Revert commit 0c2d64fb6c because it causes
(arguably poorly designed) existing userspace to spend interminable
periods closing billions of not-open file descriptors.
We could bring this back, with some sort of opt-in tunable in /proc, which
defaults to "off".
Peter's alanysis follows:
: I spent several hours trying to get to the bottom of a serious
: performance issue that appeared on one of our servers after upgrading to
: 2.6.28. In the end it's what could be considered a userspace bug that
: was triggered by a change in 2.6.28. Since this might also affect other
: people I figured I'd at least document what I found here, and maybe we
: can even do something about it:
:
:
: So, I upgraded some of debian.org's machines to 2.6.28.1 and immediately
: the team maintaining our ftp archive complained that one of their
: scripts that previously ran in a few minutes still hadn't even come
: close to being done after an hour or so. Downgrading to 2.6.27 fixed
: that.
:
: Turns out that script is forking a lot and something in it or python or
: whereever closes all the file descriptors it doesn't want to pass on.
: That is, it starts at zero and goes up to ulimit -n/RLIMIT_NOFILE and
: closes them all with a few exceptions.
:
: Turns out that takes a long time when your limit -n is now 2^20 (1048576).
:
: With 2.6.27.* the ulimit -n was the standard 1024, but with 2.6.28 it is
: now a thousand times that.
:
: 2.6.28 included a patch titled "rlimit: permit setting RLIMIT_NOFILE to
: RLIM_INFINITY" (0c2d64fb6c)[1] that
: allows, as the title implies, to set the limit for number of files to
: infinity.
:
: Closer investigation showed that the broken default ulimit did not apply
: to "system" processes (like stuff started from init). In the end I
: could establish that all processes that passed through pam_limit at one
: point had the bad resource limit.
:
: Apparently the pam library in Debian etch (4.0) initializes the limits
: to some default values when it doesn't have any settings in limit.conf
: to override them. Turns out that for nofiles this is RLIM_INFINITY.
: Commenting out "case RLIMIT_NOFILE" in pam_limit.c:267 of our pam
: package version 0.79-5 fixes that - tho I'm not sure what side effects
: that has.
:
: Debian lenny (the upcoming 5.0 version) doesn't have this issue as it
: uses a different pam (version).
Reported-by: Peter Palfrader <weasel@debian.org>
Cc: Adam Tkac <vonsch@gmail.com>
Cc: Michael Kerrisk <mtk.manpages@googlemail.com>
Cc: <stable@kernel.org> [2.6.28.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
shm_get_stat() assumes that the inode is a "struct shmem_inode_info",
which is incorrect for !CONFIG_SHMEM (see fs/ramfs/inode.c:
ramfs_get_inode() vs. mm/shmem.c: shmem_get_inode()).
This bad assumption can cause shmctl(SHM_INFO) to lockup when
shm_get_stat() tries to spin_lock(&info->lock). Users of !CONFIG_SHMEM
may encounter this lockup simply by invoking the 'ipcs' command.
Reported by Jiri Olsa back in February 2008:
http://lkml.org/lkml/2008/2/29/74
Signed-off-by: Tony Battersby <tonyb@cybernetics.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Reported-by: Jiri Olsa <olsajiri@gmail.com>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: <stable@kernel.org> [2.6.everything]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Avoid calling copy_from/to_user() with fb_info->lock mutex held in fbmem
ioctl().
fb_mmap() is called under mm->mmap_sem (A) held, that also acquires
fb_info->lock (B); fb_ioctl() takes fb_info->lock (B) and does
copy_from/to_user() that might acquire mm->mmap_sem (A), causing a
deadlock.
NOTE: it doesn't push down the fb_info->lock in each own driver's
fb_ioctl(), so there are still potential deadlocks elsewhere.
Signed-off-by: Andrea Righi <righi.andrea@gmail.com>
Cc: Dave Jones <davej@redhat.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Johannes Weiner <hannes@saeurebad.de>
Cc: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: Harvey Harrison <harvey.harrison@gmail.com>
Cc: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Simple RTC driver for the MSP430 firmware on the DM355 EVM board. Other
than not supporting atomic reads/writes of all four bytes, this is
reasonable as a basic no-alarm RTC.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Acked-by: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
dell-laptop makes use of the power supply class information to choose
which backlight interface to change. Add a depends on it.
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Len Brown <lenb@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The swap() macro is accidentally retuning the value of its first argument.
Change it into a doesn't-return-anything macro before someone goes and
relies upon this behaviour.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Wu Fengguang <wfg@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The device can take a while to respond to an open/close request, so
increase the time kernel will wait for response (1 ms to 10ms).
Also, properly clean up a channel on a failed open, by calling the channel
close routine. Just freeing the memory isn't sufficient, the device needs
to be informed that the channel is no longer open, and the device memory
cleared of references to freed dma buffer.
Signed-off-by: David Altobelli <david.altobelli@hp.com>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
alpha:
kernel/async.c: In function 'run_one_entry':
kernel/async.c:141: warning: format '%lli' expects type 'long long int', but argument 2 has type 'async_cookie_t'
kernel/async.c:149: warning: format '%lli' expects type 'long long int', but argument 2 has type 'async_cookie_t'
kernel/async.c:149: warning: format '%lld' expects type 'long long int', but argument 4 has type 's64'
kernel/async.c: In function 'async_synchronize_cookie_special':
kernel/async.c:250: warning: format '%lli' expects type 'long long int', but argument 3 has type 's64'
Cc: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Some fixes regarding snd-hda-intel workqueue:
- Use create_singlethread_workqueue() instead of create_workqueue()
as per-CPU work isn't required.
- Allocate workq name string properly
- Renamed the workq name to "hd-audio*" to be more obvious.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
We were freeing an offset into the slab object instead of the
start. This patch fixes it by calling crypto_destroy_tfm which
allows the correct address to be given.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Geert Uytterhoeven pointed out that we're not zeroing all the
memory when freeing a transform. This patch fixes it by calling
ksize to ensure that we zero everything in sight.
Reported-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Module reference counting for shash is incorrect: when
a new shash transformation is created the refcount is not
increased as it should.
Signed-off-by: Adrian-Ken Rueegsegger <rueegsegger@swiss-it.ch>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Currently, the PM core always attempts to manage devices with drivers
that use the new PM framework. In particular, it attempts to disable
the devices (which is unnecessary), to save their state (which may be
undesirable if the driver has done that already) and to put them into
low power states (again, this may be undesirable if the driver has
already put the device into a low power state). That need not be
the right thing to do, so make the core be more careful in this
respect.
Generally, there are the following categories of devices to consider:
* bridge devices without drivers
* non-bridge devices without drivers
* bridge devices with drivers
* non-bridge devices with drivers
and each of them should be handled differently.
For bridge devices without drivers the PCI PM core will save their
state on suspend and restore it (early) during resume, after putting
them into D0 if necessary. It will not attempt to do anything else
to these devices.
For non-bridge devices without drivers the PCI PM core will disable
them and save their state on suspend. During resume, it will put
them into D0, if necessary, restore their state (early) and reenable
them.
For bridge devices with drivers the PCI PM core will only save
their state on suspend if the driver hasn't done that already.
Still, the core will restore their state (early) during resume,
after putting them into D0, if necessary.
For non-bridge devices with drivers the PCI PM core will only save
their state on suspend if the driver hasn't done that already. Also,
if the state of the device hasn't been saved by the driver, the core
will attempt to put the device into a low power state. During
resume the core will restore the state of the device (early), after
putting it into D0, if necessary.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
pci_restore_standard_config() unconditionally changes current_state
to PCI_D0 after attempting to change the device's power state, but
it should rather read the actual current power state from the
device.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
It is a mistake to disable and enable PCI bridges and PCI Express
ports during suspend-resume, at least at the time when it is
currently done. Disabling them may lead to problems with accessing
devices behind them and they should be automatically enabled when
their standard config spaces are restored. Fix this by not attempting
to disable bridges during suspend and enable them during resume.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Simplify suspend and resume of the PCI Express port driver. It no
longer needs to save and restore the standard configuration space of the
device; this is now done by the PCI PM core layer.
This patch is reported to fix the regression tracked as
http://bugzilla.kernel.org/show_bug.cgi?id=12598
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-and-tested-by: Parag Warudkar <parag.lkml@gmail.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Make pci_legacy_suspend() save the state of the device if it is
in PCI_UNKNOWN after its suspend callback has run and warn only if
the power state of the device has been changed by its suspend
callback.
Also, use WARN_ONCE(), which is more useful, in pci_legacy_suspend(),
so that the name of the offending function is printed.
Additionally, remove the unnecessary line of code setting
pci_dev->state_saved.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Check if the standard configuration registers of a PCI device have
been saved during suspend before trying to restore them during
resume.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-By: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Suspend to RAM is reported to break on some machines as a result of
attempting to put one of driverless PCI devices into a low power
state. Avoid that by not attepmting to power manage driverless
devices during suspend.
Fix up pci_pm_poweroff() after a previous incomplete fix for the same
thing during hibernation.
This patch is reported to fix the regression from 2.6.28 tracked as
http://bugzilla.kernel.org/show_bug.cgi?id=12605
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Reported-and-tested-by: Eric Sesterhenn <snakebyte@gmx.de>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This patch makes the ROM reading code return an error to user space if
the size of the ROM read is equal to 0.
The patch also emits a warnings if the contents of the ROM are invalid,
and documents the effects of the "enable" file on ROM reading.
Signed-off-by: Timothy S. Nelson <wayland@wayland.id.au>
Acked-by: Alex Villacis-Lasso <a_villacis@palosanto.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
We only want to disable ASPM when the last function is removed from
the parent's device list. We determine this by checking to see if
the parent's device list is completely empty.
Unfortunately, we never hit that code because the parent is considered
an upstream port, and never had an ASPM link_state associated with it.
The early check for !link_state causes us to return early, we never
discover that our device list is empty, and thus we never remove the
downstream ports' link_state nodes.
Instead of checking to see if the parent's device list is empty, we can
check to see if we are the last device on the list, and if so, then we
know that we can clean up properly.
Cc: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
SOFT_RESET must be asserted for at least 3 TX clocks in order for it to work
properly. The syncs in the gfar_write() commands have been hiding this, but
we need to guarantee it.
Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
BD_LENGTH_MASK is supposed to catch the low 16-bits of the status field, not
the low byte. The old way, we would never be able to clean up tx packets with
sizes divisible by 256.
Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The LRO switch is always set to 1 in the rx processing loop.
It breaks the accelerated iSCSI receive traffic.
Fix its computation.
Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
x86: APIC: enable workaround on AMD Fam10h CPUs
xen: disable interrupts before saving in percpu
x86: add x86@kernel.org to MAINTAINERS
x86: push old stack address on irqstack for unwinder
irq, x86: fix lock status with numa_migrate_irq_desc
x86: add cache descriptors for Intel Core i7
x86/Voyager: make it build and boot
Cleans uCode key table bit map iwl_clear_stations_table
since all stations are cleared also the key table must be.
Since the keys are not removed properly on suspend by mac80211
this may result in exhausting key table on resume leading
to memory corruption during removal
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This reverts commit 0e0333429a.
I committed this by accident - Joel and Louis are working with the lockdep
maintainer to provide a better solution than just turning lockdep off.
Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Acked-by: <Joel Becker <joel.becker@oracle.com>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
ALSA: pcm_oss: AFMT_S24_LE is set twice in return value
ALSA: ASoC: email - update email addresses.
OMAP: ASoC: Fix spinlock misuse in omap-pcm.c
ALSA: hda - No widget selection for volume knob widgets in proc output
ALSA: hda - Add support of iMac 24 Aluminium
ALSA: alsa: time reaches -1, tested 0
ALSA: hda - Add quirk for another HP dv5 model
AFMT_S24_LE is set twice in return value
vi sound/core/oss/pcm_oss.c +640
#define AFMT_S24_LE 0x00008000
#define AFMT_S24_BE 0x00010000
Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney/blackfin-2.6: (40 commits)
Blackfin arch: Remove outdated code
Blackfin arch: Fix udelay implementation
Blackfin arch: Update Copyright information
Blackfin arch: Add BF561 PPI POLS, POLC Masks
Blackfin arch: Update CM-BF527 kernel config
Blackfin arch: define bfin_memmap as static since it is only used here
Blackfin arch: cplb mananger: use a do...while loop rather than a for loop
Blackfin arch: fix bug - traps test case 19 for exception 0x2d fails
Blackfin arch: add platform device bfin_mii-bus and KSZ8893M switch driver platform resources to board files
Blackfin arch: build jtag tty driver as a module by default
Blackfin arch: fix 2 bugs related to debug
Blackfin arch: Add ANOMALY_05000380 to BF54x to kill the compile warning
Blackfin arch: Fix bug - 561 SMP kernel can't boot from jffs2
Blackfin arch: base SIC_IWR# programming on whether the MMR exists
Blackfin arch: read SYSCR on newer parts that mirror the bits of SWRST in it
Blackfin arch: fixup board init function name
Blackfin arch: drop CONFIG_I2C_BOARDINFO ifdefs
Blackfin arch: bfin_reset->_bfin_reset redirection no longer needed
Blackfin arch: sync reboot handler with version in u-boot
Blackfin arch: Faster Implementation of csum_tcpudp_nofold()
...
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
sparc64: Kill bogus TPC/address truncation during 32-bit faults.
sparc: fixup for sparseirq changes
sparc64: Validate kernel generated fault addresses on sparc64.
sparc64: On non-Niagara, need to touch NMI watchdog in NOHZ mode.
sparc64: Implement NMI watchdog on capable cpus.
sparc: Probe PMU type and record in sparc_pmu_type.
sparc64: Move generic PCR support code to seperate file.
On fast devices that go from congested to uncongested very quickly, pdflush
is waiting too often in congestion_wait, and the FS is backing off to
easily in write_cache_pages.
For now, fix this on the btrfs side by only checking congestion after
some bios have already gone down. Longer term a real fix is needed
for pdflush, but that is a larger project.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Whenever an item deletion is done, we need to balance all the nodes
in the tree to make sure we don't end up with an empty node if a pointer
is deleted. This balance prep happens from the root of the tree down
so we can drop our locks as we go.
reada_for_balance was triggering read-ahead on neighboring nodes even
when no balancing was required. This adds an extra check to avoid
calling balance_level() and avoid reada_for_balance() when a balance
won't be required.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_unlock_up_safe would break out at the first NULL node entry or
unlocked node it found in the path.
Some of the callers have missing nodes at the lower levels of the path, so this
commit fixes things to check all the nodes in the path before returning.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_del_leaf does two things. First it removes the pointer in the
parent, and then it frees the block that has the leaf. It has the
parent node locked for both operations.
But, it only needs the parent locked while it is deleting the pointer.
After that it can safely free the block without the parent locked.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
btrfs_truncate_inode_items is setup to stop doing btree searches when
it has finished removing the items for the inode. It used to detect the
end of the inode by looking for an objectid that didn't match the
one we were searching for.
But, this would result in an extra search through the btree, which
adds extra balancing and cow costs to the operation.
This commit adds a check to see if we found the inode item, which means
we can stop searching early.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
The compression code had some checks to make sure we were only
compressing bytes inside of i_size, but it wasn't catching every
case. To make things worse, some incorrect math about the number
of bytes remaining would make it try to compress more pages than the
file really had.
The fix used here is to fall back to the non-compression code in this
case, which does all the proper cleanup of delalloc and other accounting.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
With selinux on we end up calling __btrfs_setxattr when we create an inode,
which calls btrfs_start_transaction(). The problem is we've already called
that in btrfs_new_inode, and in btrfs_start_transaction we end up doing a
wait_current_trans(). If btrfs-transaction has started committing it will wait
for all handles to finish, while the other process is waiting for the
transaction to commit. This is fixed by using btrfs_join_transaction, which
won't wait for the transaction to commit. Thanks,
Signed-off-by: Josef Bacik <jbacik@redhat.com>
Before this patch, new files/dirs would ignore the SGID bit on their
parent directory and always be owned by the creating user's uid/gid.
Signed-off-by: Chris Ball <cjb@laptop.org>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Every transaction in btrfs creates a new snapshot, and then schedules the
snapshot from the last transaction for deletion. Snapshot deletion
works by walking down the btree and dropping the reference counts
on each btree block during the walk.
If if a given leaf or node has a reference count greater than one,
the reference count is decremented and the subtree pointed to by that
node is ignored.
If the reference count is one, walking continues down into that node
or leaf, and the references of everything it points to are decremented.
The old code would try to work in small pieces, walking down the tree
until it found the lowest leaf or node to free and then returning. This
was very friendly to the rest of the FS because it didn't have a huge
impact on other operations.
But it wouldn't always keep up with the rate that new commits added new
snapshots for deletion, and it wasn't very optimal for the extent
allocation tree because it wasn't finding leaves that were close together
on disk and processing them at the same time.
This changes things to walk down to a level 1 node and then process it
in bulk. All the leaf pointers are sorted and the leaves are dropped
in order based on their extent number.
The extent allocation tree and commit code are now fast enough for
this kind of bulk processing to work without slowing the rest of the FS
down. Overall it does less IO and is better able to keep up with
snapshot deletions under high load.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Most of the btrfs metadata operations can be protected by a spinlock,
but some operations still need to schedule.
So far, btrfs has been using a mutex along with a trylock loop,
most of the time it is able to avoid going for the full mutex, so
the trylock loop is a big performance gain.
This commit is step one for getting rid of the blocking locks entirely.
btrfs_tree_lock takes a spinlock, and the code explicitly switches
to a blocking lock when it starts an operation that can schedule.
We'll be able get rid of the blocking locks in smaller pieces over time.
Tracing allows us to find the most common cause of blocking, so we
can start with the hot spots first.
The basic idea is:
btrfs_tree_lock() returns with the spin lock held
btrfs_set_lock_blocking() sets the EXTENT_BUFFER_BLOCKING bit in
the extent buffer flags, and then drops the spin lock. The buffer is
still considered locked by all of the btrfs code.
If btrfs_tree_lock gets the spinlock but finds the blocking bit set, it drops
the spin lock and waits on a wait queue for the blocking bit to go away.
Much of the code that needs to set the blocking bit finishes without actually
blocking a good percentage of the time. So, an adaptive spin is still
used against the blocking bit to avoid very high context switch rates.
btrfs_clear_lock_blocking() clears the blocking bit and returns
with the spinlock held again.
btrfs_tree_unlock() can be called on either blocking or spinning locks,
it does the right thing based on the blocking bit.
ctree.c has a helper function to set/clear all the locked buffers in a
path as blocking.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
