Currently we disable barriers as soon as we get a buffer in xlog_iodone
that has the XBF_ORDERED flag cleared. But this can be the case not only
for buffers where the barrier failed, but also the first buffer of a
split log write in case of a log wraparound. Due to the disabled
barriers we can easily get directory corruption on unclean shutdowns.
So instead of using this check add a new buffer flag for failed barrier
writes.
This is a regression vs 2.6.26 caused by patch to use the right macro
to check for the ORDERED flag, as we previously got true returned for
every buffer.
Thanks to Toei Rei for reporting the bug.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Sandeen <sandeen@sandeen.net>
Reviewed-by: David Chinner <david@fromorbit.com>
Signed-off-by: Tim Shimmin <tes@sgi.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-nmw:
GFS2: Support for I/O barriers
GFS2: Add UUID to GFS2 sb
GFS2: high time to take some time over atime
GFS2: The war on bloat
GFS2: GFS2 will panic if you misspell any mount options
GFS2: Direct IO write at end of file error
GFS2: Use an IS_ERR test rather than a NULL test
GFS2: Fix race relating to glock min-hold time
GFS2: Fix & clean up GFS2 rename
GFS2: rm on multiple nodes causes panic
GFS2: Fix metafs mounts
GFS2: Fix debugfs glock file iterator
* 'for-2.6.28' of git://git.kernel.dk/linux-2.6-block: (132 commits)
doc/cdrom: Trvial documentation error, file not present
block_dev: fix kernel-doc in new functions
block: add some comments around the bio read-write flags
block: mark bio_split_pool static
block: Find bio sector offset given idx and offset
block: gendisk integrity wrapper
block: Switch blk_integrity_compare from bdev to gendisk
block: Fix double put in blk_integrity_unregister
block: Introduce integrity data ownership flag
block: revert part of d7533ad0e132f92e75c1b2eb7c26387b25a583c1
bio.h: Remove unused conditional code
block: remove end_{queued|dequeued}_request()
block: change elevator to use __blk_end_request()
gdrom: change to use __blk_end_request()
memstick: change to use __blk_end_request()
virtio_blk: change to use __blk_end_request()
blktrace: use BLKTRACE_BDEV_SIZE as the name size for setup structure
block: add lld busy state exporting interface
block: Fix blk_start_queueing() to not kick a stopped queue
include blktrace_api.h in headers_install
...
This is debatable, but while we're debating it, let's disallow the
combination of splice and an O_APPEND destination.
It's not entirely clear what the semantics of O_APPEND should be, and
POSIX apparently expects pwrite() to ignore O_APPEND, for example. So
we could make up any semantics we want, including the old ones.
But Miklos convinced me that we should at least give it some thought,
and that accepting writes at arbitrary offsets is wrong at least for
IS_APPEND() files (which always have O_APPEND set, even if the reverse
isn't true: you can obviously have O_APPEND set on a regular file).
So disallow O_APPEND entirely for now. I doubt anybody cares, and this
way we have one less gray area to worry about.
Reported-and-argued-for-by: Miklos Szeredi <miklos@szeredi.hu>
Acked-by: Jens Axboe <ens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix kernel-doc in new functions:
Error(mmotm-2008-1002-1617//fs/block_dev.c:895): duplicate section name 'Description'
Error(mmotm-2008-1002-1617//fs/block_dev.c:924): duplicate section name 'Description'
Warning(mmotm-2008-1002-1617//fs/block_dev.c:1282): No description found for parameter 'pathname'
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
cc: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Since all bio_split calls refer the same single bio_split_pool, the bio_split
function can use bio_split_pool directly instead of the mempool_t parameter;
then the mempool_t parameter can be removed from bio_split param list, and
bio_split_pool is only referred in fs/bio.c file, can be marked static.
Signed-off-by: Denis ChengRq <crquan@gmail.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Helper function to find the sector offset in a bio given bvec index
and page offset.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
A filesystem might supply its own integrity metadata. Introduce a
flag that indicates whether the filesystem or the block layer owns the
integrity buffer.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Don't put functions that are only used in fs/bio-integrity.c in
blkdev.h, it's much cleaner to just keep it in there. Also kill
completely unused bdev_get_tag_size()
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Not all callers need (or want!) the mempool backing guarentee, it
essentially means that you can only use bio_alloc() for short allocations
and not for preallocating some bio's at setup or init time.
So add bio_kmalloc() which does the same thing as bio_alloc(), except
it just uses kmalloc() as the backing instead of the bio mempools.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
We call flush_disk() to make sure the buffer cache for the disk is
flushed after a disk resize. There are two resize cases, growing and
shrinking. Given that users can shrink/then grow a disk before
revalidate_disk() is called, we treat the grow case identically to
shrinking. We need to flush the buffer cache after an online shrink
because, as James Bottomley puts it,
The two use cases for shrinking I can see are
1. planned: the fs is already shrunk to within the new boundaries
and all data is relocated, so invalidate is fine (any dirty
buffers that might exist in the shrunk region are there only
because they were relocated but not yet written to their
original location).
2. unplanned: In this case, the fs is probably toast, so whether
we invalidate or not isn't going to make a whole lot of
difference; it's still going to try to read or write from
sectors beyond the new size and get I/O errors.
Immediately invalidating shrunk disks will cause errors for outstanding
I/Os for reads/write beyond the new end of the disk to be generated
earlier then if we waited for the normal buffer cache operation. It also
removes a potential security hole where we might keep old data around
from beyond the end of the shrunk disk if the disk was not invalidated.
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
We need to be able to flush the buffer cache for for more than
just when a disk is changed, so we factor out common cache flush code
in check_disk_change() to an internal flush_disk() routine. This
routine will then be used for both disk changes and disk resizes (in a
later patch).
Include the disk name in the text indicating that there are busy
inodes on the device and increase the KERN severity of the message.
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Check for device resize in the rescan_partitions() routine. If the device
has been resized, the bdev size is set to match. The rescan_partitions()
routine is called when opening the device and when calling the
BLKRRPART ioctl.
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
The revalidate_disk routine now checks if a disk has been resized by
comparing the gendisk capacity to the bdev inode size. If they are
different (usually because the disk has been resized underneath the kernel)
the bdev inode size is adjusted to match the capacity.
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This is a wrapper for the lower-level revalidate_disk call-backs such
as sd_revalidate_disk(). It allows us to perform pre and post
operations when calling them.
We will use this wrapper in a later patch to adjust block device sizes
after an online resize (a _post_ operation).
Signed-off-by: Andrew Patterson <andrew.patterson@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This patch changes blk_rq_map_user to accept a NULL user-space buffer
with a READ command if rq_map_data is not NULL. Thus a caller can pass
page frames to lk_rq_map_user to just set up a request and bios with
page frames propely. bio_uncopy_user (called via blk_rq_unmap_user)
doesn't copy data to user space with such request.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
bio_copy_kern and bio_copy_user are very similar. This converts
bio_copy_kern to use bio_copy_user.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This patch introduces struct rq_map_data to enable bio_copy_use_iov()
use reserved pages.
Currently, bio_copy_user_iov allocates bounce pages but
drivers/scsi/sg.c wants to allocate pages by itself and use
them. struct rq_map_data can be used to pass allocated pages to
bio_copy_user_iov.
The current users of bio_copy_user_iov simply passes NULL (they don't
want to use pre-allocated pages).
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Douglas Gilbert <dougg@torque.net>
Cc: Mike Christie <michaelc@cs.wisc.edu>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Currently, blk_rq_map_user and blk_rq_map_user_iov always do
GFP_KERNEL allocation.
This adds gfp_mask argument to blk_rq_map_user and blk_rq_map_user_iov
so sg can use it (sg always does GFP_ATOMIC allocation).
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Douglas Gilbert <dougg@torque.net>
Cc: Mike Christie <michaelc@cs.wisc.edu>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This patch adds support for controlling the IO completion CPU of
either all requests on a queue, or on a per-request basis. We export
a sysfs variable (rq_affinity) which, if set, migrates completions
of requests to the CPU that originally submitted it. A bio helper
(bio_set_completion_cpu()) is also added, so that queuers can ask
for completion on that specific CPU.
In testing, this has been show to cut the system time by as much
as 20-40% on synthetic workloads where CPU affinity is desired.
This requires a little help from the architecture, so it'll only
work as designed for archs that are using the new generic smp
helper infrastructure.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Now that disk and partition handlings are mostly unified, it's easy to
allow disk to have extended device number. This patch makes
add_disk() use extended device number if disk->minors is zero. Both
sd and ide-disk are updated to use this.
* sd_format_disk_name() is implemented which can generically determine
the drive name. This removes disk number restriction stemming from
limited device names.
* If sd index goes over SD_MAX_DISKS (which can be increased now BTW),
sd simply doesn't initialize minors letting block layer choose
extended device number.
* If CONFIG_DEBUG_EXT_DEVT is set, both sd and ide-disk always set
minors to 0 and use extended device numbers.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
With previous changes, it's meaningless to limit the number of
partitions. Replace @ext_minors with GENHD_FL_EXT_DEVT such that
setting the flag allows the disk to have maximum number of allowed
partitions (only limited by the number of entries in parsed_partitions
as determined by MAX_PART constant).
This kills not-too-pretty alloc_disk_ext[_node]() functions and makes
@minors parameter to alloc_disk[_node]() unnecessary. The parameter
is left alone to avoid disturbing the users.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
disk->__part used to be statically allocated to the maximum possible
number of partitions. This patch makes partition array allocation
dynamic. The added overhead is minimal as only real change is one
memory dereference changed to RCU one. This saves both a bit of
memory and cpu cycles iterating through unoccupied slots and makes
increasing partition limit easier.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Move stats related fields - stamp, in_flight, dkstats - from disk to
part0 and unify stat handling such that...
* part_stat_*() now updates part0 together if the specified partition
is not part0. ie. part_stat_*() are now essentially all_stat_*().
* {disk|all}_stat_*() are gone.
* part_round_stats() is updated similary. It handles part0 stats
automatically and disk_round_stats() is killed.
* part_{inc|dec}_in_fligh() is implemented which automatically updates
part0 stats for parts other than part0.
* disk_map_sector_rcu() is updated to return part0 if no part matches.
Combined with the above changes, this makes NULL special case
handling in callers unnecessary.
* Separate stats show code paths for disk are collapsed into part
stats show code paths.
* Rename disk_stat_lock/unlock() to part_stat_lock/unlock()
While at it, reposition stat handling macros a bit and add missing
parentheses around macro parameters.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
GENHD_FL_FAIL for disk is what make_it_fail is for parts. Kill it and
use part0->make_it_fail. Sysfs node handling is unified too.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Till now, bdev->bd_part is set only if the bdev was for parts other
than part0. This patch makes bdev->bd_part always set so that code
paths don't have to differenciate common handling.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Move disk->holder_dir to part0->holder_dir. Kill now mostly
superflous bdev_get_holder().
While at it, kill superflous kobject_get/put() around holder_dir,
slave_dir and cmd_filter creation and collapse
disk_sysfs_add_subdirs() into register_disk(). These serve no purpose
but obfuscating the code.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Now that capacity and __dev are moved to part0, part0 and others can
share the same method.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Move disk->capacity to part0->nr_sects and convert all users who
directly accessed the field to use {get|set}_capacity(). This is done
early to allow the __dev field to be moved.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
genhd and partition code handled disk and partitions separately. All
information about the whole disk was in struct genhd and partitions in
struct hd_struct. However, the whole disk (part0) and other
partitions have a lot in common and the data structures end up having
good number of common fields and thus separate code paths doing the
same thing. Also, the partition array was indexed by partno - 1 which
gets pretty confusing at times.
This patch introduces partition 0 and makes the partition array
indexed by partno. Following patches will unify the handling of disk
and parts piece-by-piece.
This patch also implements disk_partitionable() which tests whether a
disk is partitionable. With coming dynamic partition array change,
the most common usage of disk_max_parts() will be testing whether a
disk is partitionable and the number of max partitions will become
much less important.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Implement {disk|part}_to_dev() and use them to access generic device
instead of directly dereferencing {disk|part}->dev. To make sure no
user is left behind, rename generic devices fields to __dev.
This is in preparation of unifying partition 0 handling with other
partitions.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Implement extended device numbers. A block driver can tell block
layer that it wants to use extended device numbers. After the usual
minor space is used up, block layer automatically allocates devt's
from EXT_BLOCK_MAJOR.
Currently only one major number is allocated for this but as the
allocation is strictly on-demand, ~1mil minor space under it should
suffice unless the system actually has more than ~1mil partitions and
if that ever happens adding more majors to the extended devt area is
easy.
Due to internal implementation issues, the first partition can't be
allocated on the extended area. In other words, genhd->minors should
at least be 1. This limitation will be lifted by later changes.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
There are two variants of stat functions - ones prefixed with double
underbars which don't care about preemption and ones without which
disable preemption before manipulating per-cpu counters. It's unclear
whether the underbarred ones assume that preemtion is disabled on
entry as some callers don't do that.
This patch unifies diskstats access by implementing disk_stat_lock()
and disk_stat_unlock() which take care of both RCU (for partition
access) and preemption (for per-cpu counter access). diskstats access
should always be enclosed between the two functions. As such, there's
no need for the versions which disables preemption. They're removed
and double underbars ones are renamed to drop the underbars. As an
extra argument is added, there's no danger of using the old version
unconverted.
disk_stat_lock() uses get_cpu() and returns the cpu index and all
diskstat functions which access per-cpu counters now has @cpu
argument to help RT.
This change adds RCU or preemption operations at some places but also
collapses several preemption ops into one at others. Overall, the
performance difference should be negligible as all involved ops are
very lightweight per-cpu ones.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
disk->part[] is protected by its matching bdev's lock. However,
non-critical accesses like collecting stats and printing out sysfs and
proc information used to be performed without any locking. As
partitions can come and go dynamically, partitions can go away
underneath those non-critical accesses. As some of those accesses are
writes, this theoretically can lead to silent corruption.
This patch fixes the race by using RCU for the partition array and dev
reference counter to hold partitions.
* Rename disk->part[] to disk->__part[] to make sure no one outside
genhd layer proper accesses it directly.
* Use RCU for disk->__part[] dereferencing.
* Implement disk_{get|put}_part() which can be used to get and put
partitions from gendisk respectively.
* Iterators are implemented to help iterate through all partitions
safely.
* Functions which require RCU readlock are marked with _rcu suffix.
* Use disk_put_part() in __blkdev_put() instead of directly putting
the contained kobject.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
* Implement disk_devt() and part_devt() and use them to directly
access devt instead of computing it from ->major and ->first_minor.
Note that all references to ->major and ->first_minor outside of
block layer is used to determine devt of the disk (the part0) and as
->major and ->first_minor will continue to represent devt for the
disk, converting these users aren't strictly necessary. However,
convert them for consistency.
* Implement disk_max_parts() to avoid directly deferencing
genhd->minors.
* Update bdget_disk() such that it doesn't assume consecutive minor
space.
* Move devt computation from register_disk() to add_disk() and make it
the only one (all other usages use the initially determined value).
These changes clean up the code and will help disk->part dereference
fix and extended block device numbers.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
In hd_struct, @partno is used to denote partition number and a number
of other places use @part to denote hd_struct. Functions use @part
and @index instead. This causes confusion and makes it difficult to
use consistent variable names for hd_struct. Always use @partno if a
variable represents partition number.
Also, print out functions use @f or @part for seq_file argument. Use
@seqf uniformly instead.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
d805dda4 tried to fix error case handling in add_partition() but had a
few problems.
* disk->part[] entry is set early and left dangling if operation
fails.
* Once device initialized, the last put_device() is responsible for
freeing all the resources. The failure path freed part_stats and p
regardless of put_device() causing double free.
* holders subdir holds reference to the disk device, so failure path
should remove it to release resources properly which was missing.
This patch fixes the above problems and while at it move partition
slot busy check into add_partition() for completeness and inlines
holders subdirectory creation. Using separate function for it just
obfuscates the code.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Abdel Benamrouche <draconux@gmail.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
delete_partition() was noop for zero length partition. As the
addition code allows creating zero lenght partition and deletion is
assumed to always succeed, this causes memory leak for zero length
partitions. Allow zero length partitions to end their meaningless
lives.
While at it, allow deleting zero lenght partition via
BLKPG_DEL_PARTITION ioctl too.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Remove hw_segments field from struct bio and struct request. Without virtual
merge accounting they have no purpose.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
The previous patch db203d53d4 ("mm:
tiny-shmem fix lock ordering: mmap_sem vs i_mutex") to fix the lock
ordering in tiny-shmem breaks shared anonymous and IPC memory on NOMMU
architectures because it was using the expanding truncate to signal ramfs
to allocate a physically contiguous RAM backing the inode (otherwise it is
unusable for "memory mapping" it to userspace).
However do_truncate is what caused the lock ordering error, due to it
taking i_mutex. In this case, we can actually just call ramfs directly to
allocate memory for the mapping, rather than go via truncate.
Acked-by: David Howells <dhowells@redhat.com>
Acked-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Matt Mackall <mpm@selenic.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
There's a race between mm->owner assignment and swapoff, more easily
seen when task slab poisoning is turned on. The condition occurs when
try_to_unuse() runs in parallel with an exiting task. A similar race
can occur with callers of get_task_mm(), such as /proc/<pid>/<mmstats>
or ptrace or page migration.
CPU0 CPU1
try_to_unuse
looks at mm = task0->mm
increments mm->mm_users
task 0 exits
mm->owner needs to be updated, but no
new owner is found (mm_users > 1, but
no other task has task->mm = task0->mm)
mm_update_next_owner() leaves
mmput(mm) decrements mm->mm_users
task0 freed
dereferencing mm->owner fails
The fix is to notify the subsystem via mm_owner_changed callback(),
if no new owner is found, by specifying the new task as NULL.
Jiri Slaby:
mm->owner was set to NULL prior to calling cgroup_mm_owner_callbacks(), but
must be set after that, so as not to pass NULL as old owner causing oops.
Daisuke Nishimura:
mm_update_next_owner() may set mm->owner to NULL, but mem_cgroup_from_task()
and its callers need to take account of this situation to avoid oops.
Hugh Dickins:
Lockdep warning and hang below exec_mmap() when testing these patches.
exit_mm() up_reads mmap_sem before calling mm_update_next_owner(),
so exec_mmap() now needs to do the same. And with that repositioning,
there's now no point in mm_need_new_owner() allowing for NULL mm.
Reported-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The VFS interface for the 'd_compare()' is a bit special (read: 'odd'),
because it really just essentially replaces a memcmp(). The filesystem
is supposed to just compare the two names with whatever case-independent
or other function.
And when I say 'is supposed to', I obviously mean that 'procfs does odd
things, and actually looks at the dentry that we don't even pass down,
rather than just the name'. Which results in problems, because we
actually call d_compare before we have even verified that the dentry is
still hashed at all.
And that causes a problm since the inode that procfs looks at may have
been free'd and the d_inode pointer is NULL. procfs just assumes that
all dentries are positive, since procfs itself never generates a
negative one. But memory pressure will still result in the dentry
getting torn down, and as it is removed by RCU, it still remains visible
on some lists - and to d_compare.
If the filesystem just did a name comparison, we wouldn't care. And we
could just fix procfs to know about negative dentries too. But rather
than have the low-level filesystems know about internal VFS details,
just move the check for a unhashed dentry up a bit, so that we will only
call d_compare on dentries that are still active.
The actual oops this caused didn't look like a NULL pointer dereference
because procfs did a 'container_of(inode, struct proc_inode, vfs_inode)'
to get at its internal proc_inode information from the inode pointer,
and accessed a field below the inode. So the oops would look something
like
BUG: unable to handle kernel paging request at fffffffffffffff0
IP: [<ffffffff802bc6c6>] proc_sys_compare+0x36/0x50
and was seen on both x86-64 (Alexey Dobriyan and Hugh Dickins) and
ppc64 (Hugh Dickins).
Reported-by: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Hugh Dickins <hugh@veritas.com>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Reviewed-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-of-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'linux-next' of git://git.infradead.org/~dedekind/ubifs-2.6:
UBIFS: fix printk format warnings
UBIFS: remove incorrect assert
UBIFS: TNC / GC race fixes
UBIFS: create the name of the background thread in every case
Linus Torvalds
Christoph Hellwig
Randy Dunlap
Denis ChengRq
Martin K. Petersen
Jens Axboe
Andrew Patterson
FUJITA Tomonori
Tejun Heo
Mikulas Patocka
David Woodhouse
Nick Piggin
Nick Piggin
Balbir Singh