d46de3ad7748228 upstream.

Linear copy works by adding the offset to the buffer address,
which may end up not being 16-byte aligned.

Some tests I've written for prime_pcopy show that the engine
allows this correctly, so the restriction on lowest 4 bits of
address can be lifted safely.

The comments added were by envyas, I think because I used
a newer version.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
[bwh: Backported to 3.2: no # prefixes in nva3_copy.fuc]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoBtrfs: call the ordered free operation without any locks held
Chris Mason [Wed, 25 Jul 2012 19:57:13 +0000 (15:57 -0400)]
Btrfs: call the ordered free operation without any locks held

commit e9fbcb42201c862fd6ab45c48ead4f47bb2dea9d upstream.

Each ordered operation has a free callback, and this was called with the
worker spinlock held.  Josef made the free callback also call iput,
which we can't do with the spinlock.

This drops the spinlock for the free operation and grabs it again before
moving through the rest of the list.  We'll circle back around to this
and find a cleaner way that doesn't bounce the lock around so much.

Signed-off-by: Chris Mason <chris.mason@fusionio.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agodrm/radeon: on hotplug force link training to happen (v2)
Jerome Glisse [Thu, 19 Jul 2012 21:25:55 +0000 (17:25 -0400)]
drm/radeon: on hotplug force link training to happen (v2)

commit ca2ccde5e2f24a792caa4cca919fc5c6f65d1887 upstream.

To have DP behave like VGA/DVI we need to retrain the link
on hotplug. For this to happen we need to force link
training to happen by setting connector dpms to off
before asking it turning it on again.

v2: agd5f
- drop the dp_get_link_status() change in atombios_dp.c
  for now.  We still need the dpms OFF change.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agodrm/radeon: fix hotplug of DP to DVI|HDMI passive adapters (v2)
Jerome Glisse [Thu, 19 Jul 2012 21:15:56 +0000 (17:15 -0400)]
drm/radeon: fix hotplug of DP to DVI|HDMI passive adapters (v2)

commit 266dcba541a1ef7e5d82d9e67c67fde2910636e8 upstream.

No need to retrain the link for passive adapters.

v2: agd5f
- no passive DP to VGA adapters, update comments
- assign radeon_connector_atom_dig after we are sure
  we have a digital connector as analog connectors
  have different private data.
- get new sink type before checking for retrain.  No
  need to check if it's no longer a DP connection.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agodrm/radeon: fix non revealent error message
Jerome Glisse [Tue, 17 Jul 2012 21:17:16 +0000 (17:17 -0400)]
drm/radeon: fix non revealent error message

commit 8d1c702aa0b2c4b22b0742b72a1149d91690674b upstream.

We want to print link status query failed only if it's
an unexepected fail. If we query to see if we need
link training it might be because there is nothing
connected and thus link status query have the right
to fail in that case.

To avoid printing failure when it's expected, move the
failure message to proper place.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agodrm/radeon: fix bo creation retry path
Jerome Glisse [Thu, 12 Jul 2012 22:23:05 +0000 (18:23 -0400)]
drm/radeon: fix bo creation retry path

commit d1c7871ddb1f588b8eb35affd9ee1a3d5e11cd0c upstream.

Retry label was at wrong place in function leading to memory
leak.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoACPI/AC: prevent OOPS on some boxes due to missing check power_supply_register()...
Lan Tianyu [Fri, 20 Jul 2012 05:29:16 +0000 (13:29 +0800)]
ACPI/AC: prevent OOPS on some boxes due to missing check power_supply_register() return value check

commit f197ac13f6eeb351b31250b9ab7d0da17434ea36 upstream.

In the ac.c, power_supply_register()'s return value is not checked.

As a result, the driver's add() ops may return success
even though the device failed to initialize.

For example, some BIOS may describe two ACADs in the same DSDT.
The second ACAD device will fail to register,
but ACPI driver's add() ops returns sucessfully.
The ACPI device will receive ACPI notification and cause OOPS.

https://bugzilla.redhat.com/show_bug.cgi?id=772730

Signed-off-by: Lan Tianyu <tianyu.lan@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agolocks: fix checking of fcntl_setlease argument
J. Bruce Fields [Mon, 23 Jul 2012 19:17:17 +0000 (15:17 -0400)]
locks: fix checking of fcntl_setlease argument

commit 0ec4f431eb56d633da3a55da67d5c4b88886ccc7 upstream.

The only checks of the long argument passed to fcntl(fd,F_SETLEASE,.)
are done after converting the long to an int.  Thus some illegal values
may be let through and cause problems in later code.

[ They actually *don't* cause problems in mainline, as of Dave Jones's
  commit 8d657eb3b438 "Remove easily user-triggerable BUG from
  generic_setlease", but we should fix this anyway.  And this patch will
  be necessary to fix real bugs on earlier kernels. ]

Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoASoC: dapm: Fix _PRE and _POST events for DAPM performance improvements
Mark Brown [Fri, 20 Jul 2012 16:29:34 +0000 (17:29 +0100)]
ASoC: dapm: Fix _PRE and _POST events for DAPM performance improvements

commit 0ff97ebf0804d2e519d578fcb4db03f104d2ca8c upstream.

Ever since the DAPM performance improvements we've been marking all widgets
as not dirty after each DAPM run. Since _PRE and _POST events aren't part
of the DAPM graph this has rendered them non-functional, they will never be
marked dirty again and thus will never be run again.

Fix this by skipping them when marking widgets as not dirty.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Liam Girdwood <lrg@ti.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoext4: undo ext4_calc_metadata_amount if we fail to claim space
Theodore Ts'o [Mon, 23 Jul 2012 04:00:20 +0000 (00:00 -0400)]
ext4: undo ext4_calc_metadata_amount if we fail to claim space

commit 03179fe92318e7934c180d96f12eff2cb36ef7b6 upstream.

The function ext4_calc_metadata_amount() has side effects, although
it's not obvious from its function name.  So if we fail to claim
space, regardless of whether we retry to claim the space again, or
return an error, we need to undo these side effects.

Otherwise we can end up incorrectly calculating the number of metadata
blocks needed for the operation, which was responsible for an xfstests
failure for test #271 when using an ext2 file system with delalloc
enabled.

Reported-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoext4: don't let i_reserved_meta_blocks go negative
Brian Foster [Mon, 23 Jul 2012 03:59:40 +0000 (23:59 -0400)]
ext4: don't let i_reserved_meta_blocks go negative

commit 97795d2a5b8d3c8dc4365d4bd3404191840453ba upstream.

If we hit a condition where we have allocated metadata blocks that
were not appropriately reserved, we risk underflow of
ei->i_reserved_meta_blocks.  In turn, this can throw
sbi->s_dirtyclusters_counter significantly out of whack and undermine
the nondelalloc fallback logic in ext4_nonda_switch().  Warn if this
occurs and set i_allocated_meta_blocks to avoid this problem.

This condition is reproduced by xfstests 270 against ext2 with
delalloc enabled:

Mar 28 08:58:02 localhost kernel: [  171.526344] EXT4-fs (loop1): delayed block allocation failed for inode 14 at logical offset 64486 with max blocks 64 with error -28
Mar 28 08:58:02 localhost kernel: [  171.526346] EXT4-fs (loop1): This should not happen!! Data will be lost

270 ultimately fails with an inconsistent filesystem and requires an
fsck to repair.  The cause of the error is an underflow in
ext4_da_update_reserve_space() due to an unreserved meta block
allocation.

Signed-off-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agommc: sdhci-pci: CaFe has broken card detection
Daniel Drake [Tue, 3 Jul 2012 22:13:39 +0000 (23:13 +0100)]
mmc: sdhci-pci: CaFe has broken card detection

commit 55fc05b7414274f17795cd0e8a3b1546f3649d5e upstream.

At http://dev.laptop.org/ticket/11980 we have determined that the
Marvell CaFe SDHCI controller reports bad card presence during
resume. It reports that no card is present even when it is.
This is a regression -- resume worked back around 2.6.37.

Around 400ms after resuming, a "card inserted" interrupt is
generated, at which point it starts reporting presence.

Work around this hardware oddity by setting the
SDHCI_QUIRK_BROKEN_CARD_DETECTION flag.
Thanks to Chris Ball for helping with diagnosis.

Signed-off-by: Daniel Drake <dsd@laptop.org>
[stable@: please apply to 3.0+]
Signed-off-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoiscsi-target: Drop bogus struct file usage for iSCSI/SCTP
Al Viro [Sat, 21 Jul 2012 07:55:18 +0000 (08:55 +0100)]
iscsi-target: Drop bogus struct file usage for iSCSI/SCTP

commit bf6932f44a7b3fa7e2246a8b18a44670e5eab6c2 upstream.

From Al Viro:

BTW, speaking of struct file treatment related to sockets -
        there's this piece of code in iscsi:
        /*
         * The SCTP stack needs struct socket->file.
         */
        if ((np->np_network_transport == ISCSI_SCTP_TCP) ||
            (np->np_network_transport == ISCSI_SCTP_UDP)) {
                if (!new_sock->file) {
                        new_sock->file = kzalloc(
                                        sizeof(struct file), GFP_KERNEL);

For one thing, as far as I can see it'not true - sctp does *not* depend on
socket->file being non-NULL; it does, in one place, check socket->file->f_flags
for O_NONBLOCK, but there it treats NULL socket->file as "flag not set".
Which is the case here anyway - the fake struct file created in
__iscsi_target_login_thread() (and in iscsi_target_setup_login_socket(), with
the same excuse) do *not* get that flag set.

Moreover, it's a bloody serious violation of a bunch of asserts in VFS;
all struct file instances should come from filp_cachep, via get_empty_filp()
(or alloc_file(), which is a wrapper for it).  FWIW, I'm very tempted to
do this and be done with the entire mess:

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Andy Grover <agrover@redhat.com>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agolibsas: fix sas_discover_devices return code handling
Dan Williams [Fri, 22 Jun 2012 06:36:20 +0000 (23:36 -0700)]
libsas: fix sas_discover_devices return code handling

commit b17caa174a7e1fd2e17b26e210d4ee91c4c28b37 upstream.

commit 198439e4 [SCSI] libsas: do not set res = 0 in sas_ex_discover_dev()
commit 19252de6 [SCSI] libsas: fix wide port hotplug issues

The above commits seem to have confused the return value of
sas_ex_discover_dev which is non-zero on failure and
sas_ex_join_wide_port which just indicates short circuiting discovery on
already established ports.  The result is random discovery failures
depending on configuration.

Calls to sas_ex_join_wide_port are the source of the trouble as its
return value is errantly assigned to 'res'.  Convert it to bool and stop
returning its result up the stack.

Tested-by: Dan Melnic <dan.melnic@amd.com>
Reported-by: Dan Melnic <dan.melnic@amd.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Jack Wang <jack_wang@usish.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agolibsas: continue revalidation
Dan Williams [Fri, 22 Jun 2012 06:36:15 +0000 (23:36 -0700)]
libsas: continue revalidation

commit 26f2f199ff150d8876b2641c41e60d1c92d2fb81 upstream.

Continue running revalidation until no more broadcast devices are
discovered.  Fixes cases where re-discovery completes too early in a
domain with multiple expanders with pending re-discovery events.
Servicing BCNs can get backed up behind error recovery.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agofix eh wakeup (scsi_schedule_eh vs scsi_restart_operations)
Dan Williams [Fri, 22 Jun 2012 06:25:32 +0000 (23:25 -0700)]
fix eh wakeup (scsi_schedule_eh vs scsi_restart_operations)

commit 57fc2e335fd3c2f898ee73570dc81426c28dc7b4 upstream.

Rapid ata hotplug on a libsas controller results in cases where libsas
is waiting indefinitely on eh to perform an ata probe.

A race exists between scsi_schedule_eh() and scsi_restart_operations()
in the case when scsi_restart_operations() issues i/o to other devices
in the sas domain.  When this happens the host state transitions from
SHOST_RECOVERY (set by scsi_schedule_eh) back to SHOST_RUNNING and
->host_busy is non-zero so we put the eh thread to sleep even though
->host_eh_scheduled is active.

Before putting the error handler to sleep we need to check if the
host_state needs to return to SHOST_RECOVERY for another trip through
eh.  Since i/o that is released by scsi_restart_operations has been
blocked for at least one eh cycle, this implementation allows those
i/o's to run before another eh cycle starts to discourage hung task
timeouts.

Reported-by: Tom Jackson <thomas.p.jackson@intel.com>
Tested-by: Tom Jackson <thomas.p.jackson@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agofix hot unplug vs async scan race
Dan Williams [Fri, 22 Jun 2012 06:47:28 +0000 (23:47 -0700)]
fix hot unplug vs async scan race

commit 3b661a92e869ebe2358de8f4b3230ad84f7fce51 upstream.

The following crash results from cases where the end_device has been
removed before scsi_sysfs_add_sdev has had a chance to run.

 BUG: unable to handle kernel NULL pointer dereference at 0000000000000098
 IP: [<ffffffff8115e100>] sysfs_create_dir+0x32/0xb6
 ...
 Call Trace:
  [<ffffffff8125e4a8>] kobject_add_internal+0x120/0x1e3
  [<ffffffff81075149>] ? trace_hardirqs_on+0xd/0xf
  [<ffffffff8125e641>] kobject_add_varg+0x41/0x50
  [<ffffffff8125e70b>] kobject_add+0x64/0x66
  [<ffffffff8131122b>] device_add+0x12d/0x63a
  [<ffffffff814b65ea>] ? _raw_spin_unlock_irqrestore+0x47/0x56
  [<ffffffff8107de15>] ? module_refcount+0x89/0xa0
  [<ffffffff8132f348>] scsi_sysfs_add_sdev+0x4e/0x28a
  [<ffffffff8132dcbb>] do_scan_async+0x9c/0x145

...teach scsi_sysfs_add_devices() to check for deleted devices() before
trying to add them, and teach scsi_remove_target() how to remove targets
that have not been added via device_add().

Reported-by: Dariusz Majchrzak <dariusz.majchrzak@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoAvoid dangling pointer in scsi_requeue_command()
Bart Van Assche [Fri, 29 Jun 2012 15:34:26 +0000 (15:34 +0000)]
Avoid dangling pointer in scsi_requeue_command()

commit 940f5d47e2f2e1fa00443921a0abf4822335b54d upstream.

When we call scsi_unprep_request() the command associated with the request
gets destroyed and therefore drops its reference on the device.  If this was
the only reference, the device may get released and we end up with a NULL
pointer deref when we call blk_requeue_request.

Reported-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Reviewed-by: Tejun Heo <tj@kernel.org>
[jejb: enhance commend and add commit log for stable]
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoFix device removal NULL pointer dereference
Bart Van Assche [Fri, 29 Jun 2012 15:33:22 +0000 (15:33 +0000)]
Fix device removal NULL pointer dereference

commit 67bd94130015c507011af37858989b199c52e1de upstream.

Use blk_queue_dead() to test whether the queue is dead instead
of !sdev. Since scsi_prep_fn() may be invoked concurrently with
__scsi_remove_device(), keep the queuedata (sdev) pointer in
__scsi_remove_device(). This patch fixes a kernel oops that
can be triggered by USB device removal. See also
http://www.spinics.net/lists/linux-scsi/msg56254.html.

Other changes included in this patch:
- Swap the blk_cleanup_queue() and kfree() calls in
  scsi_host_dev_release() to make that code easier to grasp.
- Remove the queue dead check from scsi_run_queue() since the
  queue state can change anyway at any point in that function
  where the queue lock is not held.
- Remove the queue dead check from the start of scsi_request_fn()
  since it is redundant with the scsi_device_online() check.

Reported-by: Jun'ichi Nomura <j-nomura@ce.jp.nec.com>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Reviewed-by: Tejun Heo <tj@kernel.org>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoblock: add blk_queue_dead()
Tejun Heo [Tue, 13 Dec 2011 23:33:37 +0000 (00:33 +0100)]
block: add blk_queue_dead()

commit 34f6055c80285e4efb3f602a9119db75239744dc upstream.

There are a number of QUEUE_FLAG_DEAD tests.  Add blk_queue_dead()
macro and use it.

This patch doesn't introduce any functional difference.

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoALSA: hda - Turn on PIN_OUT from hdmi playback prepare.
Dylan Reid [Fri, 20 Jul 2012 00:52:58 +0000 (17:52 -0700)]
ALSA: hda - Turn on PIN_OUT from hdmi playback prepare.

commit 9e76e6d031482194a5b24d8e9ab88063fbd6b4b5 upstream.

Turn on the pin widget's PIN_OUT bit from playback prepare. The pin is
enabled in open, but is disabled in hdmi_init_pin which is called during
system resume.  This causes a system suspend/resume during playback to
mute HDMI/DP. Enabling the pin in prepare instead of open allows calling
snd_pcm_prepare after a system resume to restore audio.

Signed-off-by: Dylan Reid <dgreid@chromium.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agodrm/radeon: Try harder to avoid HW cursor ending on a multiple of 128 columns.
Michel Dänzer [Tue, 17 Jul 2012 17:02:09 +0000 (19:02 +0200)]
drm/radeon: Try harder to avoid HW cursor ending on a multiple of 128 columns.

commit f60ec4c7df043df81e62891ac45383d012afe0da upstream.

This could previously fail if either of the enabled displays was using a
horizontal resolution that is a multiple of 128, and only the leftmost column
of the cursor was (supposed to be) visible at the right edge of that display.

The solution is to move the cursor one pixel to the left in that case.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=33183

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoiommu/amd: Fix hotplug with iommu=pt
Joerg Roedel [Thu, 19 Jul 2012 11:42:54 +0000 (13:42 +0200)]
iommu/amd: Fix hotplug with iommu=pt

commit 2c9195e990297068d0f1f1bd8e2f1d09538009da upstream.

This did not work because devices are not put into the
pt_domain. Fix this.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
[bwh: Backported to 3.2: do not use iommu_dev_data::passthrough]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoALSA: hda - Add support for Realtek ALC282
David Henningsson [Wed, 18 Jul 2012 05:38:46 +0000 (07:38 +0200)]
ALSA: hda - Add support for Realtek ALC282

commit 4e01ec636e64707d202a1ca21a47bbc6d53085b7 upstream.

This codec has a separate dmic path (separate dmic only ADC),
and thus it looks mostly like ALC275.

BugLink: https://bugs.launchpad.net/bugs/1025377
Tested-by: Ray Chen <ray.chen@canonical.com>
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoworkqueue: perform cpu down operations from low priority cpu_notifier()
Tejun Heo [Tue, 17 Jul 2012 19:39:26 +0000 (12:39 -0700)]
workqueue: perform cpu down operations from low priority cpu_notifier()

commit 6575820221f7a4dd6eadecf7bf83cdd154335eda upstream.

Currently, all workqueue cpu hotplug operations run off
CPU_PRI_WORKQUEUE which is higher than normal notifiers.  This is to
ensure that workqueue is up and running while bringing up a CPU before
other notifiers try to use workqueue on the CPU.

Per-cpu workqueues are supposed to remain working and bound to the CPU
for normal CPU_DOWN_PREPARE notifiers.  This holds mostly true even
with workqueue offlining running with higher priority because
workqueue CPU_DOWN_PREPARE only creates a bound trustee thread which
runs the per-cpu workqueue without concurrency management without
explicitly detaching the existing workers.

However, if the trustee needs to create new workers, it creates
unbound workers which may wander off to other CPUs while
CPU_DOWN_PREPARE notifiers are in progress.  Furthermore, if the CPU
down is cancelled, the per-CPU workqueue may end up with workers which
aren't bound to the CPU.

While reliably reproducible with a convoluted artificial test-case
involving scheduling and flushing CPU burning work items from CPU down
notifiers, this isn't very likely to happen in the wild, and, even
when it happens, the effects are likely to be hidden by the following
successful CPU down.

Fix it by using different priorities for up and down notifiers - high
priority for up operations and low priority for down operations.

Workqueue cpu hotplug operations will soon go through further cleanup.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agortlwifi: rtl8192de: Fix phy-based version calculation
Forest Bond [Fri, 13 Jul 2012 16:26:06 +0000 (12:26 -0400)]
rtlwifi: rtl8192de: Fix phy-based version calculation

commit f1b00f4dab29b57bdf1bc03ef12020b280fd2a72 upstream.

Commit d83579e2a50ac68389e6b4c58b845c702cf37516 incorporated some
changes from the vendor driver that made it newly important that the
calculated hardware version correctly include the CHIP_92D bit, as all
of the IS_92D_* macros were changed to depend on it.  However, this bit
was being unset for dual-mac, dual-phy devices.  The vendor driver
behavior was modified to not do this, but unfortunately this change was
not picked up along with the others.  This caused scanning in the 2.4GHz
band to be broken, and possibly other bugs as well.

This patch brings the version calculation logic in parity with the
vendor driver in this regard, and in doing so fixes the regression.
However, the version calculation code in general continues to be largely
incoherent and messy, and needs to be cleaned up.

Signed-off-by: Forest Bond <forest.bond@rapidrollout.com>
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agos390/idle: fix sequence handling vs cpu hotplug
Heiko Carstens [Fri, 13 Jul 2012 13:45:33 +0000 (15:45 +0200)]
s390/idle: fix sequence handling vs cpu hotplug

commit 0008204ffe85d23382d6fd0f971f3f0fbe70bae2 upstream.

The s390 idle accounting code uses a sequence counter which gets used
when the per cpu idle statistics get updated and read.

One assumption on read access is that only when the sequence counter is
even and did not change while reading all values the result is valid.
On cpu hotplug however the per cpu data structure gets initialized via
a cpu hotplug notifier on CPU_ONLINE.
CPU_ONLINE however is too late, since the onlined cpu is already running
and might access the per cpu data. Worst case is that the data structure
gets initialized while an idle thread is updating its idle statistics.
This will result in an uneven sequence counter after an update.

As a result user space tools like top, which access /proc/stat in order
to get idle stats, will busy loop waiting for the sequence counter to
become even again, which will never happen until the queried cpu will
update its idle statistics again. And even then the sequence counter
will only have an even value for a couple of cpu cycles.

Fix this by moving the initialization of the per cpu idle statistics
to cpu_init(). I prefer that solution in favor of changing the
notifier to CPU_UP_PREPARE, which would be a different solution to
the problem.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agotarget: Check number of unmap descriptors against our limit
Roland Dreier [Mon, 16 Jul 2012 22:34:25 +0000 (15:34 -0700)]
target: Check number of unmap descriptors against our limit

commit 7409a6657aebf8be74c21d0eded80709b27275cb upstream.

Fail UNMAP commands that have more than our reported limit on unmap
descriptors.

Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
[bwh: Backported to 3.2: adjust filename]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agotarget: Fix possible integer underflow in UNMAP emulation
Roland Dreier [Mon, 16 Jul 2012 22:34:24 +0000 (15:34 -0700)]
target: Fix possible integer underflow in UNMAP emulation

commit b7fc7f3777582dea85156a821d78a522a0c083aa upstream.

It's possible for an initiator to send us an UNMAP command with a
descriptor that is less than 8 bytes; in that case it's really bad for
us to set an unsigned int to that value, subtract 8 from it, and then
use that as a limit for our loop (since the value will wrap around to
a huge positive value).

Fix this by making size be signed and only looping if size >= 16 (ie
if we have at least a full descriptor available).

Also remove offset as an obfuscated name for the constant 8.

Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
[bwh: Backported to 3.2: adjust filename, context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agotarget: Fix reading of data length fields for UNMAP commands
Roland Dreier [Mon, 16 Jul 2012 22:34:23 +0000 (15:34 -0700)]
target: Fix reading of data length fields for UNMAP commands

commit 1a5fa4576ec8a462313c7516b31d7453481ddbe8 upstream.

The UNMAP DATA LENGTH and UNMAP BLOCK DESCRIPTOR DATA LENGTH fields
are in the unmap descriptor (the payload transferred to our data out
buffer), not in the CDB itself.  Read them from the correct place in
target_emulated_unmap.

Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
[bwh: Backported to 3.2: adjust filename, context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agotarget: Add range checking to UNMAP emulation
Roland Dreier [Mon, 16 Jul 2012 22:34:22 +0000 (15:34 -0700)]
target: Add range checking to UNMAP emulation

commit 2594e29865c291db162313187612cd9f14538f33 upstream.

When processing an UNMAP command, we need to make sure that the number
of blocks we're asked to UNMAP does not exceed our reported maximum
number of blocks per UNMAP, and that the range of blocks we're
unmapping doesn't go past the end of the device.

Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
[bwh: Backported to 3.2: adjust filename, context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agotarget: Add generation of LOGICAL BLOCK ADDRESS OUT OF RANGE
Roland Dreier [Mon, 16 Jul 2012 22:34:21 +0000 (15:34 -0700)]
target: Add generation of LOGICAL BLOCK ADDRESS OUT OF RANGE

commit e2397c704429025bc6b331a970f699e52f34283e upstream.

Many SCSI commands are defined to return a CHECK CONDITION / ILLEGAL
REQUEST with ASC set to LOGICAL BLOCK ADDRESS OUT OF RANGE if the
initiator sends a command that accesses a too-big LBA.  Add an enum
value and case entries so that target code can return this status.

Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoUSB: option: add ZTE MF821D
Bjørn Mork [Thu, 12 Jul 2012 10:37:32 +0000 (12:37 +0200)]
USB: option: add ZTE MF821D

commit 09110529780890804b22e997ae6b4fe3f0b3b158 upstream.

Sold by O2 (telefonica germany) under the name "LTE4G"

Tested-by: Thomas Schäfer <tschaefer@t-online.de>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoUSB: option: Ignore ZTE (Vodafone) K3570/71 net interfaces
Andrew Bird (Sphere Systems) [Sun, 25 Mar 2012 00:10:28 +0000 (00:10 +0000)]
USB: option: Ignore ZTE (Vodafone) K3570/71 net interfaces

commit f264ddea0109bf7ce8aab920d64a637e830ace5b upstream.

These interfaces need to be handled by QMI/WWAN driver

Signed-off-by: Andrew Bird <ajb@spheresystems.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomwifiex: correction in mcs index check
Amitkumar Karwar [Thu, 12 Jul 2012 01:12:57 +0000 (18:12 -0700)]
mwifiex: correction in mcs index check

commit fe020120cb863ba918c6d603345342a880272c4d upstream.

mwifiex driver supports 2x2 chips as well. Hence valid mcs values
are 0 to 15. The check for mcs index is corrected in this patch.

For example: if 40MHz is enabled and mcs index is 11, "iw link"
command would show "tx bitrate: 108.0 MBit/s" without this patch.
Now it shows "tx bitrate: 108.0 MBit/s MCS 11 40Mhz" with the patch.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agopowerpc: Add "memory" attribute for mfmsr()
Tiejun Chen [Wed, 11 Jul 2012 04:22:46 +0000 (14:22 +1000)]
powerpc: Add "memory" attribute for mfmsr()

commit b416c9a10baae6a177b4f9ee858b8d309542fbef upstream.

Add "memory" attribute in inline assembly language as a compiler
barrier to make sure 4.6.x GCC don't reorder mfmsr().

Signed-off-by: Tiejun Chen <tiejun.chen@windriver.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoudf: Improve table length check to avoid possible overflow
Jan Kara [Tue, 10 Jul 2012 15:58:04 +0000 (17:58 +0200)]
udf: Improve table length check to avoid possible overflow

commit 57b9655d01ef057a523e810d29c37ac09b80eead upstream.

When a partition table length is corrupted to be close to 1 << 32, the
check for its length may overflow on 32-bit systems and we will think
the length is valid. Later on the kernel can crash trying to read beyond
end of buffer. Fix the check to avoid possible overflow.

Reported-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoext4: fix overhead calculation used by ext4_statfs()
Theodore Ts'o [Mon, 9 Jul 2012 20:27:05 +0000 (16:27 -0400)]
ext4: fix overhead calculation used by ext4_statfs()

commit 952fc18ef9ec707ebdc16c0786ec360295e5ff15 upstream.

Commit f975d6bcc7a introduced bug which caused ext4_statfs() to
miscalculate the number of file system overhead blocks.  This causes
the f_blocks field in the statfs structure to be larger than it should
be.  This would in turn cause the "df" output to show the number of
data blocks in the file system and the number of data blocks used to
be larger than they should be.

Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agousbdevfs: Correct amount of data copied to user in processcompl_compat
Hans de Goede [Wed, 4 Jul 2012 07:18:01 +0000 (09:18 +0200)]
usbdevfs: Correct amount of data copied to user in processcompl_compat

commit 2102e06a5f2e414694921f23591f072a5ba7db9f upstream.

iso data buffers may have holes in them if some packets were short, so for
iso urbs we should always copy the entire buffer, just like the regular
processcompl does.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agox86, microcode: Sanitize per-cpu microcode reloading interface
Borislav Petkov [Thu, 21 Jun 2012 12:07:16 +0000 (14:07 +0200)]
x86, microcode: Sanitize per-cpu microcode reloading interface

commit c9fc3f778a6a215ace14ee556067c73982b6d40f upstream.

Microcode reloading in a per-core manner is a very bad idea for both
major x86 vendors. And the thing is, we have such interface with which
we can end up with different microcode versions applied on different
cores of an otherwise homogeneous wrt (family,model,stepping) system.

So turn off the possibility of doing that per core and allow it only
system-wide.

This is a minimal fix which we'd like to see in stable too thus the
more-or-less arbitrary decision to allow system-wide reloading only on
the BSP:

$ echo 1 > /sys/devices/system/cpu/cpu0/microcode/reload
...

and disable the interface on the other cores:

$ echo 1 > /sys/devices/system/cpu/cpu23/microcode/reload
-bash: echo: write error: Invalid argument

Also, allowing the reload only from one CPU (the BSP in
that case) doesn't allow the reload procedure to degenerate
into an O(n^2) deal when triggering reloads from all
/sys/devices/system/cpu/cpuX/microcode/reload sysfs nodes
simultaneously.

A more generic fix will follow.

Cc: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
Link: http://lkml.kernel.org/r/1340280437-7718-2-git-send-email-bp@amd64.org
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agox86, microcode: microcode_core.c simple_strtoul cleanup
Shuah Khan [Sun, 6 May 2012 17:11:04 +0000 (11:11 -0600)]
x86, microcode: microcode_core.c simple_strtoul cleanup

commit e826abd523913f63eb03b59746ffb16153c53dc4 upstream.

Change reload_for_cpu() in kernel/microcode_core.c to call kstrtoul()
instead of calling obsoleted simple_strtoul().

Signed-off-by: Shuah Khan <shuahkhan@gmail.com>
Reviewed-by: Borislav Petkov <bp@alien8.de>
Link: http://lkml.kernel.org/r/1336324264.2897.9.camel@lorien2
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoftrace: Disable function tracing during suspend/resume and hibernation, again
Srivatsa S. Bhat [Sat, 16 Jun 2012 13:30:45 +0000 (15:30 +0200)]
ftrace: Disable function tracing during suspend/resume and hibernation, again

commit 443772d408a25af62498793f6f805ce3c559309a upstream.

If function tracing is enabled for some of the low-level suspend/resume
functions, it leads to triple fault during resume from suspend, ultimately
ending up in a reboot instead of a resume (or a total refusal to come out
of suspended state, on some machines).

This issue was explained in more detail in commit f42ac38c59e0a03d (ftrace:
disable tracing for suspend to ram). However, the changes made by that commit
got reverted by commit cbe2f5a6e84eebb (tracing: allow tracing of
suspend/resume & hibernation code again). So, unfortunately since things are
not yet robust enough to allow tracing of low-level suspend/resume functions,
suspend/resume is still broken when ftrace is enabled.

So fix this by disabling function tracing during suspend/resume & hibernation.

Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@linux.vnet.ibm.com>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoext4: pass a char * to ext4_count_free() instead of a buffer_head ptr
Theodore Ts'o [Sat, 30 Jun 2012 23:14:57 +0000 (19:14 -0400)]
ext4: pass a char * to ext4_count_free() instead of a buffer_head ptr

commit f6fb99cadcd44660c68e13f6eab28333653621e6 upstream.

Make it possible for ext4_count_free to operate on buffers and not
just data in buffer_heads.

Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agousb: gadget: Fix g_ether interface link status
Kevin Cernekee [Mon, 25 Jun 2012 04:11:22 +0000 (21:11 -0700)]
usb: gadget: Fix g_ether interface link status

commit 31bde1ceaa873bcaecd49e829bfabceacc4c512d upstream.

A "usb0" interface that has never been connected to a host has an unknown
operstate, and therefore the IFF_RUNNING flag is (incorrectly) asserted
when queried by ifconfig, ifplugd, etc.  This is a result of calling
netif_carrier_off() too early in the probe function; it should be called
after register_netdev().

Similar problems have been fixed in many other drivers, e.g.:

    e826eafa6 (bonding: Call netif_carrier_off after register_netdevice)
    0d672e9f8 (drivers/net: Call netif_carrier_off at the end of the probe)
    6a3c869a6 (cxgb4: fix reported state of interfaces without link)

Fix is to move netif_carrier_off() to the end of the function.

Signed-off-by: Kevin Cernekee <cernekee@gmail.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agort2800usb: 2001:3c17 is an RT3370 device
Albert Pool [Mon, 14 May 2012 16:08:32 +0000 (18:08 +0200)]
rt2800usb: 2001:3c17 is an RT3370 device

commit 8fd9d059af12786341dec5a688e607bcdb372238 upstream.

D-Link DWA-123 rev A1

Signed-off-by: Albert Pool<albertpool@solcon.nl>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agowireless: rt2x00: rt2800usb more devices were identified
Xose Vazquez Perez [Tue, 17 Apr 2012 14:28:05 +0000 (16:28 +0200)]
wireless: rt2x00: rt2800usb more devices were identified

commit e828b9fb4f6c3513950759d5fb902db5bd054048 upstream.

found in 2012_03_22_RT5572_Linux_STA_v2.6.0.0_DPO

RT3070:
(0x2019,0x5201)  Planex Communications, Inc. RT8070
(0x7392,0x4085)  2L Central Europe BV 8070
7392 is Edimax

RT35xx:
(0x1690,0x0761) Askey
was Fujitsu Stylistic 550, but 1690 is Askey

Signed-off-by: Xose Vazquez Perez <xose.vazquez@gmail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agowireless: rt2x00: rt2800usb add more devices ids
Xose Vazquez Perez [Mon, 16 Apr 2012 23:50:32 +0000 (01:50 +0200)]
wireless: rt2x00: rt2800usb add more devices ids

commit 63b376411173c343bbcb450f95539da91f079e0c upstream.

They were taken from ralink drivers:
2011_0719_RT3070_RT3370_RT5370_RT5372_Linux_STA_V2.5.0.3_DPO
2012_03_22_RT5572_Linux_STA_v2.6.0.0_DPO

0x1eda,0x2210 RT3070 Airties

0x083a,0xb511 RT3370 Panasonic
0x0471,0x20dd RT3370 Philips

0x1690,0x0764 RT35xx Askey
0x0df6,0x0065 RT35xx Sitecom
0x0df6,0x0066 RT35xx Sitecom
0x0df6,0x0068 RT35xx Sitecom

0x2001,0x3c1c RT5370 DLink
0x2001,0x3c1d RT5370 DLink

2001 is D-Link not Alpha

Signed-off-by: Xose Vazquez Perez <xose.vazquez@gmail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
[bwh: Backported to 3.2: drop the 5372 devices]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agocifs: when CONFIG_HIGHMEM is set, serialize the read/write kmaps
Jeff Layton [Wed, 11 Jul 2012 13:09:36 +0000 (09:09 -0400)]
cifs: when CONFIG_HIGHMEM is set, serialize the read/write kmaps

commit 3cf003c08be785af4bee9ac05891a15bcbff856a upstream.

Jian found that when he ran fsx on a 32 bit arch with a large wsize the
process and one of the bdi writeback kthreads would sometimes deadlock
with a stack trace like this:

crash> bt
PID: 2789   TASK: f02edaa0  CPU: 3   COMMAND: "fsx"
 #0 [eed63cbc] schedule at c083c5b3
 #1 [eed63d80] kmap_high at c0500ec8
 #2 [eed63db0] cifs_async_writev at f7fabcd7 [cifs]
 #3 [eed63df0] cifs_writepages at f7fb7f5c [cifs]
 #4 [eed63e50] do_writepages at c04f3e32
 #5 [eed63e54] __filemap_fdatawrite_range at c04e152a
 #6 [eed63ea4] filemap_fdatawrite at c04e1b3e
 #7 [eed63eb4] cifs_file_aio_write at f7fa111a [cifs]
 #8 [eed63ecc] do_sync_write at c052d202
 #9 [eed63f74] vfs_write at c052d4ee
#10 [eed63f94] sys_write at c052df4c
#11 [eed63fb0] ia32_sysenter_target at c0409a98
    EAX: 00000004  EBX: 00000003  ECX: abd73b73  EDX: 012a65c6
    DS:  007b      ESI: 012a65c6  ES:  007b      EDI: 00000000
    SS:  007b      ESP: bf8db178  EBP: bf8db1f8  GS:  0033
    CS:  0073      EIP: 40000424  ERR: 00000004  EFLAGS: 00000246

Each task would kmap part of its address array before getting stuck, but
not enough to actually issue the write.

This patch fixes this by serializing the marshal_iov operations for
async reads and writes. The idea here is to ensure that cifs
aggressively tries to populate a request before attempting to fulfill
another one. As soon as all of the pages are kmapped for a request, then
we can unlock and allow another one to proceed.

There's no need to do this serialization on non-CONFIG_HIGHMEM arches
however, so optimize all of this out when CONFIG_HIGHMEM isn't set.

Reported-by: Jian Li <jiali@redhat.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agor8169: RxConfig hack for the 8168evl.
françois romieu [Wed, 20 Jun 2012 12:09:18 +0000 (12:09 +0000)]
r8169: RxConfig hack for the 8168evl.

commit eb2dc35d99028b698cdedba4f5522bc43e576bd2 upstream.

The 8168evl (RTL_GIGA_MAC_VER_34) based Gigabyte GA-990FXA motherboards
are very prone to NETDEV watchdog problems without this change. See
https://bugzilla.kernel.org/show_bug.cgi?id=42899 for instance.

I don't know why it *works*. It's depressingly effective though.

For the record:
- the problem may go along IOMMU (AMD-Vi) errors but it really looks
  like a red herring.
- the patch sets the RX_MULTI_EN bit. If the 8168c doc is any guide,
  the chipset now fetches several Rx descriptors at a time.
- long ago the driver ignored the RX_MULTI_EN bit.
  e542a2269f232d61270ceddd42b73a4348dee2bb changed the RxConfig
  settings. Whatever the problem it's now labeled a regression.
- Realtek's own driver can identify two different 8168evl devices
  (CFG_METHOD_16 and CFG_METHOD_17) where the r8169 driver only
  sees one. It sucks.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agox86: Fix boot on Twinhead H12Y
Alan Cox [Tue, 15 May 2012 17:44:15 +0000 (18:44 +0100)]
x86: Fix boot on Twinhead H12Y

commit 80b3e557371205566a71e569fbfcce5b11f92dbe upstream.

Despite lots of investigation into why this is needed we don't
know or have an elegant cure. The only answer found on this
laptop is to mark a problem region as used so that Linux doesn't
put anything there.

Currently all the users add reserve= command lines and anyone
not knowing this needs to find the magic page that documents it.
Automate it instead.

Signed-off-by: Alan Cox <alan@linux.intel.com>
Tested-and-bugfixed-by: Arne Fitzenreiter <arne@fitzenreiter.de>
Resolves-bug: https://bugzilla.kernel.org/show_bug.cgi?id=10231
Link: http://lkml.kernel.org/r/20120515174347.5109.94551.stgit@bluebook
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agocx25821: Remove bad strcpy to read-only char*
Ezequiel Garcia [Wed, 18 Jul 2012 13:05:26 +0000 (10:05 -0300)]
cx25821: Remove bad strcpy to read-only char*

commit 380e99fc44d79bc35af9ff1d3316ef4027ce775e upstream.

The strcpy was being used to set the name of the board.  Since the
destination char* was read-only and the name is set statically at
compile time; this was both wrong and redundant.

The type of char* is changed to const char* to prevent future errors.

Reported-by: Radek Masin <radek@masin.eu>
Signed-off-by: Ezequiel Garcia <elezegarcia@gmail.com>
[ Taking directly due to vacations   - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agopowerpc/ftrace: Fix assembly trampoline register usage
roger blofeld [Thu, 21 Jun 2012 05:27:14 +0000 (05:27 +0000)]
powerpc/ftrace: Fix assembly trampoline register usage

commit fd5a42980e1cf327b7240adf5e7b51ea41c23437 upstream.

Just like the module loader, ftrace needs to be updated to use r12
instead of r11 with newer gcc's.

Signed-off-by: Roger Blofeld <blofeldus@yahoo.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agosched/nohz: Fix rq->cpu_load calculations some more
Peter Zijlstra [Thu, 17 May 2012 15:15:29 +0000 (17:15 +0200)]
sched/nohz: Fix rq->cpu_load calculations some more

commit 5aaa0b7a2ed5b12692c9ffb5222182bd558d3146 upstream.

Follow up on commit 556061b00 ("sched/nohz: Fix rq->cpu_load[]
calculations") since while that fixed the busy case it regressed the
mostly idle case.

Add a callback from the nohz exit to also age the rq->cpu_load[]
array. This closes the hole where either there was no nohz load
balance pass during the nohz, or there was a 'significant' amount of
idle time between the last nohz balance and the nohz exit.

So we'll update unconditionally from the tick to not insert any
accidental 0 load periods while busy, and we try and catch up from
nohz idle balance and nohz exit. Both these are still prone to missing
a jiffy, but that has always been the case.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: pjt@google.com
Cc: Venkatesh Pallipadi <venki@google.com>
Link: http://lkml.kernel.org/n/tip-kt0trz0apodbf84ucjfdbr1a@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
[bwh: Backported to 3.2: adjust filenames and context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agosched/nohz: Fix rq->cpu_load[] calculations
Peter Zijlstra [Fri, 11 May 2012 15:31:26 +0000 (17:31 +0200)]
sched/nohz: Fix rq->cpu_load[] calculations

commit 556061b00c9f2fd6a5524b6bde823ef12f299ecf upstream.

While investigating why the load-balancer did funny I found that the
rq->cpu_load[] tables were completely screwy.. a bit more digging
revealed that the updates that got through were missing ticks followed
by a catchup of 2 ticks.

The catchup assumes the cpu was idle during that time (since only nohz
can cause missed ticks and the machine is idle etc..) this means that
esp. the higher indices were significantly lower than they ought to
be.

The reason for this is that its not correct to compare against jiffies
on every jiffy on any other cpu than the cpu that updates jiffies.

This patch cludges around it by only doing the catch-up stuff from
nohz_idle_balance() and doing the regular stuff unconditionally from
the tick.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: pjt@google.com
Cc: Venkatesh Pallipadi <venki@google.com>
Link: http://lkml.kernel.org/n/tip-tp4kj18xdd5aj4vvj0qg55s2@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
[bwh: Backported to 3.2: adjust filenames and context; keep functions static]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoFix NULL dereferences in scsi_cmd_to_driver
Mark Rustad [Thu, 21 Jun 2012 19:23:42 +0000 (12:23 -0700)]
Fix NULL dereferences in scsi_cmd_to_driver

commit 222a806af830fda34ad1f6bc991cd226916de060 upstream.

Avoid crashing if the private_data pointer happens to be NULL. This has
been seen sometimes when a host reset happens, notably when there are
many LUNs:

host3: Assigned Port ID 0c1601
scsi host3: libfc: Host reset succeeded on port (0c1601)
BUG: unable to handle kernel NULL pointer dereference at 0000000000000350
IP: [<ffffffff81352bb8>] scsi_send_eh_cmnd+0x58/0x3a0
<snip>
Process scsi_eh_3 (pid: 4144, threadinfo ffff88030920c000, task ffff880326b160c0)
Stack:
 000000010372e6ba 0000000000000282 000027100920dca0 ffffffffa0038ee0
 0000000000000000 0000000000030003 ffff88030920dc80 ffff88030920dc80
 00000002000e0000 0000000a00004000 ffff8803242f7760 ffff88031326ed80
Call Trace:
 [<ffffffff8105b590>] ? lock_timer_base+0x70/0x70
 [<ffffffff81352fbe>] scsi_eh_tur+0x3e/0xc0
 [<ffffffff81353a36>] scsi_eh_test_devices+0x76/0x170
 [<ffffffff81354125>] scsi_eh_host_reset+0x85/0x160
 [<ffffffff81354291>] scsi_eh_ready_devs+0x91/0x110
 [<ffffffff813543fd>] scsi_unjam_host+0xed/0x1f0
 [<ffffffff813546a8>] scsi_error_handler+0x1a8/0x200
 [<ffffffff81354500>] ? scsi_unjam_host+0x1f0/0x1f0
 [<ffffffff8106ec3e>] kthread+0x9e/0xb0
 [<ffffffff81509264>] kernel_thread_helper+0x4/0x10
 [<ffffffff8106eba0>] ? kthread_freezable_should_stop+0x70/0x70
 [<ffffffff81509260>] ? gs_change+0x13/0x13
Code: 25 28 00 00 00 48 89 45 c8 31 c0 48 8b 87 80 00 00 00 48 8d b5 60 ff ff ff 89 d1 48 89 fb 41 89 d6 4c 89 fa 48 8b 80 b8 00 00 00
 <48> 8b 80 50 03 00 00 48 8b 00 48 89 85 38 ff ff ff 48 8b 07 4c
RIP  [<ffffffff81352bb8>] scsi_send_eh_cmnd+0x58/0x3a0
 RSP <ffff88030920dc50>
CR2: 0000000000000350

Signed-off-by: Mark Rustad <mark.d.rustad@intel.com>
Tested-by: Marcus Dennis <marcusx.e.dennis@intel.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
[bwh: Backported to 3.2: adjust filename, context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm/hugetlb: fix warning in alloc_huge_page/dequeue_huge_page_vma
Konstantin Khlebnikov [Wed, 25 Apr 2012 23:01:46 +0000 (16:01 -0700)]
mm/hugetlb: fix warning in alloc_huge_page/dequeue_huge_page_vma

commit b1c12cbcd0a02527c180a862e8971e249d3b347d upstream.

Stable note: Not tracked in Bugzilla. [get|put]_mems_allowed() is extremely
expensive and severely impacted page allocator performance. This
is part of a series of patches that reduce page allocator overhead.

Fix a gcc warning (and bug?) introduced in cc9a6c877 ("cpuset: mm: reduce
large amounts of memory barrier related damage v3")

Local variable "page" can be uninitialized if the nodemask from vma policy
does not intersects with nodemask from cpuset.  Even if it doesn't happens
it is better to initialize this variable explicitly than to introduce
a kernel oops in a weird corner case.

mm/hugetlb.c: In function `alloc_huge_page':
mm/hugetlb.c:1135:5: warning: `page' may be used uninitialized in this function

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agocpuset: mm: reduce large amounts of memory barrier related damage v3
Mel Gorman [Wed, 21 Mar 2012 23:34:11 +0000 (16:34 -0700)]
cpuset: mm: reduce large amounts of memory barrier related damage v3

commit cc9a6c8776615f9c194ccf0b63a0aa5628235545 upstream.

Stable note:  Not tracked in Bugzilla. [get|put]_mems_allowed() is extremely
expensive and severely impacted page allocator performance. This
is part of a series of patches that reduce page allocator overhead.

Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.

[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths.  This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32.  The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.

For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.

This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side.  This is much cheaper on some architectures, including x86.  The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.

While updating the nodemask, a check is made to see if a false failure
is a risk.  If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.

In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%.  The
actual results were

                             3.3.0-rc3          3.3.0-rc3
                             rc3-vanilla        nobarrier-v2r1
    Clients   1 UserTime       0.07 (  0.00%)   0.08 (-14.19%)
    Clients   2 UserTime       0.07 (  0.00%)   0.07 (  2.72%)
    Clients   4 UserTime       0.08 (  0.00%)   0.07 (  3.29%)
    Clients   1 SysTime        0.70 (  0.00%)   0.65 (  6.65%)
    Clients   2 SysTime        0.85 (  0.00%)   0.82 (  3.65%)
    Clients   4 SysTime        1.41 (  0.00%)   1.41 (  0.32%)
    Clients   1 WallTime       0.77 (  0.00%)   0.74 (  4.19%)
    Clients   2 WallTime       0.47 (  0.00%)   0.45 (  3.73%)
    Clients   4 WallTime       0.38 (  0.00%)   0.37 (  1.58%)
    Clients   1 Flt/sec/cpu  497620.28 (  0.00%) 520294.53 (  4.56%)
    Clients   2 Flt/sec/cpu  414639.05 (  0.00%) 429882.01 (  3.68%)
    Clients   4 Flt/sec/cpu  257959.16 (  0.00%) 258761.48 (  0.31%)
    Clients   1 Flt/sec      495161.39 (  0.00%) 517292.87 (  4.47%)
    Clients   2 Flt/sec      820325.95 (  0.00%) 850289.77 (  3.65%)
    Clients   4 Flt/sec      1020068.93 (  0.00%) 1022674.06 (  0.26%)
    MMTests Statistics: duration
    Sys Time Running Test (seconds)             135.68    132.17
    User+Sys Time Running Test (seconds)         164.2    160.13
    Total Elapsed Time (seconds)                123.46    120.87

The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected).  The
actual number of page faults is noticeably improved.

For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.

To test the actual bug the commit fixed I opened two terminals.  The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data.  In a second window, the nodemask of the
cpuset was continually randomised in a loop.

Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Mel Gorman <mgorman@suse.de>
[bwh: Forward-ported from 3.0 to 3.2: apply the upstream changes
 to get_any_partial()]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: vmscan: convert global reclaim to per-memcg LRU lists
Johannes Weiner [Fri, 13 Jan 2012 01:18:06 +0000 (17:18 -0800)]
mm: vmscan: convert global reclaim to per-memcg LRU lists

commit b95a2f2d486d0d768a92879c023a03757b9c7e58 upstream - WARNING: this is a substitute patch.

Stable note: Not tracked in Bugzilla. This is a partial backport of an
upstream commit addressing a completely different issue
that accidentally contained an important fix. The workload
this patch helps was memcached when IO is started in the
background. memcached should stay resident but without this patch
it gets swapped. Sometimes this manifests as a drop in throughput
but mostly it was observed through /proc/vmstat.

Commit [246e87a9: memcg: fix get_scan_count() for small targets] was meant
to fix a problem whereby small scan targets on memcg were ignored causing
priority to raise too sharply. It forced scanning to take place if the
target was small, memcg or kswapd.

From the time it was introduced it caused excessive reclaim by kswapd
with workloads being pushed to swap that previously would have stayed
resident. This was accidentally fixed in commit [b95a2f2d: mm: vmscan:
convert global reclaim to per-memcg LRU lists] by making it harder for
kswapd to force scan small targets but that patchset is not suitable for
backporting. This was later changed again by commit [90126375: mm/vmscan:
push lruvec pointer into get_scan_count()] into a format that looks
like it would be a straight-forward backport but there is a subtle
difference due to the use of lruvecs.

The impact of the accidental fix is to make it harder for kswapd to force
scan small targets by taking zone->all_unreclaimable into account. This
patch is the closest equivalent available based on what is backported.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: test PageSwapBacked in lumpy reclaim
Hugh Dickins [Tue, 10 Jan 2012 23:08:33 +0000 (15:08 -0800)]
mm: test PageSwapBacked in lumpy reclaim

commit 043bcbe5ec51e0478ef2b44acef17193e01d7f70 upstream.

Stable note: Not tracked in Bugzilla. There were reports of shared
mapped pages being unfairly reclaimed in comparison to older kernels.
This is being addressed over time. Even though the subject
refers to lumpy reclaim, it impacts compaction as well.

Lumpy reclaim does well to stop at a PageAnon when there's no swap, but
better is to stop at any PageSwapBacked, which includes shmem/tmpfs too.

Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm/vmscan.c: consider swap space when deciding whether to continue reclaim
Minchan Kim [Tue, 10 Jan 2012 23:08:18 +0000 (15:08 -0800)]
mm/vmscan.c: consider swap space when deciding whether to continue reclaim

commit 86cfd3a45042ab242d47f3935a02811a402beab6 upstream.

Stable note: Not tracked in Bugzilla. This patch reduces kswapd CPU
usage on swapless systems with high anonymous memory usage.

It's pointless to continue reclaiming when we have no swap space and lots
of anon pages in the inactive list.

Without this patch, it is possible when swap is disabled to continue
trying to reclaim when there are only anonymous pages in the system even
though that will not make any progress.

Signed-off-by: Minchan Kim <minchan@kernel.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agovmscan: activate executable pages after first usage
Konstantin Khlebnikov [Tue, 10 Jan 2012 23:07:03 +0000 (15:07 -0800)]
vmscan: activate executable pages after first usage

commit c909e99364c8b6ca07864d752950b6b4ecf6bef4 upstream.

Stable note: Not tracked in Bugzilla. There were reports of shared
mapped pages being unfairly reclaimed in comparison to older kernels.
This is being addressed over time.

Logic added in commit 8cab4754d24a0 ("vmscan: make mapped executable pages
the first class citizen") was noticeably weakened in commit
645747462435d84 ("vmscan: detect mapped file pages used only once").

Currently these pages can become "first class citizens" only after second
usage.  After this patch page_check_references() will activate they after
first usage, and executable code gets yet better chance to stay in memory.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agovmscan: promote shared file mapped pages
Konstantin Khlebnikov [Tue, 10 Jan 2012 23:06:59 +0000 (15:06 -0800)]
vmscan: promote shared file mapped pages

commit 34dbc67a644f11ab3475d822d72e25409911e760 upstream.

Stable note: Not tracked in Bugzilla. There were reports of shared
mapped pages being unfairly reclaimed in comparison to older kernels.
This is being addressed over time. The specific workload being
addressed here in described in paragraph four and while paragraph
five says it did not help performance as such, it made a difference
to major page faults. I'm aware of at least one bug for a large
vendor that was due to increased major faults.

Commit 645747462435 ("vmscan: detect mapped file pages used only once")
greatly decreases lifetime of single-used mapped file pages.
Unfortunately it also decreases life time of all shared mapped file
pages.  Because after commit bf3f3bc5e7347 ("mm: don't mark_page_accessed
in fault path") page-fault handler does not mark page active or even
referenced.

Thus page_check_references() activates file page only if it was used twice
while it stays in inactive list, meanwhile it activates anon pages after
first access.  Inactive list can be small enough, this way reclaimer can
accidentally throw away any widely used page if it wasn't used twice in
short period.

After this patch page_check_references() also activate file mapped page at
first inactive list scan if this page is already used multiple times via
several ptes.

I found this while trying to fix degragation in rhel6 (~2.6.32) from rhel5
(~2.6.18).  There a complete mess with >100 web/mail/spam/ftp containers,
they share all their files but there a lot of anonymous pages: ~500mb
shared file mapped memory and 15-20Gb non-shared anonymous memory.  In
this situation major-pagefaults are very costly, because all containers
share the same page.  In my load kernel created a disproportionate
pressure on the file memory, compared with the anonymous, they equaled
only if I raise swappiness up to 150 =)

These patches actually wasn't helped a lot in my problem, but I saw
noticable (10-20 times) reduce in count and average time of
major-pagefault in file-mapped areas.

Actually both patches are fixes for commit v2.6.33-5448-g6457474, because
it was aimed at one scenario (singly used pages), but it breaks the logic
in other scenarios (shared and/or executable pages)

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Acked-by: Pekka Enberg <penberg@kernel.org>
Acked-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Shaohua Li <shaohua.li@intel.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: vmscan: check if reclaim should really abort even if compaction_ready() is true...
Mel Gorman [Fri, 13 Jan 2012 01:19:49 +0000 (17:19 -0800)]
mm: vmscan: check if reclaim should really abort even if compaction_ready() is true for one zone

commit 0cee34fd72c582b4f8ad8ce00645b75fb4168199 upstream.

Stable note: Not tracked on Bugzilla. THP and compaction was found to
aggressively reclaim pages and stall systems under different
situations that was addressed piecemeal over time.

If compaction can proceed for a given zone, shrink_zones() does not
reclaim any more pages from it.  After commit [e0c2327: vmscan: abort
reclaim/compaction if compaction can proceed], do_try_to_free_pages()
tries to finish as soon as possible once one zone can compact.

This was intended to prevent slabs being shrunk unnecessarily but there
are side-effects.  One is that a small zone that is ready for compaction
will abort reclaim even if the chances of successfully allocating a THP
from that zone is small.  It also means that reclaim can return too early
even though sc->nr_to_reclaim pages were not reclaimed.

This partially reverts the commit until it is proven that slabs are really
being shrunk unnecessarily but preserves the check to return 1 to avoid
OOM if reclaim was aborted prematurely.

[aarcange@redhat.com: This patch replaces a revert from Andrea]
Signed-off-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Andy Isaacson <adi@hexapodia.org>
Cc: Nai Xia <nai.xia@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: vmscan: do not OOM if aborting reclaim to start compaction
Mel Gorman [Fri, 13 Jan 2012 01:19:33 +0000 (17:19 -0800)]
mm: vmscan: do not OOM if aborting reclaim to start compaction

commit 7335084d446b83cbcb15da80497d03f0c1dc9e21 upstream.

Stable note: Not tracked in Bugzilla. This patch makes later patches
easier to apply but otherwise has little to justify it. The
problem it fixes was never observed but the source of the
theoretical problem did not exist for very long.

During direct reclaim it is possible that reclaim will be aborted so that
compaction can be attempted to satisfy a high-order allocation.  If this
decision is made before any pages are reclaimed, it is possible that 0 is
returned to the page allocator potentially triggering an OOM.  This has
not been observed but it is a possibility so this patch addresses it.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Andy Isaacson <adi@hexapodia.org>
Cc: Nai Xia <nai.xia@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: vmscan: when reclaiming for compaction, ensure there are sufficient free pages...
Mel Gorman [Fri, 13 Jan 2012 01:19:45 +0000 (17:19 -0800)]
mm: vmscan: when reclaiming for compaction, ensure there are sufficient free pages available

commit fe4b1b244bdb96136855f2c694071cb09d140766 upstream.

Stable note: Not tracked on Bugzilla. THP and compaction was found to
aggressively reclaim pages and stall systems under different
situations that was addressed piecemeal over time. This patch
addresses a problem where the fix regressed THP allocation
success rates.

In commit e0887c19 ("vmscan: limit direct reclaim for higher order
allocations"), Rik noted that reclaim was too aggressive when THP was
enabled.  In his initial patch he used the number of free pages to decide
if reclaim should abort for compaction.  My feedback was that reclaim and
compaction should be using the same logic when deciding if reclaim should
be aborted.

Unfortunately, this had the effect of reducing THP success rates when the
workload included something like streaming reads that continually
allocated pages.  The window during which compaction could run and return
a THP was too small.

This patch combines Rik's two patches together.  compaction_suitable() is
still used to decide if reclaim should be aborted to allow compaction is
used.  However, it will also ensure that there is a reasonable buffer of
free pages available.  This improves upon the THP allocation success rates
but bounds the number of pages that are freed for compaction.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel<riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Andy Isaacson <adi@hexapodia.org>
Cc: Nai Xia <nai.xia@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: compaction: introduce sync-light migration for use by compaction
Mel Gorman [Fri, 13 Jan 2012 01:19:43 +0000 (17:19 -0800)]
mm: compaction: introduce sync-light migration for use by compaction

commit a6bc32b899223a877f595ef9ddc1e89ead5072b8 upstream.

Stable note: Not tracked in Buzilla. This was part of a series that
reduced interactivity stalls experienced when THP was enabled.
These stalls were particularly noticable when copying data
to a USB stick but the experiences for users varied a lot.

This patch adds a lightweight sync migrate operation MIGRATE_SYNC_LIGHT
mode that avoids writing back pages to backing storage.  Async compaction
maps to MIGRATE_ASYNC while sync compaction maps to MIGRATE_SYNC_LIGHT.
For other migrate_pages users such as memory hotplug, MIGRATE_SYNC is
used.

This avoids sync compaction stalling for an excessive length of time,
particularly when copying files to a USB stick where there might be a
large number of dirty pages backed by a filesystem that does not support
->writepages.

[aarcange@redhat.com: This patch is heavily based on Andrea's work]
[akpm@linux-foundation.org: fix fs/nfs/write.c build]
[akpm@linux-foundation.org: fix fs/btrfs/disk-io.c build]
Signed-off-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Andy Isaacson <adi@hexapodia.org>
Cc: Nai Xia <nai.xia@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: compaction: make isolate_lru_page() filter-aware again
Mel Gorman [Fri, 13 Jan 2012 01:19:38 +0000 (17:19 -0800)]
mm: compaction: make isolate_lru_page() filter-aware again

commit c82449352854ff09e43062246af86bdeb628f0c3 upstream.

Stable note: Not tracked in Bugzilla. A fix aimed at preserving page aging
information by reducing LRU list churning had the side-effect of
reducing THP allocation success rates. This was part of a series
to restore the success rates while preserving the reclaim fix.

Commit 39deaf85 ("mm: compaction: make isolate_lru_page() filter-aware")
noted that compaction does not migrate dirty or writeback pages and that
is was meaningless to pick the page and re-add it to the LRU list.  This
had to be partially reverted because some dirty pages can be migrated by
compaction without blocking.

This patch updates "mm: compaction: make isolate_lru_page" by skipping
over pages that migration has no possibility of migrating to minimise LRU
disruption.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel<riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Reviewed-by: Minchan Kim <minchan@kernel.org>
Cc: Dave Jones <davej@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Andy Isaacson <adi@hexapodia.org>
Cc: Nai Xia <nai.xia@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: page allocator: do not call direct reclaim for THP allocations while compaction...
Mel Gorman [Fri, 13 Jan 2012 01:19:41 +0000 (17:19 -0800)]
mm: page allocator: do not call direct reclaim for THP allocations while compaction is deferred

commit 66199712e9eef5aede09dbcd9dfff87798a66917 upstream.

Stable note: Not tracked in Buzilla. This was part of a series that
reduced interactivity stalls experienced when THP was enabled.

If compaction is deferred, direct reclaim is used to try to free enough
pages for the allocation to succeed.  For small high-orders, this has a
reasonable chance of success.  However, if the caller has specified
__GFP_NO_KSWAPD to limit the disruption to the system, it makes more sense
to fail the allocation rather than stall the caller in direct reclaim.
This patch skips direct reclaim if compaction is deferred and the caller
specifies __GFP_NO_KSWAPD.

Async compaction only considers a subset of pages so it is possible for
compaction to be deferred prematurely and not enter direct reclaim even in
cases where it should.  To compensate for this, this patch also defers
compaction only if sync compaction failed.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Acked-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: Rik van Riel<riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Andy Isaacson <adi@hexapodia.org>
Cc: Nai Xia <nai.xia@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: compaction: determine if dirty pages can be migrated without blocking within...
Mel Gorman [Fri, 13 Jan 2012 01:19:34 +0000 (17:19 -0800)]
mm: compaction: determine if dirty pages can be migrated without blocking within ->migratepage

commit b969c4ab9f182a6e1b2a0848be349f99714947b0 upstream.

Stable note: Not tracked in Bugzilla. A fix aimed at preserving page
aging information by reducing LRU list churning had the side-effect
of reducing THP allocation success rates. This was part of a series
to restore the success rates while preserving the reclaim fix.

Asynchronous compaction is used when allocating transparent hugepages to
avoid blocking for long periods of time.  Due to reports of stalling,
there was a debate on disabling synchronous compaction but this severely
impacted allocation success rates.  Part of the reason was that many dirty
pages are skipped in asynchronous compaction by the following check;

if (PageDirty(page) && !sync &&
mapping->a_ops->migratepage != migrate_page)
rc = -EBUSY;

This skips over all mapping aops using buffer_migrate_page() even though
it is possible to migrate some of these pages without blocking.  This
patch updates the ->migratepage callback with a "sync" parameter.  It is
the responsibility of the callback to fail gracefully if migration would
block.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Andy Isaacson <adi@hexapodia.org>
Cc: Nai Xia <nai.xia@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: compaction: allow compaction to isolate dirty pages
Mel Gorman [Fri, 13 Jan 2012 01:19:22 +0000 (17:19 -0800)]
mm: compaction: allow compaction to isolate dirty pages

commit a77ebd333cd810d7b680d544be88c875131c2bd3 upstream.

Stable note: Not tracked in Bugzilla. A fix aimed at preserving page aging
information by reducing LRU list churning had the side-effect of
reducing THP allocation success rates. This was part of a series
to restore the success rates while preserving the reclaim fix.

Short summary: There are severe stalls when a USB stick using VFAT is
used with THP enabled that are reduced by this series.  If you are
experiencing this problem, please test and report back and considering I
have seen complaints from openSUSE and Fedora users on this as well as a
few private mails, I'm guessing it's a widespread issue.  This is a new
type of USB-related stall because it is due to synchronous compaction
writing where as in the past the big problem was dirty pages reaching
the end of the LRU and being written by reclaim.

Am cc'ing Andrew this time and this series would replace
mm-do-not-stall-in-synchronous-compaction-for-thp-allocations.patch.
I'm also cc'ing Dave Jones as he might have merged that patch to Fedora
for wider testing and ideally it would be reverted and replaced by this
series.

That said, the later patches could really do with some review.  If this
series is not the answer then a new direction needs to be discussed
because as it is, the stalls are unacceptable as the results in this
leader show.

For testers that try backporting this to 3.1, it won't work because
there is a non-obvious dependency on not writing back pages in direct
reclaim so you need those patches too.

Changelog since V5
o Rebase to 3.2-rc5
o Tidy up the changelogs a bit

Changelog since V4
o Added reviewed-bys, credited Andrea properly for sync-light
o Allow dirty pages without mappings to be considered for migration
o Bound the number of pages freed for compaction
o Isolate PageReclaim pages on their own LRU list

This is against 3.2-rc5 and follows on from discussions on "mm: Do
not stall in synchronous compaction for THP allocations" and "[RFC
PATCH 0/5] Reduce compaction-related stalls". Initially, the proposed
patch eliminated stalls due to compaction which sometimes resulted in
user-visible interactivity problems on browsers by simply never using
sync compaction. The downside was that THP success allocation rates
were lower because dirty pages were not being migrated as reported by
Andrea. His approach at fixing this was nacked on the grounds that
it reverted fixes from Rik merged that reduced the amount of pages
reclaimed as it severely impacted his workloads performance.

This series attempts to reconcile the requirements of maximising THP
usage, without stalling in a user-visible fashion due to compaction
or cheating by reclaiming an excessive number of pages.

Patch 1 partially reverts commit 39deaf85 to allow migration to isolate
dirty pages. This is because migration can move some dirty
pages without blocking.

Patch 2 notes that the /proc/sys/vm/compact_memory handler is not using
synchronous compaction when it should be. This is unrelated
to the reported stalls but is worth fixing.

Patch 3 checks if we isolated a compound page during lumpy scan and
account for it properly. For the most part, this affects
tracing so it's unrelated to the stalls but worth fixing.

Patch 4 notes that it is possible to abort reclaim early for compaction
and return 0 to the page allocator potentially entering the
"may oom" path. This has not been observed in practice but
the rest of the series potentially makes it easier to happen.

Patch 5 adds a sync parameter to the migratepage callback and gives
the callback responsibility for migrating the page without
blocking if sync==false. For example, fallback_migrate_page
will not call writepage if sync==false. This increases the
number of pages that can be handled by asynchronous compaction
thereby reducing stalls.

Patch 6 restores filter-awareness to isolate_lru_page for migration.
In practice, it means that pages under writeback and pages
without a ->migratepage callback will not be isolated
for migration.

Patch 7 avoids calling direct reclaim if compaction is deferred but
makes sure that compaction is only deferred if sync
compaction was used.

Patch 8 introduces a sync-light migration mechanism that sync compaction
uses. The objective is to allow some stalls but to not call
->writepage which can lead to significant user-visible stalls.

Patch 9 notes that while we want to abort reclaim ASAP to allow
compation to go ahead that we leave a very small window of
opportunity for compaction to run. This patch allows more pages
to be freed by reclaim but bounds the number to a reasonable
level based on the high watermark on each zone.

Patch 10 allows slabs to be shrunk even after compaction_ready() is
true for one zone. This is to avoid a problem whereby a single
small zone can abort reclaim even though no pages have been
reclaimed and no suitably large zone is in a usable state.

Patch 11 fixes a problem with the rate of page scanning. As reclaim is
rarely stalling on pages under writeback it means that scan
rates are very high. This is particularly true for direct
reclaim which is not calling writepage. The vmstat figures
implied that much of this was busy work with PageReclaim pages
marked for immediate reclaim. This patch is a prototype that
moves these pages to their own LRU list.

This has been tested and other than 2 USB keys getting trashed,
nothing horrible fell out. That said, I am a bit unhappy with the
rescue logic in patch 11 but did not find a better way around it. It
does significantly reduce scan rates and System CPU time indicating
it is the right direction to take.

What is of critical importance is that stalls due to compaction
are massively reduced even though sync compaction was still
allowed. Testing from people complaining about stalls copying to USBs
with THP enabled are particularly welcome.

The following tests all involve THP usage and USB keys in some
way. Each test follows this type of pattern

1. Read from some fast fast storage, be it raw device or file. Each time
   the copy finishes, start again until the test ends
2. Write a large file to a filesystem on a USB stick. Each time the copy
   finishes, start again until the test ends
3. When memory is low, start an alloc process that creates a mapping
   the size of physical memory to stress THP allocation. This is the
   "real" part of the test and the part that is meant to trigger
   stalls when THP is enabled. Copying continues in the background.
4. Record the CPU usage and time to execute of the alloc process
5. Record the number of THP allocs and fallbacks as well as the number of THP
   pages in use a the end of the test just before alloc exited
6. Run the test 5 times to get an idea of variability
7. Between each run, sync is run and caches dropped and the test
   waits until nr_dirty is a small number to avoid interference
   or caching between iterations that would skew the figures.

The individual tests were then

writebackCPDeviceBasevfat
Disable THP, read from a raw device (sda), vfat on USB stick
writebackCPDeviceBaseext4
Disable THP, read from a raw device (sda), ext4 on USB stick
writebackCPDevicevfat
THP enabled, read from a raw device (sda), vfat on USB stick
writebackCPDeviceext4
THP enabled, read from a raw device (sda), ext4 on USB stick
writebackCPFilevfat
THP enabled, read from a file on fast storage and USB, both vfat
writebackCPFileext4
THP enabled, read from a file on fast storage and USB, both ext4

The kernels tested were

3.1 3.1
vanilla 3.2-rc5
freemore Patches 1-10
immediate Patches 1-11
andrea The 8 patches Andrea posted as a basis of comparison

The results are very long unfortunately. I'll start with the case
where we are not using THP at all

writebackCPDeviceBasevfat
                   3.1.0-vanilla         rc5-vanilla       freemore-v6r1        isolate-v6r1         andrea-v2r1
System Time         1.28 (    0.00%)   54.49 (-4143.46%)   48.63 (-3687.69%)    4.69 ( -265.11%)   51.88 (-3940.81%)
+/-                 0.06 (    0.00%)    2.45 (-4305.55%)    4.75 (-8430.57%)    7.46 (-13282.76%)    4.76 (-8440.70%)
User Time           0.09 (    0.00%)    0.05 (   40.91%)    0.06 (   29.55%)    0.07 (   15.91%)    0.06 (   27.27%)
+/-                 0.02 (    0.00%)    0.01 (   45.39%)    0.02 (   25.07%)    0.00 (   77.06%)    0.01 (   52.24%)
Elapsed Time      110.27 (    0.00%)   56.38 (   48.87%)   49.95 (   54.70%)   11.77 (   89.33%)   53.43 (   51.54%)
+/-                 7.33 (    0.00%)    3.77 (   48.61%)    4.94 (   32.63%)    6.71 (    8.50%)    4.76 (   35.03%)
THP Active          0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)
+/-                 0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)
Fault Alloc         0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)
+/-                 0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)
Fault Fallback      0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)
+/-                 0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)    0.00 (    0.00%)

The THP figures are obviously all 0 because THP was enabled. The
main thing to watch is the elapsed times and how they compare to
times when THP is enabled later. It's also important to note that
elapsed time is improved by this series as System CPu time is much
reduced.

writebackCPDevicevfat

                   3.1.0-vanilla         rc5-vanilla       freemore-v6r1        isolate-v6r1         andrea-v2r1
System Time         1.22 (    0.00%)   13.89 (-1040.72%)   46.40 (-3709.20%)    4.44 ( -264.37%)   47.37 (-3789.33%)
+/-                 0.06 (    0.00%)   22.82 (-37635.56%)    3.84 (-6249.44%)    6.48 (-10618.92%)    6.60
(-10818.53%)
User Time           0.06 (    0.00%)    0.06 (   -6.90%)    0.05 (   17.24%)    0.05 (   13.79%)    0.04 (   31.03%)
+/-                 0.01 (    0.00%)    0.01 (   33.33%)    0.01 (   33.33%)    0.01 (   39.14%)    0.01 (   25.46%)
Elapsed Time     10445.54 (    0.00%) 2249.92 (   78.46%)   70.06 (   99.33%)   16.59 (   99.84%)  472.43 (
95.48%)
+/-               643.98 (    0.00%)  811.62 (  -26.03%)   10.02 (   98.44%)    7.03 (   98.91%)   59.99 (   90.68%)
THP Active         15.60 (    0.00%)   35.20 (  225.64%)   65.00 (  416.67%)   70.80 (  453.85%)   62.20 (  398.72%)
+/-                18.48 (    0.00%)   51.29 (  277.59%)   15.99 (   86.52%)   37.91 (  205.18%)   22.02 (  119.18%)
Fault Alloc       121.80 (    0.00%)   76.60 (   62.89%)  155.40 (  127.59%)  181.20 (  148.77%)  286.60 (  235.30%)
+/-                73.51 (    0.00%)   61.11 (   83.12%)   34.89 (   47.46%)   31.88 (   43.36%)   68.13 (   92.68%)
Fault Fallback    881.20 (    0.00%)  926.60 (   -5.15%)  847.60 (    3.81%)  822.00 (    6.72%)  716.60 (   18.68%)
+/-                73.51 (    0.00%)   61.26 (   16.67%)   34.89 (   52.54%)   31.65 (   56.94%)   67.75 (    7.84%)
MMTests Statistics: duration
User/Sys Time Running Test (seconds)       3540.88   1945.37    716.04     64.97   1937.03
Total Elapsed Time (seconds)              52417.33  11425.90    501.02    230.95   2520.28

The first thing to note is the "Elapsed Time" for the vanilla kernels
of 2249 seconds versus 56 with THP disabled which might explain the
reports of USB stalls with THP enabled. Applying the patches brings
performance in line with THP-disabled performance while isolating
pages for immediate reclaim from the LRU cuts down System CPU time.

The "Fault Alloc" success rate figures are also improved. The vanilla
kernel only managed to allocate 76.6 pages on average over the course
of 5 iterations where as applying the series allocated 181.20 on
average albeit it is well within variance. It's worth noting that
applies the series at least descreases the amount of variance which
implies an improvement.

Andrea's series had a higher success rate for THP allocations but
at a severe cost to elapsed time which is still better than vanilla
but still much worse than disabling THP altogether. One can bring my
series close to Andrea's by removing this check

        /*
         * If compaction is deferred for high-order allocations, it is because
         * sync compaction recently failed. In this is the case and the caller
         * has requested the system not be heavily disrupted, fail the
         * allocation now instead of entering direct reclaim
         */
        if (deferred_compaction && (gfp_mask & __GFP_NO_KSWAPD))
                goto nopage;

I didn't include a patch that removed the above check because hurting
overall performance to improve the THP figure is not what the average
user wants. It's something to consider though if someone really wants
to maximise THP usage no matter what it does to the workload initially.

This is summary of vmstat figures from the same test.

                                       3.1.0-vanilla rc5-vanilla freemore-v6r1 isolate-v6r1 andrea-v2r1
Page Ins                                  3257266139  1111844061    17263623    10901575   161423219
Page Outs                                   81054922    30364312     3626530     3657687     8753730
Swap Ins                                        3294        2851        6560        4964        4592
Swap Outs                                     390073      528094      620197      790912      698285
Direct pages scanned                      1077581700  3024951463  1764930052   115140570  5901188831
Kswapd pages scanned                        34826043     7112868     2131265     1686942     1893966
Kswapd pages reclaimed                      28950067     4911036     1246044      966475     1497726
Direct pages reclaimed                     805148398   280167837     3623473     2215044    40809360
Kswapd efficiency                                83%         69%         58%         57%         79%
Kswapd velocity                              664.399     622.521    4253.852    7304.360     751.490
Direct efficiency                                74%          9%          0%          1%          0%
Direct velocity                            20557.737  264745.137 3522673.849  498551.938 2341481.435
Percentage direct scans                          96%         99%         99%         98%         99%
Page writes by reclaim                        722646      529174      620319      791018      699198
Page writes file                              332573        1080         122         106         913
Page writes anon                              390073      528094      620197      790912      698285
Page reclaim immediate                             0  2552514720  1635858848   111281140  5478375032
Page rescued immediate                             0           0           0       87848           0
Slabs scanned                                  23552       23552        9216        8192        9216
Direct inode steals                              231           0           0           0           0
Kswapd inode steals                                0           0           0           0           0
Kswapd skipped wait                            28076         786           0          61           6
THP fault alloc                                  609         383         753         906        1433
THP collapse alloc                                12           6           0           0           6
THP splits                                       536         211         456         593        1136
THP fault fallback                              4406        4633        4263        4110        3583
THP collapse fail                                120         127           0           0           4
Compaction stalls                               1810         728         623         779        3200
Compaction success                               196          53          60          80         123
Compaction failures                             1614         675         563         699        3077
Compaction pages moved                        193158       53545      243185      333457      226688
Compaction move failure                         9952        9396       16424       23676       45070

The main things to look at are

1. Page In/out figures are much reduced by the series.

2. Direct page scanning is incredibly high (264745.137 pages scanned
   per second on the vanilla kernel) but isolating PageReclaim pages
   on their own list reduces the number of pages scanned significantly.

3. The fact that "Page rescued immediate" is a positive number implies
   that we sometimes race removing pages from the LRU_IMMEDIATE list
   that need to be put back on a normal LRU but it happens only for
   0.07% of the pages marked for immediate reclaim.

writebackCPDeviceext4
                   3.1.0-vanilla         rc5-vanilla       freemore-v6r1        isolate-v6r1         andrea-v2r1
System Time         1.51 (    0.00%)    1.77 (  -17.66%)    1.46 (    2.92%)    1.15 (   23.77%)    1.89 (  -25.63%)
+/-                 0.27 (    0.00%)    0.67 ( -148.52%)    0.33 (  -22.76%)    0.30 (  -11.15%)    0.19 (   30.16%)
User Time           0.03 (    0.00%)    0.04 (  -37.50%)    0.05 (  -62.50%)    0.07 ( -112.50%)    0.04 (  -18.75%)
+/-                 0.01 (    0.00%)    0.02 ( -146.64%)    0.02 (  -97.91%)    0.02 (  -75.59%)    0.02 (  -63.30%)
Elapsed Time      124.93 (    0.00%)  114.49 (    8.36%)   96.77 (   22.55%)   27.48 (   78.00%)  205.70 (  -64.65%)
+/-                20.20 (    0.00%)   74.39 ( -268.34%)   59.88 ( -196.48%)    7.72 (   61.79%)   25.03 (  -23.95%)
THP Active        161.80 (    0.00%)   83.60 (   51.67%)  141.20 (   87.27%)   84.60 (   52.29%)   82.60 (   51.05%)
+/-                71.95 (    0.00%)   43.80 (   60.88%)   26.91 (   37.40%)   59.02 (   82.03%)   52.13 (   72.45%)
Fault Alloc       471.40 (    0.00%)  228.60 (   48.49%)  282.20 (   59.86%)  225.20 (   47.77%)  388.40 (   82.39%)
+/-                88.07 (    0.00%)   87.42 (   99.26%)   73.79 (   83.78%)  109.62 (  124.47%)   82.62 (   93.81%)
Fault Fallback    531.60 (    0.00%)  774.60 (  -45.71%)  720.80 (  -35.59%)  777.80 (  -46.31%)  614.80 (  -15.65%)
+/-                88.07 (    0.00%)   87.26 (    0.92%)   73.79 (   16.22%)  109.62 (  -24.47%)   82.29 (    6.56%)
MMTests Statistics: duration
User/Sys Time Running Test (seconds)         50.22     33.76     30.65     24.14    128.45
Total Elapsed Time (seconds)               1113.73   1132.19   1029.45    759.49   1707.26

Similar test but the USB stick is using ext4 instead of vfat. As
ext4 does not use writepage for migration, the large stalls due to
compaction when THP is enabled are not observed. Still, isolating
PageReclaim pages on their own list helped completion time largely
by reducing the number of pages scanned by direct reclaim although
time spend in congestion_wait could also be a factor.

Again, Andrea's series had far higher success rates for THP allocation
at the cost of elapsed time. I didn't look too closely but a quick
look at the vmstat figures tells me kswapd reclaimed 8 times more pages
than the patch series and direct reclaim reclaimed roughly three times
as many pages. It follows that if memory is aggressively reclaimed,
there will be more available for THP.

writebackCPFilevfat
                   3.1.0-vanilla         rc5-vanilla       freemore-v6r1        isolate-v6r1         andrea-v2r1
System Time         1.76 (    0.00%)   29.10 (-1555.52%)   46.01 (-2517.18%)    4.79 ( -172.35%)   54.89 (-3022.53%)
+/-                 0.14 (    0.00%)   25.61 (-18185.17%)    2.15 (-1434.83%)    6.60 (-4610.03%)    9.75
(-6863.76%)
User Time           0.05 (    0.00%)    0.07 (  -45.83%)    0.05 (   -4.17%)    0.06 (  -29.17%)    0.06 (  -16.67%)
+/-                 0.02 (    0.00%)    0.02 (   20.11%)    0.02 (   -3.14%)    0.01 (   31.58%)    0.01 (   47.41%)
Elapsed Time     22520.79 (    0.00%) 1082.85 (   95.19%)   73.30 (   99.67%)   32.43 (   99.86%)  291.84 (  98.70%)
+/-              7277.23 (    0.00%)  706.29 (   90.29%)   19.05 (   99.74%)   17.05 (   99.77%)  125.55 (   98.27%)
THP Active         83.80 (    0.00%)   12.80 (   15.27%)   15.60 (   18.62%)   13.00 (   15.51%)    0.80 (    0.95%)
+/-                66.81 (    0.00%)   20.19 (   30.22%)    5.92 (    8.86%)   15.06 (   22.54%)    1.17 (    1.75%)
Fault Alloc       171.00 (    0.00%)   67.80 (   39.65%)   97.40 (   56.96%)  125.60 (   73.45%)  133.00 (   77.78%)
+/-                82.91 (    0.00%)   30.69 (   37.02%)   53.91 (   65.02%)   55.05 (   66.40%)   21.19 (   25.56%)
Fault Fallback    832.00 (    0.00%)  935.20 (  -12.40%)  906.00 (   -8.89%)  877.40 (   -5.46%)  870.20 (   -4.59%)
+/-                82.91 (    0.00%)   30.69 (   62.98%)   54.01 (   34.86%)   55.05 (   33.60%)   20.91 (   74.78%)
MMTests Statistics: duration
User/Sys Time Running Test (seconds)       7229.81    928.42    704.52     80.68   1330.76
Total Elapsed Time (seconds)             112849.04   5618.69    571.11    360.54   1664.28

In this case, the test is reading/writing only from filesystems but as
it's vfat, it's slow due to calling writepage during compaction. Little
to observe really - the time to complete the test goes way down
with the series applied and THP allocation success rates go up in
comparison to 3.2-rc5.  The success rates are lower than 3.1.0 but
the elapsed time for that kernel is abysmal so it is not really a
sensible comparison.

As before, Andrea's series allocates more THPs at the cost of overall
performance.

writebackCPFileext4
                   3.1.0-vanilla         rc5-vanilla       freemore-v6r1        isolate-v6r1         andrea-v2r1
System Time         1.51 (    0.00%)    1.77 (  -17.66%)    1.46 (    2.92%)    1.15 (   23.77%)    1.89 (  -25.63%)
+/-                 0.27 (    0.00%)    0.67 ( -148.52%)    0.33 (  -22.76%)    0.30 (  -11.15%)    0.19 (   30.16%)
User Time           0.03 (    0.00%)    0.04 (  -37.50%)    0.05 (  -62.50%)    0.07 ( -112.50%)    0.04 (  -18.75%)
+/-                 0.01 (    0.00%)    0.02 ( -146.64%)    0.02 (  -97.91%)    0.02 (  -75.59%)    0.02 (  -63.30%)
Elapsed Time      124.93 (    0.00%)  114.49 (    8.36%)   96.77 (   22.55%)   27.48 (   78.00%)  205.70 (  -64.65%)
+/-                20.20 (    0.00%)   74.39 ( -268.34%)   59.88 ( -196.48%)    7.72 (   61.79%)   25.03 (  -23.95%)
THP Active        161.80 (    0.00%)   83.60 (   51.67%)  141.20 (   87.27%)   84.60 (   52.29%)   82.60 (   51.05%)
+/-                71.95 (    0.00%)   43.80 (   60.88%)   26.91 (   37.40%)   59.02 (   82.03%)   52.13 (   72.45%)
Fault Alloc       471.40 (    0.00%)  228.60 (   48.49%)  282.20 (   59.86%)  225.20 (   47.77%)  388.40 (   82.39%)
+/-                88.07 (    0.00%)   87.42 (   99.26%)   73.79 (   83.78%)  109.62 (  124.47%)   82.62 (   93.81%)
Fault Fallback    531.60 (    0.00%)  774.60 (  -45.71%)  720.80 (  -35.59%)  777.80 (  -46.31%)  614.80 (  -15.65%)
+/-                88.07 (    0.00%)   87.26 (    0.92%)   73.79 (   16.22%)  109.62 (  -24.47%)   82.29 (    6.56%)
MMTests Statistics: duration
User/Sys Time Running Test (seconds)         50.22     33.76     30.65     24.14    128.45
Total Elapsed Time (seconds)               1113.73   1132.19   1029.45    759.49   1707.26

Same type of story - elapsed times go down. In this case, allocation
success rates are roughtly the same. As before, Andrea's has higher
success rates but takes a lot longer.

Overall the series does reduce latencies and while the tests are
inherency racy as alloc competes with the cp processes, the variability
was included. The THP allocation rates are not as high as they could
be but that is because we would have to be more aggressive about
reclaim and compaction impacting overall performance.

This patch:

Commit 39deaf85 ("mm: compaction: make isolate_lru_page() filter-aware")
noted that compaction does not migrate dirty or writeback pages and that
is was meaningless to pick the page and re-add it to the LRU list.

What was missed during review is that asynchronous migration moves dirty
pages if their ->migratepage callback is migrate_page() because these can
be moved without blocking.  This potentially impacted hugepage allocation
success rates by a factor depending on how many dirty pages are in the
system.

This patch partially reverts 39deaf85 to allow migration to isolate dirty
pages again.  This increases how much compaction disrupts the LRU but that
is addressed later in the series.

Signed-off-by: Mel Gorman <mgorman@suse.de>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Andy Isaacson <adi@hexapodia.org>
Cc: Nai Xia <nai.xia@gmail.com>
Cc: Johannes Weiner <jweiner@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agomm: reduce the amount of work done when updating min_free_kbytes
Mel Gorman [Tue, 10 Jan 2012 23:07:14 +0000 (15:07 -0800)]
mm: reduce the amount of work done when updating min_free_kbytes

commit 938929f14cb595f43cd1a4e63e22d36cab1e4a1f upstream.

Stable note: Fixes https://bugzilla.novell.com/show_bug.cgi?id=726210 .
        Large machines with 1TB or more of RAM take a long time to boot
        without this patch and may spew out soft lockup warnings.

When min_free_kbytes is updated, some pageblocks are marked
MIGRATE_RESERVE.  Ordinarily, this work is unnoticable as it happens early
in boot but on large machines with 1TB of memory, this has been reported
to delay boot times, probably due to the NUMA distances involved.

The bulk of the work is due to calling calling pageblock_is_reserved() an
unnecessary amount of times and accessing far more struct page metadata
than is necessary.  This patch significantly reduces the amount of work
done by setup_zone_migrate_reserve() improving boot times on 1TB machines.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoLinux 3.2.24 v3.2.24
Ben Hutchings [Wed, 25 Jul 2012 03:11:50 +0000 (04:11 +0100)]
Linux 3.2.24

9 years agoHID: add support for 2012 MacBook Pro Retina
Ryan Bourgeois [Tue, 10 Jul 2012 16:43:33 +0000 (09:43 -0700)]
HID: add support for 2012 MacBook Pro Retina

commit b2e6ad7dfe26aac5bf136962d0b11d180b820d44 upstream.

Add support for the 15'' MacBook Pro Retina. The keyboard is
the same as recent models.

The patch needs to be synchronized with the bcm5974 patch for
the trackpad - as usual.

Patch originally written by clipcarl (forums.opensuse.org).

[rydberg@euromail.se: Amended mouse ignore lines]
Signed-off-by: Ryan Bourgeois <bluedragonx@gmail.com>
Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoInput: xpad - add Andamiro Pump It Up pad
Yuri Khan [Thu, 12 Jul 2012 05:12:31 +0000 (22:12 -0700)]
Input: xpad - add Andamiro Pump It Up pad

commit e76b8ee25e034ab601b525abb95cea14aa167ed3 upstream.

I couldn't find the vendor ID in any of the online databases, but this
mat has a Pump It Up logo on the top side of the controller compartment,
and a disclaimer stating that Andamiro will not be liable on the bottom.

Signed-off-by: Yuri Khan <yurivkhan@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoInput: xpad - add signature for Razer Onza Tournament Edition
Ilia Katsnelson [Wed, 11 Jul 2012 07:54:20 +0000 (00:54 -0700)]
Input: xpad - add signature for Razer Onza Tournament Edition

commit cc71a7e899cc6b2ff41e1be48756782ed004d802 upstream.

Signed-off-by: Ilia Katsnelson <k0009000@gmail.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoInput: xpad - handle all variations of Mad Catz Beat Pad
Yuri Khan [Wed, 11 Jul 2012 07:49:18 +0000 (00:49 -0700)]
Input: xpad - handle all variations of Mad Catz Beat Pad

commit 3ffb62cb9ac2430c2504c6ff9727d0f2476ef0bd upstream.

The device should be handled by xpad driver instead of generic HID driver.

Signed-off-by: Yuri Khan <yurivkhan@gmail.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoInput: bcm5974 - Add support for 2012 MacBook Pro Retina
Henrik Rydberg [Tue, 10 Jul 2012 16:43:57 +0000 (09:43 -0700)]
Input: bcm5974 - Add support for 2012 MacBook Pro Retina

commit 3dde22a98e94eb18527f0ff0068fb2fb945e58d4 upstream.

Add support for the 15'' MacBook Pro Retina model (MacBookPro10,1).

Patch originally written by clipcarl (forums.opensuse.org).

Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agobonding: Manage /proc/net/bonding/ entries from the netdev events
Eric W. Biederman [Mon, 9 Jul 2012 10:51:45 +0000 (10:51 +0000)]
bonding: Manage /proc/net/bonding/ entries from the netdev events

commit a64d49c3dd504b685f9742a2f3dcb11fb8e4345f upstream.

It was recently reported that moving a bonding device between network
namespaces causes warnings from /proc.  It turns out after the move we
were trying to add and to remove the /proc/net/bonding entries from the
wrong network namespace.

Move the bonding /proc registration code into the NETDEV_REGISTER and
NETDEV_UNREGISTER events where the proc registration and unregistration
will always happen at the right time.

Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agobonding: debugfs and network namespaces are incompatible
Eric W. Biederman [Mon, 9 Jul 2012 10:52:43 +0000 (10:52 +0000)]
bonding: debugfs and network namespaces are incompatible

commit 96ca7ffe748bf91f851e6aa4479aa11c8b1122ba upstream.

The bonding debugfs support has been broken in the presence of network
namespaces since it has been added.  The debugfs support does not handle
multiple bonding devices with the same name in different network
namespaces.

I haven't had any bug reports, and I'm not interested in getting any.
Disable the debugfs support when network namespaces are enabled.

Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agostmmac: Fix for nfs hang on multiple reboot
Deepak Sikri [Sun, 8 Jul 2012 21:14:45 +0000 (21:14 +0000)]
stmmac: Fix for nfs hang on multiple reboot

commit 8e83989106562326bfd6aaf92174fe138efd026b upstream.

It was observed that during multiple reboots nfs hangs. The status of
receive descriptors shows that all the descriptors were in control of
CPU, and none were assigned to DMA.
Also the DMA status register confirmed that the Rx buffer is
unavailable.

This patch adds the fix for the same by adding the memory barriers to
ascertain that the all instructions before enabling the Rx or Tx DMA are
completed which involves the proper setting of the ownership bit in DMA
descriptors.

Signed-off-by: Deepak Sikri <deepak.sikri@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoipheth: add support for iPad
Davide Gerhard [Mon, 25 Jun 2012 07:04:47 +0000 (09:04 +0200)]
ipheth: add support for iPad

commit 6de0298ec9c1edaf330b71b57346241ece8f3346 upstream.

This adds support for the iPad to the ipheth driver.
(product id = 0x129a)

Signed-off-by: Davide Gerhard <rainbow@irh.it>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoACPI / PM: Make acpi_pm_device_sleep_state() follow the specification
Rafael J. Wysocki [Tue, 29 May 2012 19:21:07 +0000 (21:21 +0200)]
ACPI / PM: Make acpi_pm_device_sleep_state() follow the specification

commit dbe9a2edd17d843d80faf2b99f20a691c1853418 upstream.

The comparison between the system sleep state being entered
and the lowest system sleep state the given device may wake up
from in acpi_pm_device_sleep_state() is reversed, because the
specification (ACPI 5.0) says that for wakeup to work:

"The sleeping state being entered must be less than or equal to the
 power state declared in element 1 of the _PRW object."

In other words, the state returned by _PRW is the deepest
(lowest-power) system sleep state the device is capable of waking up
the system from.

Moreover, acpi_pm_device_sleep_state() also should check if the
wakeup capability is supported through ACPI, because in principle it
may be done via native PCIe PME, for example, in which case _SxW
should not be evaluated.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoeCryptfs: Properly check for O_RDONLY flag before doing privileged open
Tyler Hicks [Tue, 12 Jun 2012 18:17:01 +0000 (11:17 -0700)]
eCryptfs: Properly check for O_RDONLY flag before doing privileged open

commit 9fe79d7600497ed8a95c3981cbe5b73ab98222f0 upstream.

If the first attempt at opening the lower file read/write fails,
eCryptfs will retry using a privileged kthread. However, the privileged
retry should not happen if the lower file's inode is read-only because a
read/write open will still be unsuccessful.

The check for determining if the open should be retried was intended to
be based on the access mode of the lower file's open flags being
O_RDONLY, but the check was incorrectly performed. This would cause the
open to be retried by the privileged kthread, resulting in a second
failed open of the lower file. This patch corrects the check to
determine if the open request should be handled by the privileged
kthread.

Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoeCryptfs: Fix lockdep warning in miscdev operations
Tyler Hicks [Mon, 11 Jun 2012 17:21:34 +0000 (10:21 -0700)]
eCryptfs: Fix lockdep warning in miscdev operations

commit 60d65f1f07a7d81d3eb3b91fc13fca80f2fdbb12 upstream.

Don't grab the daemon mutex while holding the message context mutex.
Addresses this lockdep warning:

 ecryptfsd/2141 is trying to acquire lock:
  (&ecryptfs_msg_ctx_arr[i].mux){+.+.+.}, at: [<ffffffffa029c213>] ecryptfs_miscdev_read+0x143/0x470 [ecryptfs]

 but task is already holding lock:
  (&(*daemon)->mux){+.+...}, at: [<ffffffffa029c2ec>] ecryptfs_miscdev_read+0x21c/0x470 [ecryptfs]

 which lock already depends on the new lock.

 the existing dependency chain (in reverse order) is:

 -> #1 (&(*daemon)->mux){+.+...}:
        [<ffffffff810a3b8d>] lock_acquire+0x9d/0x220
        [<ffffffff8151c6da>] __mutex_lock_common+0x5a/0x4b0
        [<ffffffff8151cc64>] mutex_lock_nested+0x44/0x50
        [<ffffffffa029c5d7>] ecryptfs_send_miscdev+0x97/0x120 [ecryptfs]
        [<ffffffffa029b744>] ecryptfs_send_message+0x134/0x1e0 [ecryptfs]
        [<ffffffffa029a24e>] ecryptfs_generate_key_packet_set+0x2fe/0xa80 [ecryptfs]
        [<ffffffffa02960f8>] ecryptfs_write_metadata+0x108/0x250 [ecryptfs]
        [<ffffffffa0290f80>] ecryptfs_create+0x130/0x250 [ecryptfs]
        [<ffffffff811963a4>] vfs_create+0xb4/0x120
        [<ffffffff81197865>] do_last+0x8c5/0xa10
        [<ffffffff811998f9>] path_openat+0xd9/0x460
        [<ffffffff81199da2>] do_filp_open+0x42/0xa0
        [<ffffffff81187998>] do_sys_open+0xf8/0x1d0
        [<ffffffff81187a91>] sys_open+0x21/0x30
        [<ffffffff81527d69>] system_call_fastpath+0x16/0x1b

 -> #0 (&ecryptfs_msg_ctx_arr[i].mux){+.+.+.}:
        [<ffffffff810a3418>] __lock_acquire+0x1bf8/0x1c50
        [<ffffffff810a3b8d>] lock_acquire+0x9d/0x220
        [<ffffffff8151c6da>] __mutex_lock_common+0x5a/0x4b0
        [<ffffffff8151cc64>] mutex_lock_nested+0x44/0x50
        [<ffffffffa029c213>] ecryptfs_miscdev_read+0x143/0x470 [ecryptfs]
        [<ffffffff811887d3>] vfs_read+0xb3/0x180
        [<ffffffff811888ed>] sys_read+0x4d/0x90
        [<ffffffff81527d69>] system_call_fastpath+0x16/0x1b

Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoeCryptfs: Gracefully refuse miscdev file ops on inherited/passed files
Tyler Hicks [Mon, 11 Jun 2012 16:24:11 +0000 (09:24 -0700)]
eCryptfs: Gracefully refuse miscdev file ops on inherited/passed files

commit 8dc6780587c99286c0d3de747a2946a76989414a upstream.

File operations on /dev/ecryptfs would BUG() when the operations were
performed by processes other than the process that originally opened the
file. This could happen with open files inherited after fork() or file
descriptors passed through IPC mechanisms. Rather than calling BUG(), an
error code can be safely returned in most situations.

In ecryptfs_miscdev_release(), eCryptfs still needs to handle the
release even if the last file reference is being held by a process that
didn't originally open the file. ecryptfs_find_daemon_by_euid() will not
be successful, so a pointer to the daemon is stored in the file's
private_data. The private_data pointer is initialized when the miscdev
file is opened and only used when the file is released.

https://launchpad.net/bugs/994247

Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Reported-by: Sasha Levin <levinsasha928@gmail.com>
Tested-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoACPI sysfs.c strlen fix
Pavel Vasilyev [Tue, 5 Jun 2012 04:02:05 +0000 (00:02 -0400)]
ACPI sysfs.c strlen fix

commit 9f132652d94c96476b0b0a8caf0c10e96ab10fa8 upstream.

Current code is ignoring the last character of "enable" and "disable"
in comparisons.

https://bugzilla.kernel.org/show_bug.cgi?id=33732

Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoACPI, x86: fix Dell M6600 ACPI reboot regression via DMI
Zhang Rui [Mon, 20 Feb 2012 06:20:06 +0000 (14:20 +0800)]
ACPI, x86: fix Dell M6600 ACPI reboot regression via DMI

commit 76eb9a30db4bc8fd172f9155247264b5f2686d7b upstream.

Dell Precision M6600 is known to require PCI reboot, so add it to
the reboot blacklist in pci_reboot_dmi_table[].

https://bugzilla.kernel.org/show_bug.cgi?id=42749

cc: x86@kernel.org
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoACPI: Add a quirk for "AMILO PRO V2030" to ignore the timer overriding
Feng Tang [Mon, 4 Jun 2012 07:00:06 +0000 (15:00 +0800)]
ACPI: Add a quirk for "AMILO PRO V2030" to ignore the timer overriding

commit b939c2acf1dc42b08407ef5174f2e8d6f43dd5ea upstream.

commit f6b54f083cc66cf9b11d2120d8df3c2ad4e0836d upstream.

This is the 2nd part of fix for kernel bugzilla 40002:
    "IRQ 0 assigned to VGA"
https://bugzilla.kernel.org/show_bug.cgi?id=40002

The root cause is the buggy FW, whose ACPI tables assign the GSI 16
to 2 irqs 0 and 16(VGA), and the VGA is the right owner of GSI 16.
So add a quirk to ignore the irq0 overriding GSI 16 for the
FUJITSU SIEMENS AMILO PRO V2030 platform will solve this issue.

Reported-and-tested-by: Szymon Kowalczyk <fazerxlo@o2.pl>
Signed-off-by: Feng Tang <feng.tang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoACPI: Remove one board specific WARN when ignoring timer overriding
Feng Tang [Mon, 4 Jun 2012 07:00:05 +0000 (15:00 +0800)]
ACPI: Remove one board specific WARN when ignoring timer overriding

commit 5752cdb805ff89942d99d12118e2844e7db34df8 upstream.

commit 7f68b4c2e158019c2ec494b5cfbd9c83b4e5b253 upstream.

Current WARN msg is only for the ati_ixp4x0 board, while this function
is used by mulitple platforms. So this one board specific warning
is not appropriate any more.

Signed-off-by: Feng Tang <feng.tang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agoACPI: Make acpi_skip_timer_override cover all source_irq==0 cases
Feng Tang [Mon, 4 Jun 2012 07:00:04 +0000 (15:00 +0800)]
ACPI: Make acpi_skip_timer_override cover all source_irq==0 cases

commit ae10ccdc3093486f8c2369d227583f9d79f628e5 upstream.

Currently when acpi_skip_timer_override is set, it only cover the
(source_irq == 0 && global_irq == 2) cases. While there is also
platform which need use this option and its global_irq is not 2.
This patch will extend acpi_skip_timer_override to cover all
timer overriding cases as long as the source irq is 0.

This is the first part of a fix to kernel bug bugzilla 40002:
"IRQ 0 assigned to VGA"
https://bugzilla.kernel.org/show_bug.cgi?id=40002

Reported-and-tested-by: Szymon Kowalczyk <fazerxlo@o2.pl>
Signed-off-by: Feng Tang <feng.tang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agonet: remove skb_orphan_try()
Eric Dumazet [Thu, 14 Jun 2012 06:42:44 +0000 (06:42 +0000)]
net: remove skb_orphan_try()

commit 62b1a8ab9b3660bb820d8dfe23148ed6cda38574 upstream.

Orphaning skb in dev_hard_start_xmit() makes bonding behavior
unfriendly for applications sending big UDP bursts : Once packets
pass the bonding device and come to real device, they might hit a full
qdisc and be dropped. Without orphaning, the sender is automatically
throttled because sk->sk_wmemalloc reaches sk->sk_sndbuf (assuming
sk_sndbuf is not too big)

We could try to defer the orphaning adding another test in
dev_hard_start_xmit(), but all this seems of little gain,
now that BQL tends to make packets more likely to be parked
in Qdisc queues instead of NIC TX ring, in cases where performance
matters.

Reverts commits :
fc6055a5ba31 net: Introduce skb_orphan_try()
87fd308cfc6b net: skb_tx_hash() fix relative to skb_orphan_try()
and removes SKBTX_DRV_NEEDS_SK_REF flag

Reported-and-bisected-by: Jean-Michel Hautbois <jhautbois@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Tested-by: Oliver Hartkopp <socketcan@hartkopp.net>
Acked-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
[bwh: Backported to 3.2:
 - Adjust context
 - SKBTX_WIFI_STATUS is not defined]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agobnx2x: fix panic when TX ring is full
Eric Dumazet [Wed, 13 Jun 2012 09:45:16 +0000 (09:45 +0000)]
bnx2x: fix panic when TX ring is full

commit bc14786a100cc6a81cd060e8031ec481241b418c upstream.

There is a off by one error in the minimal number of BD in
bnx2x_start_xmit() and bnx2x_tx_int() before stopping/resuming tx queue.

A full size GSO packet, with data included in skb->head really needs
(MAX_SKB_FRAGS + 4) BDs, because of bnx2x_tx_split()

This error triggers if BQL is disabled and heavy TCP transmit traffic
occurs.

bnx2x_tx_split() definitely can be called, remove a wrong comment.

Reported-by: Tomas Hruby <thruby@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Eilon Greenstein <eilong@broadcom.com>
Cc: Yaniv Rosner <yanivr@broadcom.com>
Cc: Merav Sicron <meravs@broadcom.com>
Cc: Tom Herbert <therbert@google.com>
Cc: Robert Evans <evansr@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agobnx2x: fix checksum validation
Eric Dumazet [Tue, 12 Jun 2012 23:50:04 +0000 (23:50 +0000)]
bnx2x: fix checksum validation

commit d6cb3e41386f20fb0777d0b59a2def82c65d37f7 upstream.

bnx2x driver incorrectly sets ip_summed to CHECKSUM_UNNECESSARY on
encapsulated segments. TCP stack happily accepts frames with bad
checksums, if they are inside a GRE or IPIP encapsulation.

Our understanding is that if no IP or L4 csum validation was done by the
hardware, we should leave ip_summed as is (CHECKSUM_NONE), since
hardware doesn't provide CHECKSUM_COMPLETE support in its cqe.

Then, if IP/L4 checksumming was done by the hardware, set
CHECKSUM_UNNECESSARY if no error was flagged.

Patch based on findings and analysis from Robert Evans

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Eilon Greenstein <eilong@broadcom.com>
Cc: Yaniv Rosner <yanivr@broadcom.com>
Cc: Merav Sicron <meravs@broadcom.com>
Cc: Tom Herbert <therbert@google.com>
Cc: Robert Evans <evansr@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
[bwh: Backported to 3.2: adjust context, indentation]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agor8169: call netif_napi_del at errpaths and at driver unload
Devendra Naga [Thu, 31 May 2012 01:51:20 +0000 (01:51 +0000)]
r8169: call netif_napi_del at errpaths and at driver unload

commit ad1be8d345416a794dea39761a374032aa471a76 upstream.

when register_netdev fails, the init'ed NAPIs by netif_napi_add must be
deleted with netif_napi_del, and also when driver unloads, it should
delete the NAPI before unregistering netdevice using unregister_netdev.

Signed-off-by: Devendra Naga <devendra.aaru@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agovhost: don't forget to schedule()
Nadav Har'El [Mon, 27 Feb 2012 13:07:29 +0000 (15:07 +0200)]
vhost: don't forget to schedule()

commit d550dda192c1bd039afb774b99485e88b70d7cb8 upstream.

This is a tiny, but important, patch to vhost.

Vhost's worker thread only called schedule() when it had no work to do, and
it wanted to go to sleep. But if there's always work to do, e.g., the guest
is running a network-intensive program like netperf with small message sizes,
schedule() was *never* called. This had several negative implications (on
non-preemptive kernels):

 1. Passing time was not properly accounted to the "vhost" process (ps and
    top would wrongly show it using zero CPU time).

 2. Sometimes error messages about RCU timeouts would be printed, if the
    core running the vhost thread didn't schedule() for a very long time.

 3. Worst of all, a vhost thread would "hog" the core. If several vhost
    threads need to share the same core, typically one would get most of the
    CPU time (and its associated guest most of the performance), while the
    others hardly get any work done.

The trivial solution is to add

if (need_resched())
schedule();

After doing every piece of work. This will not do the heavy schedule() all
the time, just when the timer interrupt decided a reschedule is warranted
(so need_resched returns true).

Thanks to Abel Gordon for this patch.

Signed-off-by: Nadav Har'El <nyh@il.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
9 years agopowerpc: Fix wrong divisor in usecs_to_cputime
Andreas Schwab [Fri, 9 Dec 2011 11:35:08 +0000 (11:35 +0000)]
powerpc: Fix wrong divisor in usecs_to_cputime

commit 9f5072d4f63f28d30d343573830ac6c85fc0deff upstream.

Commit d57af9b (taskstats: use real microsecond granularity for CPU times)
renamed msecs_to_cputime to usecs_to_cputime, but failed to update all
numbers on the way.  This causes nonsensical cpu idle/iowait values to be
displayed in /proc/stat (the only user of usecs_to_cputime so far).

This also renames __cputime_msec_factor to __cputime_usec_factor, adapting
its value and using it directly in cputime_to_usecs instead of doing two
multiplications.

Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Acked-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>