GitHub/LineageOS/android_kernel_motorola_exynos9610.git
16 years agomd: replace R5_WantPrexor with R5_WantDrain, add 'prexor' reconstruct_states
Dan Williams [Fri, 27 Jun 2008 22:32:06 +0000 (08:32 +1000)]
md: replace R5_WantPrexor with R5_WantDrain, add 'prexor' reconstruct_states

From: Dan Williams <dan.j.williams@intel.com>

Currently ops_run_biodrain and other locations have extra logic to determine
which blocks are processed in the prexor and non-prexor cases.  This can be
eliminated if handle_write_operations5 flags the blocks to be processed in all
cases via R5_Wantdrain.  The presence of the prexor operation is tracked in
sh->reconstruct_state.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agomd: replace STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} with 'reconstruct_states'
Dan Williams [Fri, 27 Jun 2008 22:32:05 +0000 (08:32 +1000)]
md: replace STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} with 'reconstruct_states'

From: Dan Williams <dan.j.williams@intel.com>

Track the state of reconstruct operations (recalculating the parity block
usually due to incoming writes, or as part of array expansion)  Reduces the
scope of the STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} flags to only tracking whether
a reconstruct operation has been requested via the ops_request field of struct
stripe_head_state.

This is the final step in the removal of ops.{pending,ack,complete,count}, i.e.
the STRIPE_OP_{BIODRAIN,PREXOR,POSTXOR} flags only request an operation and do
not track the state of the operation.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agomd: replace STRIPE_OP_COMPUTE_BLK with STRIPE_COMPUTE_RUN
Dan Williams [Fri, 27 Jun 2008 22:32:03 +0000 (08:32 +1000)]
md: replace STRIPE_OP_COMPUTE_BLK with STRIPE_COMPUTE_RUN

From: Dan Williams <dan.j.williams@intel.com>

Track the state of compute operations (recalculating a block from all the other
blocks in a stripe) with a state flag.  Reduces the scope of the
STRIPE_OP_COMPUTE_BLK flag to only tracking whether a compute operation has
been requested via the ops_request field of struct stripe_head_state.

Note, the compute operation that is performed in the course of doing a 'repair'
operation (check the parity block, recalculate it and write it back if the
check result is not zero) is tracked separately with the 'check_state'
variable.  Compute operations are held off while a 'check' is in progress, and
moving this check out to handle_issuing_new_read_requests5 the helper routine
__handle_issuing_new_read_requests5 can be simplified.

This is another step towards the removal of ops.{pending,ack,complete,count},
i.e. STRIPE_OP_COMPUTE_BLK only requests an operation and does not track the
state of the operation.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agomd: replace STRIPE_OP_BIOFILL with STRIPE_BIOFILL_RUN
Dan Williams [Fri, 27 Jun 2008 22:31:58 +0000 (08:31 +1000)]
md: replace STRIPE_OP_BIOFILL with STRIPE_BIOFILL_RUN

From: Dan Williams <dan.j.williams@intel.com>

Track the state of read operations (copying data from the stripe cache to bio
buffers outside the lock) with a state flag.  Reduce the scope of the
STRIPE_OP_BIOFILL flag to only tracking whether a biofill operation has been
requested via the ops_request field of struct stripe_head_state.

This is another step towards the removal of ops.{pending,ack,complete,count},
i.e. STRIPE_OP_BIOFILL only requests an operation and does not track the state
of the operation.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agomd: replace STRIPE_OP_CHECK with 'check_states'
Dan Williams [Fri, 27 Jun 2008 22:31:57 +0000 (08:31 +1000)]
md: replace STRIPE_OP_CHECK with 'check_states'

From: Dan Williams <dan.j.williams@intel.com>

The STRIPE_OP_* flags record the state of stripe operations which are
performed outside the stripe lock.  Their use in indicating which
operations need to be run is straightforward; however, interpolating what
the next state of the stripe should be based on a given combination of
these flags is not straightforward, and has led to bugs.  An easier to read
implementation with minimal degrees of freedom is needed.

Towards this goal, this patch introduces explicit states to replace what was
previously interpolated from the STRIPE_OP_* flags.  For now this only converts
the handle_parity_checks5 path, removing a user of the
ops.{pending,ack,complete,count} fields of struct stripe_operations.

This conversion also found a remaining issue with the current code.  There is
a small window for a drive to fail between when we schedule a repair and when
the parity calculation for that repair completes.  When this happens we will
writeback to 'failed_num' when we really want to write back to 'pd_idx'.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agomd: unify raid5/6 i/o submission
Dan Williams [Fri, 27 Jun 2008 22:31:55 +0000 (08:31 +1000)]
md: unify raid5/6 i/o submission

From: Dan Williams <dan.j.williams@intel.com>

Let the raid6 path call ops_run_io to get pending i/o submitted.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agomd: use stripe_head_state in ops_run_io()
Dan Williams [Fri, 27 Jun 2008 22:31:53 +0000 (08:31 +1000)]
md: use stripe_head_state in ops_run_io()

From: Dan Williams <dan.j.williams@intel.com>

In handle_stripe after taking sh->lock we sample some bits into 's' (struct
stripe_head_state):

s.syncing = test_bit(STRIPE_SYNCING, &sh->state);
s.expanding = test_bit(STRIPE_EXPAND_SOURCE, &sh->state);
s.expanded = test_bit(STRIPE_EXPAND_READY, &sh->state);

Use these values from 's' in ops_run_io() rather than re-sampling the bits.
This ensures a consistent snapshot (as seen under sh->lock) is used.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agomd: kill STRIPE_OP_IO flag
Dan Williams [Fri, 27 Jun 2008 22:31:52 +0000 (08:31 +1000)]
md: kill STRIPE_OP_IO flag

From: Dan Williams <dan.j.williams@intel.com>

The R5_Want{Read,Write} flags already gate i/o.  So, this flag is
superfluous and we can unconditionally call ops_run_io().

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agomd: kill STRIPE_OP_MOD_DMA in raid5 offload
Dan Williams [Fri, 27 Jun 2008 22:31:50 +0000 (08:31 +1000)]
md: kill STRIPE_OP_MOD_DMA in raid5 offload

From: Dan Williams <dan.j.williams@intel.com>

This micro-optimization allowed the raid code to skip a re-read of the
parity block after checking parity.  It took advantage of the fact that
xor-offload-engines have their own internal result buffer and can check
parity without writing to memory.  Remove it for the following reasons:

1/ It is a layering violation for MD to need to manage the DMA and
   non-DMA paths within async_xor_zero_sum
2/ Bad precedent to toggle the 'ops' flags outside the lock
3/ Hard to realize a performance gain as reads will not need an updated
   parity block and writes will dirty it anyways.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoSupport changing rdev size on running arrays.
Chris Webb [Fri, 27 Jun 2008 22:31:46 +0000 (08:31 +1000)]
Support changing rdev size on running arrays.

From: Chris Webb <chris@arachsys.com>

Allow /sys/block/mdX/md/rdY/size to change on running arrays, moving the
superblock if necessary for this metadata version. We prevent the available
space from shrinking to less than the used size, and allow it to be set to zero
to fill all the available space on the underlying device.

Signed-off-by: Chris Webb <chris@arachsys.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoMake sure all changes to md/dev-XX/state are notified
Neil Brown [Fri, 27 Jun 2008 22:31:44 +0000 (08:31 +1000)]
Make sure all changes to md/dev-XX/state are notified

The important state change happens during an interrupt
in md_error.  So just set a flag there and call sysfs_notify
later in process context.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoMake sure all changes to md/degraded are notified.
Neil Brown [Fri, 27 Jun 2008 22:31:43 +0000 (08:31 +1000)]
Make sure all changes to md/degraded are notified.

When a device fails, when a spare is activated, when
an array is reshaped, or when an array is started,
the extent to which the array is degraded can change.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoMake sure all changes to md/sync_action are notified.
Neil Brown [Fri, 27 Jun 2008 22:31:41 +0000 (08:31 +1000)]
Make sure all changes to md/sync_action are notified.

When the 'resync' thread starts or stops, when we explicitly
set sync_action, or when we determine that there is definitely nothing
to do, we notify sync_action.

To stop "sync_action" from occasionally showing the wrong value,
we introduce a new flags - MD_RECOVERY_RECOVER - to say that a
recovery is probably needed or happening, and we make sure
that we set MD_RECOVERY_RUNNING before clearing MD_RECOVERY_NEEDED.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoMake sure all changes to md/array_state are notified.
Neil Brown [Fri, 27 Jun 2008 22:31:36 +0000 (08:31 +1000)]
Make sure all changes to md/array_state are notified.

Changes in md/array_state could be of interest to a monitoring
program.  So make sure all changes trigger a notification.

Exceptions:
   changing active_idle to active is not reported because it
      is frequent and not interesting.
   changing active to active_idle is only reported on arrays
      with externally managed metadata, as it is not interesting
      otherwise.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoDon't reject HOT_REMOVE_DISK request for an array that is not yet started.
Neil Brown [Fri, 27 Jun 2008 22:31:34 +0000 (08:31 +1000)]
Don't reject HOT_REMOVE_DISK request for an array that is not yet started.

There is really no need for this test here, and there are valid
cases for selectively removing devices from an array that
it not actually active.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agorationalise return value for ->hot_add_disk method.
Neil Brown [Fri, 27 Jun 2008 22:31:33 +0000 (08:31 +1000)]
rationalise return value for ->hot_add_disk method.

For all array types but linear, ->hot_add_disk returns 1 on
success, 0 on failure.
For linear, it returns 0 on success and -errno on failure.

This doesn't cause a functional problem because the ->hot_add_disk
function of linear is used quite differently to the others.
However it is confusing.

So convert all to return 0 for success or -errno on failure
and fix call sites to match.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoSupport adding a spare to a live md array with external metadata.
Neil Brown [Fri, 27 Jun 2008 22:31:31 +0000 (08:31 +1000)]
Support adding a spare to a live md array with external metadata.

i.e. extend the 'md/dev-XXX/slot' attribute so that you can
tell a device to fill an vacant slot in an and md array.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoEnable setting of 'offset' and 'size' of a hot-added spare.
Neil Brown [Fri, 27 Jun 2008 22:31:29 +0000 (08:31 +1000)]
Enable setting of 'offset' and 'size' of a hot-added spare.

offset_store and rdev_size_store allow control of the region of a
device which is to be using in an md/raid array.
They only allow these values to be set when an array is being assembled,
as changing them on an active array could be dangerous.
However when adding a spare device to an array, we might need to
set the offset and size before starting recovery.  So allow
these values to be set also if "->raid_disk < 0" which indicates that
the device is still a spare.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoDon't try to make md arrays dirty if that is not meaningful.
Neil Brown [Fri, 27 Jun 2008 22:31:27 +0000 (08:31 +1000)]
Don't try to make md arrays dirty if that is not meaningful.

Arrays personalities such as 'raid0' and 'linear' have no redundancy,
and so marking them as 'clean' or 'dirty' is not meaningful.
So always allow write requests without requiring a superblock update.

Such arrays types are detected by ->sync_request being NULL.  If it is
not possible to send a sync request we don't need a 'dirty' flag because
all a dirty flag does is trigger some sync_requests.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoClose race in md_probe
Neil Brown [Fri, 27 Jun 2008 22:31:26 +0000 (08:31 +1000)]
Close race in md_probe

There is a possible race in md_probe.  If two threads call md_probe
for the same device, then one could exit (having checked that
->gendisk exists) before the other has called kobject_init_and_add,
thus returning an incomplete kobj which will cause problems when
we try to add children to it.

So extend the range of protection of disks_mutex slightly to
avoid this possibility.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoAllow setting start point for requested check/repair
Neil Brown [Fri, 27 Jun 2008 22:31:24 +0000 (08:31 +1000)]
Allow setting start point for requested check/repair

This makes it possible to just resync a small part of an array.
e.g. if a drive reports that it has questionable sectors,
a 'repair' of just the region covering those sectors will
cause them to be read and, if there is an error, re-written
with correct data.

Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoImprove setting of "events_cleared" for write-intent bitmaps.
Neil Brown [Fri, 27 Jun 2008 22:31:22 +0000 (08:31 +1000)]
Improve setting of "events_cleared" for write-intent bitmaps.

When an array is degraded, bits in the write-intent bitmap are not
cleared, so that if the missing device is re-added, it can be synced
by only updated those parts of the device that have changed since
it was removed.

The enable this a 'events_cleared' value is stored. It is the event
counter for the array the last time that any bits were cleared.

Sometimes - if a device disappears from an array while it is 'clean' -
the events_cleared value gets updated incorrectly (there are subtle
ordering issues between updateing events in the main metadata and the
bitmap metadata) resulting in the missing device appearing to require
a full resync when it is re-added.

With this patch, we update events_cleared precisely when we are about
to clear a bit in the bitmap.  We record events_cleared when we clear
the bit internally, and copy that to the superblock which is written
out before the bit on storage.  This makes it more "obviously correct".

We also need to update events_cleared when the event_count is going
backwards (as happens on a dirty->clean transition of a non-degraded
array).

Thanks to Mike Snitzer for identifying this problem and testing early
"fixes".

Cc: "Mike Snitzer" <snitzer@gmail.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agouse bio_endio instead of a call to bi_end_io
Neil Brown [Fri, 27 Jun 2008 22:31:20 +0000 (08:31 +1000)]
use bio_endio instead of a call to bi_end_io

Turn calls to bi->bi_end_io() into bio_endio(). Apparently bio_endio does
exactly the same error processing as is hardcoded at these places.

bio_endio() avoids recursion (or will soon), so it should be used.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agolinear: correct disk numbering error check
Nikanth Karthikesan [Fri, 27 Jun 2008 22:31:19 +0000 (08:31 +1000)]
linear: correct disk numbering error check

From: "Nikanth Karthikesan" <knikanth@novell.com>

Correct disk numbering problem check.

Signed-off-by: Nikanth Karthikesan <knikanth@suse.de>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoFix error paths if md_probe fails.
Neil Brown [Fri, 27 Jun 2008 22:31:17 +0000 (08:31 +1000)]
Fix error paths if md_probe fails.

md_probe can fail (e.g. alloc_disk could fail) without
returning an error (as it alway returns NULL).
So when we call mddev_find immediately afterwards, we need
to check that md_probe actually succeeded.  This means checking
that mdev->gendisk is non-NULL.

cc: <stable@kernel.org>
Cc: Dave Jones <davej@redhat.com>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoDon't acknowlege that stripe-expand is complete until it really is.
Neil Brown [Fri, 27 Jun 2008 22:31:14 +0000 (08:31 +1000)]
Don't acknowlege that stripe-expand is complete until it really is.

We shouldn't acknowledge that a stripe has been expanded (When
reshaping a raid5 by adding a device) until the moved data has
actually been written out.  However we are currently
acknowledging (by calling md_done_sync) when the POST_XOR
is complete and before the write.

So track in s.locked whether there are pending writes, and don't
call md_done_sync yet if there are.

Note: we all set R5_LOCKED on devices which are are about to
read from.  This probably isn't technically necessary, but is
usually done when writing a block, and justifies the use of
s.locked here.

This bug can lead to a crash if an array is stopped while an reshape
is in progress.

Cc: <stable@kernel.org>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoEnsure interrupted recovery completed properly (v1 metadata plus bitmap)
Neil Brown [Fri, 27 Jun 2008 22:30:52 +0000 (08:30 +1000)]
Ensure interrupted recovery completed properly (v1 metadata plus bitmap)

If, while assembling an array, we find a device which is not fully
in-sync with the array, it is important to set the "fullsync" flags.
This is an exact analog to the setting of this flag in hot_add_disk
methods.

Currently, only v1.x metadata supports having devices in an array
which are not fully in-sync (it keep track of how in sync they are).
The 'fullsync' flag only makes a difference when a write-intent bitmap
is being used.  In this case it tells recovery to ignore the bitmap
and recovery all blocks.

This fix is already in place for raid1, but not raid5/6 or raid10.

So without this fix, a raid1 ir raid4/5/6 array with version 1.x
metadata and a write intent bitmaps, that is stopped in the middle
of a recovery, will appear to complete the recovery instantly
after it is reassembled, but the recovery will not be correct.

If you might have an array like that, issueing
   echo repair > /sys/block/mdXX/md/sync_action

will make sure recovery completes properly.

Cc: <stable@kernel.org>
Signed-off-by: Neil Brown <neilb@suse.de>
16 years agoLinux 2.6.26-rc8
Linus Torvalds [Wed, 25 Jun 2008 01:58:20 +0000 (18:58 -0700)]
Linux 2.6.26-rc8

16 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Wed, 25 Jun 2008 01:12:33 +0000 (18:12 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  [IA64] Eliminate NULL test after alloc_bootmem in iosapic_alloc_rte()
  [IA64] Handle count==0 in sn2_ptc_proc_write()
  [IA64] Fix boot failure on ia64/sn2

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-fixes
Linus Torvalds [Wed, 25 Jun 2008 01:09:47 +0000 (18:09 -0700)]
Merge git://git./linux/kernel/git/steve/gfs2-2.6-fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-fixes:
  [GFS2] fix gfs2 block allocation (cleaned up)
  [GFS2] BUG: unable to handle kernel paging request at ffff81002690e000

16 years agoMerge branch 'kvm-updates-2.6.26' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 25 Jun 2008 01:09:06 +0000 (18:09 -0700)]
Merge branch 'kvm-updates-2.6.26' of git://git./linux/kernel/git/avi/kvm

* 'kvm-updates-2.6.26' of git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm:
  KVM: Remove now unused structs from kvm_para.h
  x86: KVM guest: Use the paravirt clocksource structs and functions
  KVM: Make kvm host use the paravirt clocksource structs
  x86: Make xen use the paravirt clocksource structs and functions
  x86: Add structs and functions for paravirt clocksource
  KVM: VMX: Fix host msr corruption with preemption enabled
  KVM: ioapic: fix lost interrupt when changing a device's irq
  KVM: MMU: Fix oops on guest userspace access to guest pagetable
  KVM: MMU: large page update_pte issue with non-PAE 32-bit guests (resend)
  KVM: MMU: Fix rmap_write_protect() hugepage iteration bug
  KVM: close timer injection race window in __vcpu_run
  KVM: Fix race between timer migration and vcpu migration

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog
Linus Torvalds [Tue, 24 Jun 2008 18:23:35 +0000 (11:23 -0700)]
Merge git://git./linux/kernel/git/wim/linux-2.6-watchdog

* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
  Revert "[WATCHDOG] hpwdt: Add CFLAGS to get driver working"

16 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 24 Jun 2008 18:21:47 +0000 (11:21 -0700)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  xen: remove support for non-PAE 32-bit

16 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jwessel...
Linus Torvalds [Tue, 24 Jun 2008 18:20:59 +0000 (11:20 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jwessel/linux-2.6-kgdb

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jwessel/linux-2.6-kgdb:
  kgdb: sparse fix
  kgdb: documentation update - remove kgdboe

16 years agoenable bus mastering on i915 at resume time
Jie Luo [Tue, 24 Jun 2008 17:38:31 +0000 (10:38 -0700)]
enable bus mastering on i915 at resume time

On 9xx chips, bus mastering needs to be enabled at resume time for much of the
chip to function.  With this patch, vblank interrupts will work as expected
on resume, along with other chip functions.   Fixes kernel bugzilla #10844.

Signed-off-by: Jie Luo <clotho67@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoKVM: Remove now unused structs from kvm_para.h
Gerd Hoffmann [Tue, 3 Jun 2008 14:17:33 +0000 (16:17 +0200)]
KVM: Remove now unused structs from kvm_para.h

The kvm_* structs are obsoleted by the pvclock_* ones.
Now all users have been switched over and the old structs
can be dropped.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agox86: KVM guest: Use the paravirt clocksource structs and functions
Gerd Hoffmann [Tue, 3 Jun 2008 14:17:32 +0000 (16:17 +0200)]
x86: KVM guest: Use the paravirt clocksource structs and functions

This patch updates the kvm host code to use the pvclock structs
and functions, thereby making it compatible with Xen.

The patch also fixes an initialization bug: on SMP systems the
per-cpu has two different locations early at boot and after CPU
bringup.  kvmclock must take that in account when registering the
physical address within the host.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoKVM: Make kvm host use the paravirt clocksource structs
Gerd Hoffmann [Tue, 3 Jun 2008 14:17:31 +0000 (16:17 +0200)]
KVM: Make kvm host use the paravirt clocksource structs

This patch updates the kvm host code to use the pvclock structs.
It also makes the paravirt clock compatible with Xen.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agox86: Make xen use the paravirt clocksource structs and functions
Gerd Hoffmann [Tue, 3 Jun 2008 14:17:30 +0000 (16:17 +0200)]
x86: Make xen use the paravirt clocksource structs and functions

This patch updates the xen guest to use the pvclock structs
and helper functions.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agox86: Add structs and functions for paravirt clocksource
Gerd Hoffmann [Tue, 3 Jun 2008 14:17:29 +0000 (16:17 +0200)]
x86: Add structs and functions for paravirt clocksource

This patch adds structs for the paravirt clocksource ABI
used by both xen and kvm (pvclock-abi.h).

It also adds some helper functions to read system time and
wall clock time from a paravirtual clocksource (pvclock.[ch]).
They are based on the xen code.  They are enabled using
CONFIG_PARAVIRT_CLOCK.

Subsequent patches of this series will put the code in use.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years ago[GFS2] fix gfs2 block allocation (cleaned up)
Benjamin Marzinski [Tue, 24 Jun 2008 17:53:38 +0000 (12:53 -0500)]
[GFS2] fix gfs2 block allocation (cleaned up)

This patch fixes bz 450641.

This patch changes the computation for zero_metapath_length(), which it
renames to metapath_branch_start(). When you are extending the metadata
tree, The indirect blocks that point to the new data block must either
diverge from the existing tree either at the inode, or at the first
indirect block. They can diverge at the first indirect block because the
inode has room for 483 pointers while the indirect blocks have room for
509 pointers, so when the tree is grown, there is some free space in the
first indirect block. What metapath_branch_start() now computes is the
height where the first indirect block for the new data block is located.
It can either be 1 (if the indirect block diverges from the inode) or 2
(if it diverges from the first indirect block).

Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years ago[IA64] Eliminate NULL test after alloc_bootmem in iosapic_alloc_rte()
Julia Lawall [Tue, 24 Jun 2008 08:22:05 +0000 (10:22 +0200)]
[IA64] Eliminate NULL test after alloc_bootmem in iosapic_alloc_rte()

As noted by Akinobu Mita alloc_bootmem and related functions never return
NULL and always return a zeroed region of memory.  Thus a NULL test or
memset after calls to these functions is unnecessary.

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Tony Luck <tony.luck@intel.com>
16 years ago[IA64] Handle count==0 in sn2_ptc_proc_write()
Cliff Wickman [Tue, 24 Jun 2008 17:20:06 +0000 (10:20 -0700)]
[IA64] Handle count==0 in sn2_ptc_proc_write()

The fix applied in e0c6d97c65e0784aade7e97b9411f245a6c543e7
"security hole in sn2_ptc_proc_write" didn't take into account
the case where count==0 (which results in a buffer underrun
when adding the trailing '\0').  Thanks to Andi Kleen for
pointing this out.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
16 years ago[IA64] Fix boot failure on ia64/sn2
Jes Sorensen [Tue, 24 Jun 2008 15:30:09 +0000 (11:30 -0400)]
[IA64] Fix boot failure on ia64/sn2

Call check_sal_cache_flush() after platform_setup() as
check_sal_cache_flush() now relies on being able to call platform
vector code.

Problem was introduced by: 3463a93def55c309f3c0d0a8aaf216be3be42d64
"Update check_sal_cache_flush to use platform_send_ipi()"

Signed-off-by: Jes Sorensen <jes@sgi.com>
Tested-by: Alex Chiang: <achiang@hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
16 years agokgdb: sparse fix
Jason Wessel [Tue, 24 Jun 2008 15:52:55 +0000 (10:52 -0500)]
kgdb: sparse fix

- Fix warning reported by sparse
kernel/kgdb.c:1502:6: warning: symbol 'kgdb_console_write' was not declared.
Should it be static?

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
16 years agokgdb: documentation update - remove kgdboe
Jason Wessel [Tue, 24 Jun 2008 15:52:55 +0000 (10:52 -0500)]
kgdb: documentation update - remove kgdboe

kgdboe is not presently included kgdb, and there should be no
references to it.

Also fix the tcp port terminal connection example.

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
16 years agoxen: remove support for non-PAE 32-bit
Jeremy Fitzhardinge [Fri, 9 May 2008 11:05:57 +0000 (12:05 +0100)]
xen: remove support for non-PAE 32-bit

Non-PAE operation has been deprecated in Xen for a while, and is
rarely tested or used.  xen-unstable has now officially dropped
non-PAE support.  Since Xen/pvops' non-PAE support has also been
broken for a while, we may as well completely drop it altogether.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years ago[GFS2] BUG: unable to handle kernel paging request at ffff81002690e000
Bob Peterson [Wed, 18 Jun 2008 16:30:40 +0000 (11:30 -0500)]
[GFS2] BUG: unable to handle kernel paging request at ffff81002690e000

This patch fixes bugzilla bug bz448866: gfs2: BUG: unable to
handle kernel paging request at ffff81002690e000.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
16 years agoRevert "[WATCHDOG] hpwdt: Add CFLAGS to get driver working"
Wim Van Sebroeck [Tue, 24 Jun 2008 13:09:26 +0000 (13:09 +0000)]
Revert "[WATCHDOG] hpwdt: Add CFLAGS to get driver working"

After Linus fixed the inline assembly, the CFLAGS option is not
needed anymore.

Signed-off-by: Thomas Mingarelli <Thomas.Mingarelli@hp.com>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
16 years agoKVM: VMX: Fix host msr corruption with preemption enabled
Avi Kivity [Tue, 24 Jun 2008 08:48:49 +0000 (11:48 +0300)]
KVM: VMX: Fix host msr corruption with preemption enabled

Switching msrs can occur either synchronously as a result of calls to
the msr management functions (usually in response to the guest touching
virtualized msrs), or asynchronously when preempting a kvm thread that has
guest state loaded.  If we're unlucky enough to have the two at the same
time, host msrs are corrupted and the machine goes kaput on the next syscall.

Most easily triggered by Windows Server 2008, as it does a lot of msr
switching during bootup.

Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoKVM: ioapic: fix lost interrupt when changing a device's irq
Avi Kivity [Tue, 17 Jun 2008 22:36:36 +0000 (15:36 -0700)]
KVM: ioapic: fix lost interrupt when changing a device's irq

The ioapic acknowledge path translates interrupt vectors to irqs.  It
currently uses a first match algorithm, stopping when it finds the first
redirection table entry containing the vector.  That fails however if the
guest changes the irq to a different line, leaving the old redirection table
entry in place (though masked).  Result is interrupts not making it to the
guest.

Fix by always scanning the entire redirection table.

Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoKVM: MMU: Fix oops on guest userspace access to guest pagetable
Avi Kivity [Thu, 12 Jun 2008 13:54:41 +0000 (16:54 +0300)]
KVM: MMU: Fix oops on guest userspace access to guest pagetable

KVM has a heuristic to unshadow guest pagetables when userspace accesses
them, on the assumption that most guests do not allow userspace to access
pagetables directly. Unfortunately, in addition to unshadowing the pagetables,
it also oopses.

This never triggers on ordinary guests since sane OSes will clear the
pagetables before assigning them to userspace, which will trigger the flood
heuristic, unshadowing the pagetables before the first userspace access. One
particular guest, though (Xenner) will run the kernel in userspace, triggering
the oops.  Since the heuristic is incorrect in this case, we can simply
remove it.

Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoKVM: MMU: large page update_pte issue with non-PAE 32-bit guests (resend)
Marcelo Tosatti [Wed, 11 Jun 2008 23:32:40 +0000 (20:32 -0300)]
KVM: MMU: large page update_pte issue with non-PAE 32-bit guests (resend)

kvm_mmu_pte_write() does not handle 32-bit non-PAE large page backed
guests properly. It will instantiate two 2MB sptes pointing to the same
physical 2MB page when a guest large pte update is trapped.

Instead of duplicating code to handle this, disallow directory level
updates to happen through kvm_mmu_pte_write(), so the two 2MB sptes
emulating one guest 4MB pte can be correctly created by the page fault
handling path.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoKVM: MMU: Fix rmap_write_protect() hugepage iteration bug
Marcelo Tosatti [Sun, 8 Jun 2008 04:48:53 +0000 (01:48 -0300)]
KVM: MMU: Fix rmap_write_protect() hugepage iteration bug

rmap_next() does not work correctly after rmap_remove(), as it expects
the rmap chains not to change during iteration.  Fix (for now) by restarting
iteration from the beginning.

Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoKVM: close timer injection race window in __vcpu_run
Marcelo Tosatti [Fri, 6 Jun 2008 19:37:36 +0000 (16:37 -0300)]
KVM: close timer injection race window in __vcpu_run

If a timer fires after kvm_inject_pending_timer_irqs() but before
local_irq_disable() the code will enter guest mode and only inject such
timer interrupt the next time an unrelated event causes an exit.

It would be simpler if the timer->pending irq conversion could be done
with IRQ's disabled, so that the above problem cannot happen.

For now introduce a new vcpu requests bit to cancel guest entry.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoKVM: Fix race between timer migration and vcpu migration
Marcelo Tosatti [Fri, 6 Jun 2008 19:37:35 +0000 (16:37 -0300)]
KVM: Fix race between timer migration and vcpu migration

A guest vcpu instance can be scheduled to a different physical CPU
between the test for KVM_REQ_MIGRATE_TIMER and local_irq_disable().

If that happens, the timer will only be migrated to the current pCPU on
the next exit, meaning that guest LAPIC timer event can be delayed until
a host interrupt is triggered.

Fix it by cancelling guest entry if any vcpu request is pending.  This
has the side effect of nicely consolidating vcpu->requests checks.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoalpha: fix compile error in arch/alpha/mm/init.c
Thorsten Kranzkowski [Mon, 23 Jun 2008 20:57:22 +0000 (20:57 +0000)]
alpha: fix compile error in arch/alpha/mm/init.c

Commit 9267b4b3880d00dc2dab90f1d817c856939114f7 ("alpha: fix module load
failures on smp (bug #10926)") causes a regression for my ev4
uniprocessor build:

  CC      arch/alpha/mm/init.o
/export/data/repositories/linux-2.6/arch/alpha/mm/init.c:34: error: expected â€˜=’, â€˜,’, â€˜;’, â€˜asm’ or â€˜__attribute__’ before â€˜typeof’
make[2]: *** [arch/alpha/mm/init.o] Error 1
make[1]: *** [arch/alpha/mm] Error 2
make: *** [sub-make] Error 2

This fixes it for me (compile and boot tested):

Signed-off-by: Thorsten Kranzkowski <dl8bcu@dl8bcu.de>
Acked-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'hotfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6
Linus Torvalds [Mon, 23 Jun 2008 23:25:11 +0000 (16:25 -0700)]
Merge branch 'hotfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6

* 'hotfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6:
  NFS: nfs_updatepage(): don't mark page as dirty if an error occurred
  NFS: Fix filehandle size comparisons in the mount code
  NFS: Reduce the NFS mount code stack usage.

16 years agoNFS: nfs_updatepage(): don't mark page as dirty if an error occurred
Trond Myklebust [Thu, 5 Jun 2008 20:02:35 +0000 (16:02 -0400)]
NFS: nfs_updatepage(): don't mark page as dirty if an error occurred

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
16 years agoNFS: Fix filehandle size comparisons in the mount code
Trond Myklebust [Thu, 19 Jun 2008 19:21:11 +0000 (15:21 -0400)]
NFS: Fix filehandle size comparisons in the mount code

Fix a sign issue in xdr_decode_fhstatus3()
Fix incorrect comparison in nfs_validate_mount_data()

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
16 years agoNFS: Reduce the NFS mount code stack usage.
Trond Myklebust [Thu, 19 Jun 2008 18:20:11 +0000 (14:20 -0400)]
NFS: Reduce the NFS mount code stack usage.

This appears to fix the Oops reported in
  http://bugzilla.kernel.org/show_bug.cgi?id=10826

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
16 years agoMerge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 23 Jun 2008 19:49:22 +0000 (12:49 -0700)]
Merge branch 'core-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  futexes: fix fault handling in futex_lock_pi

16 years agoMerge branch 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 23 Jun 2008 19:48:50 +0000 (12:48 -0700)]
Merge branch 'sched-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  sched: refactor wait_for_completion_timeout()
  sched: fix wait_for_completion_timeout() spurious failure under heavy load
  sched: rt: dont stop the period timer when there are tasks wanting to run

16 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 23 Jun 2008 19:48:17 +0000 (12:48 -0700)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  xen: don't drop NX bit
  xen: mask unwanted pte bits in __supported_pte_mask
  xen: Use wmb instead of rmb in xen_evtchn_do_upcall().
  x86: fix NULL pointer deref in __switch_to

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Mon, 23 Jun 2008 19:45:49 +0000 (12:45 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/mthca: Clear ICM pages before handing to FW

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Mon, 23 Jun 2008 19:18:06 +0000 (12:18 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: sb - Fix wrong assertions
  ALSA: aw2 - Fix Oops at initialization

16 years agomm: fix race in COW logic
Nick Piggin [Mon, 23 Jun 2008 12:30:30 +0000 (14:30 +0200)]
mm: fix race in COW logic

There is a race in the COW logic.  It contains a shortcut to avoid the
COW and reuse the page if we have the sole reference on the page,
however it is possible to have two racing do_wp_page()ers with one
causing the other to mistakenly believe it is safe to take the shortcut
when it is not.  This could lead to data corruption.

Process 1 and process2 each have a wp pte of the same anon page (ie.
one forked the other).  The page's mapcount is 2.  Then they both
attempt to write to it around the same time...

  proc1 proc2 thr1 proc2 thr2
  CPU0 CPU1 CPU3
  do_wp_page() do_wp_page()
 trylock_page()
  can_share_swap_page()
   load page mapcount (==2)
  reuse = 0
 pte unlock
 copy page to new_page
 pte lock
 page_remove_rmap(page);
   trylock_page()
    can_share_swap_page()
     load page mapcount (==1)
    reuse = 1
   ptep_set_access_flags (allow W)

  write private key into page
read from page
ptep_clear_flush()
set_pte_at(pte of new_page)

Fix this by moving the page_remove_rmap of the old page after the pte
clear and flush.  Potentially the entire branch could be moved down
here, but in order to stay consistent, I won't (should probably move all
the *_mm_counter stuff with one patch).

Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: Hugh Dickins <hugh@veritas.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoFix ZERO_PAGE breakage with vmware
Linus Torvalds [Mon, 23 Jun 2008 18:21:37 +0000 (11:21 -0700)]
Fix ZERO_PAGE breakage with vmware

Commit 89f5b7da2a6bad2e84670422ab8192382a5aeb9f ("Reinstate ZERO_PAGE
optimization in 'get_user_pages()' and fix XIP") broke vmware, as
reported by Jeff Chua:

  "This broke vmware 6.0.4.
   Jun 22 14:53:03.845: vmx| NOT_IMPLEMENTED
   /build/mts/release/bora-93057/bora/vmx/main/vmmonPosix.c:774"

and the reason seems to be that there's an old bug in how we handle do
FOLL_ANON on VM_SHARED areas in get_user_pages(), but since it only
triggered if the whole page table was missing, nobody had apparently hit
it before.

The recent changes to 'follow_page()' made the FOLL_ANON logic trigger
not just for whole missing page tables, but for individual pages as
well, and exposed this problem.

This fixes it by making the test for when FOLL_ANON is used more
careful, and also makes the code easier to read and understand by moving
the logic to a separate inline function.

Reported-and-tested-by: Jeff Chua <jeff.chua.linux@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoremoved unused var real_tty on n_tty_ioctl()
Gustavo Fernando Padovan [Mon, 23 Jun 2008 11:07:06 +0000 (12:07 +0100)]
removed unused var real_tty on n_tty_ioctl()

I noted that the 'struct tty_struct *real_tty' is not used in this
function, so I removed the code about 'real_tty'.

Signed-off-by: Gustavo Fernando Padovan <gustavo@las.ic.unicamp.br>
Acked-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotty_driver: Update required method documentation
Alan Cox [Mon, 23 Jun 2008 11:06:52 +0000 (12:06 +0100)]
tty_driver: Update required method documentation

Some of the requirement rules are now more relaxed. Also correct a
contradiction in the previous update

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoIB/mthca: Clear ICM pages before handing to FW
Eli Cohen [Mon, 23 Jun 2008 16:29:58 +0000 (09:29 -0700)]
IB/mthca: Clear ICM pages before handing to FW

Current memfree FW has a bug which in some cases, assumes that ICM
pages passed to it are cleared.  This patch uses __GFP_ZERO to
allocate all ICM pages passed to the FW.  Once firmware with a fix is
released, we can make the workaround conditional on firmware version.

This fixes the bug reported by Arthur Kepner <akepner@sgi.com> here:
http://lists.openfabrics.org/pipermail/general/2008-May/050026.html

Cc: <stable@kernel.org>
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
[ Rewritten to be a one-liner using __GFP_ZERO instead of vmap()ing
  ICM memory and memset()ing it to 0. - Roland ]

Signed-off-by: Roland Dreier <rolandd@cisco.com>
16 years agofutexes: fix fault handling in futex_lock_pi
Thomas Gleixner [Mon, 23 Jun 2008 09:21:58 +0000 (11:21 +0200)]
futexes: fix fault handling in futex_lock_pi

This patch addresses a very sporadic pi-futex related failure in
highly threaded java apps on large SMP systems.

David Holmes reported that the pi_state consistency check in
lookup_pi_state triggered with his test application. This means that
the kernel internal pi_state and the user space futex variable are out
of sync. First we assumed that this is a user space data corruption,
but deeper investigation revieled that the problem happend because the
pi-futex code is not handling a fault in the futex_lock_pi path when
the user space variable needs to be fixed up.

The fault happens when a fork mapped the anon memory which contains
the futex readonly for COW or the page got swapped out exactly between
the unlock of the futex and the return of either the new futex owner
or the task which was the expected owner but failed to acquire the
kernel internal rtmutex. The current futex_lock_pi() code drops out
with an inconsistent in case it faults and returns -EFAULT to user
space. User space has no way to fixup that state.

When we wrote this code we thought that we could not drop the hash
bucket lock at this point to handle the fault.

After analysing the code again it turned out to be wrong because there
are only two tasks involved which might modify the pi_state and the
user space variable:

 - the task which acquired the rtmutex
 - the pending owner of the pi_state which did not get the rtmutex

Both tasks drop into the fixup_pi_state() function before returning to
user space. The first task which acquired the hash bucket lock faults
in the fixup of the user space variable, drops the spinlock and calls
futex_handle_fault() to fault in the page. Now the second task could
acquire the hash bucket lock and tries to fixup the user space
variable as well. It either faults as well or it succeeds because the
first task already faulted the page in.

One caveat is to avoid a double fixup. After returning from the fault
handling we reacquire the hash bucket lock and check whether the
pi_state owner has been modified already.

Reported-by: David Holmes <david.holmes@sun.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: David Holmes <david.holmes@sun.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: <stable@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
 kernel/futex.c |   93 ++++++++++++++++++++++++++++++++++++++++++++-------------
 1 file changed, 73 insertions(+), 20 deletions(-)

16 years agoALSA: sb - Fix wrong assertions
Takashi Iwai [Mon, 23 Jun 2008 09:58:06 +0000 (11:58 +0200)]
ALSA: sb - Fix wrong assertions

snd_assert() in save_mixer() and restore_mixer() in sb_mixer.c is
just wrong.  The debug code wasn't tested at all, obviously...

Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agoALSA: aw2 - Fix Oops at initialization
Takashi Iwai [Mon, 23 Jun 2008 09:54:05 +0000 (11:54 +0200)]
ALSA: aw2 - Fix Oops at initialization

The irq handler may be called before the proper initialization of hardware.
Call snd_aw2_saa7146_setup() before the irq handler registration.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agoMerge branch 'linus' into sched/urgent
Ingo Molnar [Mon, 23 Jun 2008 09:00:26 +0000 (11:00 +0200)]
Merge branch 'linus' into sched/urgent

16 years agoFix performance regression on lmbench select benchmark
Linus Torvalds [Sun, 22 Jun 2008 19:23:15 +0000 (12:23 -0700)]
Fix performance regression on lmbench select benchmark

Christian Borntraeger reported that reinstating cond_resched() with
CONFIG_PREEMPT caused a performance regression on lmbench:

For example select file 500:
23 microseconds
32 microseconds

and that's really because we totally unnecessarily do the cond_resched()
in the innermost loop of select(), which is just silly.

This moves it out from the innermost loop (which only ever loops ove the
bits in a single "unsigned long" anyway), which makes the performance
regression go away.

Reported-and-tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoSlab: Fix memory leak in fallback_alloc()
Christoph Lameter [Sat, 21 Jun 2008 23:46:35 +0000 (16:46 -0700)]
Slab: Fix memory leak in fallback_alloc()

The zonelist patches caused the loop that checks for available
objects in permitted zones to not terminate immediately. One object
per zone per allocation may be allocated and then abandoned.

Break the loop when we have successfully allocated one object.

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Linus Torvalds [Sat, 21 Jun 2008 23:43:56 +0000 (16:43 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/tytso/ext4

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  Ext4: Fix online resize block group descriptor corruption

16 years agoMerge branch 'release' of git://lm-sensors.org/kernel/mhoffman/hwmon-2.6
Linus Torvalds [Sat, 21 Jun 2008 19:31:02 +0000 (12:31 -0700)]
Merge branch 'release' of git://lm-sensors.org/kernel/mhoffman/hwmon-2.6

* 'release' of git://lm-sensors.org/kernel/mhoffman/hwmon-2.6:
  hwmon: (lm75) sensor reading bugfix
  hwmon: (abituguru3) update driver detection
  hwmon: (w83791d) new maintainer
  hwmon: (abituguru3) Identify Abit AW8D board as such
  hwmon: Update the sysfs interface documentation
  hwmon: (adt7473) Initialize max_duty_at_overheat before use
  hwmon: (lm85) Fix function RANGE_TO_REG()

16 years agoAdd return value to reserve_bootmem_node()
Bernhard Walle [Sat, 21 Jun 2008 17:01:02 +0000 (19:01 +0200)]
Add return value to reserve_bootmem_node()

This patch changes the function reserve_bootmem_node() from void to int,
returning -ENOMEM if the allocation fails.

This fixes a build problem on x86 with CONFIG_KEXEC=y and
CONFIG_NEED_MULTIPLE_NODES=y

Signed-off-by: Bernhard Walle <bwalle@suse.de>
Reported-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Sat, 21 Jun 2008 15:44:08 +0000 (08:44 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  netns: Don't receive new packets in a dead network namespace.
  sctp: Make sure N * sizeof(union sctp_addr) does not overflow.
  pppoe: warning fix
  ipv6: Drop packets for loopback address from outside of the box.
  ipv6: Remove options header when setsockopt's optlen is 0
  mac80211: detect driver tx bugs

16 years agonetns: Don't receive new packets in a dead network namespace.
Eric W. Biederman [Sat, 21 Jun 2008 05:16:51 +0000 (22:16 -0700)]
netns: Don't receive new packets in a dead network namespace.

Alexey Dobriyan <adobriyan@gmail.com> writes:
> Subject: ICMP sockets destruction vs ICMP packets oops

> After icmp_sk_exit() nuked ICMP sockets, we get an interrupt.
> icmp_reply() wants ICMP socket.
>
> Steps to reproduce:
>
>  launch shell in new netns
>  move real NIC to netns
>  setup routing
>  ping -i 0
>  exit from shell
>
> BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
> IP: [<ffffffff803fce17>] icmp_sk+0x17/0x30
> PGD 17f3cd067 PUD 17f3ce067 PMD 0
> Oops: 0000 [1] PREEMPT SMP DEBUG_PAGEALLOC
> CPU 0
> Modules linked in: usblp usbcore
> Pid: 0, comm: swapper Not tainted 2.6.26-rc6-netns-ct #4
> RIP: 0010:[<ffffffff803fce17>]  [<ffffffff803fce17>] icmp_sk+0x17/0x30
> RSP: 0018:ffffffff8057fc30  EFLAGS: 00010286
> RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffff81017c7db900
> RDX: 0000000000000034 RSI: ffff81017c7db900 RDI: ffff81017dc41800
> RBP: ffffffff8057fc40 R08: 0000000000000001 R09: 000000000000a815
> R10: 0000000000000000 R11: 0000000000000001 R12: ffffffff8057fd28
> R13: ffffffff8057fd00 R14: ffff81017c7db938 R15: ffff81017dc41800
> FS:  0000000000000000(0000) GS:ffffffff80525000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b
> CR2: 0000000000000000 CR3: 000000017fcda000 CR4: 00000000000006e0
> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
> Process swapper (pid: 0, threadinfo ffffffff8053a000, task ffffffff804fa4a0)
> Stack:  0000000000000000 ffff81017c7db900 ffffffff8057fcf0 ffffffff803fcfe4
>  ffffffff804faa38 0000000000000246 0000000000005a40 0000000000000246
>  000000000001ffff ffff81017dd68dc0 0000000000005a40 0000000055342436
> Call Trace:
>  <IRQ>  [<ffffffff803fcfe4>] icmp_reply+0x44/0x1e0
>  [<ffffffff803d3a0a>] ? ip_route_input+0x23a/0x1360
>  [<ffffffff803fd645>] icmp_echo+0x65/0x70
>  [<ffffffff803fd300>] icmp_rcv+0x180/0x1b0
>  [<ffffffff803d6d84>] ip_local_deliver+0xf4/0x1f0
>  [<ffffffff803d71bb>] ip_rcv+0x33b/0x650
>  [<ffffffff803bb16a>] netif_receive_skb+0x27a/0x340
>  [<ffffffff803be57d>] process_backlog+0x9d/0x100
>  [<ffffffff803bdd4d>] net_rx_action+0x18d/0x250
>  [<ffffffff80237be5>] __do_softirq+0x75/0x100
>  [<ffffffff8020c97c>] call_softirq+0x1c/0x30
>  [<ffffffff8020f085>] do_softirq+0x65/0xa0
>  [<ffffffff80237af7>] irq_exit+0x97/0xa0
>  [<ffffffff8020f198>] do_IRQ+0xa8/0x130
>  [<ffffffff80212ee0>] ? mwait_idle+0x0/0x60
>  [<ffffffff8020bc46>] ret_from_intr+0x0/0xf
>  <EOI>  [<ffffffff80212f2c>] ? mwait_idle+0x4c/0x60
>  [<ffffffff80212f23>] ? mwait_idle+0x43/0x60
>  [<ffffffff8020a217>] ? cpu_idle+0x57/0xa0
>  [<ffffffff8040f380>] ? rest_init+0x70/0x80
> Code: 10 5b 41 5c 41 5d 41 5e c9 c3 66 2e 0f 1f 84 00 00 00 00 00 55 48 89 e5 53
> 48 83 ec 08 48 8b 9f 78 01 00 00 e8 2b c7 f1 ff 89 c0 <48> 8b 04 c3 48 83 c4 08
> 5b c9 c3 66 66 66 66 66 2e 0f 1f 84 00
> RIP  [<ffffffff803fce17>] icmp_sk+0x17/0x30
>  RSP <ffffffff8057fc30>
> CR2: 0000000000000000
> ---[ end trace ea161157b76b33e8 ]---
> Kernel panic - not syncing: Aiee, killing interrupt handler!

Receiving packets while we are cleaning up a network namespace is a
racy proposition. It is possible when the packet arrives that we have
removed some but not all of the state we need to fully process it.  We
have the choice of either playing wack-a-mole with the cleanup routines
or simply dropping packets when we don't have a network namespace to
handle them.

Since the check looks inexpensive in netif_receive_skb let's just
drop the incoming packets.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agosctp: Make sure N * sizeof(union sctp_addr) does not overflow.
David S. Miller [Sat, 21 Jun 2008 05:04:34 +0000 (22:04 -0700)]
sctp: Make sure N * sizeof(union sctp_addr) does not overflow.

As noticed by Gabriel Campana, the kmalloc() length arg
passed in by sctp_getsockopt_local_addrs_old() can overflow
if ->addr_num is large enough.

Therefore, enforce an appropriate limit.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopppoe: warning fix
Stephen Hemminger [Sat, 21 Jun 2008 04:58:02 +0000 (21:58 -0700)]
pppoe: warning fix

Fix warning:
drivers/net/pppoe.c: In function 'pppoe_recvmsg':
drivers/net/pppoe.c:945: warning: comparison of distinct pointer types lacks a cast
because skb->len is unsigned int and total_len is size_t

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Sat, 21 Jun 2008 00:10:04 +0000 (17:10 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  [IA64] SN2: security hole in sn2_ptc_proc_write

16 years agoalpha: resurrect Cypress IDE quirk
Ivan Kokshaysky [Fri, 20 Jun 2008 23:28:54 +0000 (03:28 +0400)]
alpha: resurrect Cypress IDE quirk

Which was removed in the hope that generic legacy IDE quirk in
drivers/pci/probe.c is sufficient for Cypress IDE.
It isn't, as this controller has non-standard BAR layout:
secondary channel registers are in the BAR0-1 of the second
PCI function - not in the BAR2-3 of the same function, as the
generic quirk routine assumes.

Signed-off-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoalpha: fix compile failures with gcc-4.3 (bug #10438)
Ivan Kokshaysky [Fri, 20 Jun 2008 23:28:31 +0000 (03:28 +0400)]
alpha: fix compile failures with gcc-4.3 (bug #10438)

Vast majority of these build failures are gcc-4.3 warnings
about static functions and objects being referenced from
non-static (read: "extern inline") functions, in conjunction
with our -Werror.

We cannot just convert "extern inline" to "static inline",
as people keep suggesting all the time, because "extern inline"
logic is crucial for generic kernel build.
So
- just make sure that all callees of critical "extern inline"
  functions are also "extern inline";
- use "static inline", wherever it's possible.

traps.c: work around gcc-4.3 being too smart about array
bounds-checking.

TODO: add "gnu_inline" attribute to all our "extern inline"
functions to ensure desired behaviour with future compilers.

Signed-off-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoalpha: link failure fix
Ivan Kokshaysky [Fri, 20 Jun 2008 23:26:21 +0000 (03:26 +0400)]
alpha: link failure fix

With built-in scsi disk driver, the final link fails with a following
error:
`.exit.text' referenced in section `.rodata' of drivers/built-in.o:
defined in discarded section `.exit.text' of drivers/built-in.o

This happens with -Os (CONFIG_CC_OPTIMIZE_FOR_SIZE=y) with all gcc-4
versions, and also with -O2 and gcc-4.3.

The problem is in sd.c:sd_major() being inlined into __exit function
exit_sd(), and the compiler generating a jump table in .rodata section
for the 'switch' statement in sd_major(). So we have references to
discarded section.

Fixed with a big hammer in the form of -fno-jump-tables.

Note that jump tables vs. discarded sections is a generic problem,
other architectures are just lucky not to suffer from it. But with
a slightly more complex switch/case statement it can be reproduced
on x86 as well. So maybe at some point we should consider
-fno-jump-tables as a generic compile option...

Signed-off-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoalpha: fix module load failures on smp (bug #10926)
Ivan Kokshaysky [Fri, 20 Jun 2008 23:25:39 +0000 (03:25 +0400)]
alpha: fix module load failures on smp (bug #10926)

To calculate addresses of locally defined variables, GCC uses 32-bit
displacement from the GP. Which doesn't work for per cpu variables in
modules, as an offset to the kernel per cpu area is way above 4G.

The workaround is to force allocation of a GOT entry for per cpu variable
using ldq instruction with a 'literal' relocation.
I had to use custom asm/percpu.h, as a required argument magic doesn't
work with asm-generic/percpu.h macros.

Signed-off-by: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoLinux 2.6.26-rc7
Linus Torvalds [Fri, 20 Jun 2008 23:19:44 +0000 (16:19 -0700)]
Linux 2.6.26-rc7

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6
Linus Torvalds [Fri, 20 Jun 2008 19:46:47 +0000 (12:46 -0700)]
Merge git://git./linux/kernel/git/bart/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6:
  BAST: Remove old IDE driver
  pcmcia ide kingston compactflash's have a new manufacturer id
  pcmcia: add another pata/ide ID
  pcmcia: add an pata/ide ID
  ide: increase timeout in wait_drive_not_busy()
  palm_bk3710: fix resource management

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394...
Linus Torvalds [Fri, 20 Jun 2008 19:41:10 +0000 (12:41 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ieee1394/linux1394-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6:
  ieee1394: Kconfig menu touch-up
  firewire: Kconfig menu touch-up
  firewire: deadline for PHY config transmission
  firewire: fw-ohci: unify printk prefixes
  firewire: fill_bus_reset_event needs lock protection
  firewire: fw-ohci: write selfIDBufferPtr before LinkControl.rcvSelfID
  firewire: fw-ohci: disable PHY packet reception into AR context
  firewire: fw-ohci: use of uninitialized data in AR handler
  firewire: don't panic on invalid AR request buffer

16 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Fri, 20 Jun 2008 19:39:12 +0000 (12:39 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6:
  ACPI: no AC status notification
  ACPI Exception (video-1721): UNKNOWN_STATUS_CODE, Cant attach device

16 years agoMerge branch 'drm-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Fri, 20 Jun 2008 19:38:18 +0000 (12:38 -0700)]
Merge branch 'drm-patches' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (21 commits)
  drm: only trust core drm ioctls - driver ioctls are a mess.
  drm/i915: add support for Intel series 4 chipsets.
  drm/radeon: add hier-z registers for r300 and r500 chipsets
  drm/radeon: use DSTCACHE_CTLSTAT rather than RB2D_DSTCACHE_CTLSTAT
  drm/radeon: switch IGP gart to use radeon_write_agp_base()
  drm/radeon: Restore sw interrupt on resume
  drm/r500: add support for AGP based cards.
  drm/radeon: fix texture uploads with large 3d textures (bug 13980)
  drm/radeon: add initial r500 support.
  drm/radeon: init pipe setup in kernel code.
  drm/radeon: fixup radeon_do_engine_reset
  drm/radeon: fix pixcache and purge/cache flushing registers
  drm/radeon: write AGP_BASE_2 on chips that support it.
  drm/radeon: merge IGP chip setup and fixup RS400 vs RS480 support
  drm/radeon: IGP clean up register and magic numbers.
  drm/rs690: set base 2 to 0.
  drm/rs690: set all of gart base address.
  radeon: add production microcode from AMD
  drm: pcigart use proper pci map interfaces.
  drm: the sg alloc ioctl should write back the handle to userspace
  ...

16 years agoMerge branch 'agp-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Fri, 20 Jun 2008 19:37:55 +0000 (12:37 -0700)]
Merge branch 'agp-patches' of git://git./linux/kernel/git/airlied/agp-2.6

* 'agp-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/agp-2.6:
  [agp]: fixup chipset flush for new Intel G4x.
  agp: brown paper bag patch - put back the two lines it took out.

16 years agoMerge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 20 Jun 2008 19:37:13 +0000 (12:37 -0700)]
Merge branch 'core-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  softlockup: fix NMI hangs due to lock race - 2.6.26-rc regression
  rcupreempt: remove export of rcu_batches_completed_bh
  cpuset: limit the input of cpuset.sched_relax_domain_level

16 years agoMerge branch 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 20 Jun 2008 19:36:55 +0000 (12:36 -0700)]
Merge branch 'sched-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  sched, delay accounting: fix incorrect delay time when constantly waiting on runqueue
  sched: CPU hotplug events must not destroy scheduler domains created by the cpusets
  sched: rt-group: fix RR buglet
  sched: rt-group: heirarchy aware throttle
  sched: rt-group: fix hierarchy
  sched: NULL pointer dereference while setting sched_rt_period_us
  sched: fix defined-but-unused warning

16 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 20 Jun 2008 19:36:38 +0000 (12:36 -0700)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86, geode: add a VSA2 ID for General Software
  x86: use BOOTMEM_EXCLUSIVE on 32-bit
  x86, 32-bit: fix boot failure on TSC-less processors
  x86: fix NULL pointer deref in __switch_to
  x86: set PAE PHYSICAL_MASK_SHIFT to 44 bits.

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney...
Linus Torvalds [Fri, 20 Jun 2008 19:34:43 +0000 (12:34 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/cooloney/blackfin-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney/blackfin-2.6:
  Blackfin Serial Driver: Use timer to poll CTS PIN instead of workqueue.
  Blackfin arch: fix typo error in bf548 serial header file

16 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Fri, 20 Jun 2008 19:31:03 +0000 (12:31 -0700)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  ahci: sis can't do PMP
  ata_piix: add TECRA M4 to broken suspend list
  LIBATA: Add HAVE_PATA_PLATFORM to select PATA_PLATFORM driver
  sata_mv: warn on PIO with multiple DRQs
  sata_mv: enable async_notify for 60x1 Rev.C0 and higher
  libata: don't check whether to use DMA or not for no data commands
  ahci: jmb361 has only one port