Dmitry Monakhov [Tue, 16 Feb 2010 05:31:50 +0000 (08:31 +0300)]
quota: generalize quota transfer interface
Current quota transfer interface support only uid/gid.
This patch extend interface in order to support various quotas types
The goal is accomplished without changes in most frequently used
vfs_dq_transfer() func.
Jan Kara [Tue, 16 Feb 2010 19:37:12 +0000 (20:37 +0100)]
jbd: Delay discarding buffers in journal_unmap_buffer
Delay discarding buffers in journal_unmap_buffer until
we know that "add to orphan" operation has definitely been
committed, otherwise the log space of committing transation
may be freed and reused before truncate get committed, updates
may get lost if crash happens.
This patch is a backport of JBD2 fix by dingdinghua <[email protected]>.
Dmitry Monakhov [Tue, 16 Feb 2010 16:33:42 +0000 (19:33 +0300)]
ext3: quota_write cross block boundary behaviour
We always assume what dquot update result in changes in one data block
But ext3_quota_write() function may handle cross block boundary writes
In fact if this ever happen it will result in incorrect journal credits
reservation. And later bug_on triggering. As soon this never happen the
boundary cross loop is NOOP. In order to make things straight
let's remove this loop and assert cross boundary condition.
quota: remove invalid optimization from quota_sync_all
Checking the "VFS" quota enabled and dirty bits from generic code means
this code will never get called for other implementations, e.g. XFS and
GFS2. Grabbing the reference on the superblock really isn't much overhead
for a global Q_SYNC call, so just drop this optimization.
quota: move code from sync_quota_sb into vfs_quota_sync
Currenly sync_quota_sb does a lot of sync and truncate action that only
applies to "VFS" style quotas and is actively harmful for the sync
performance in XFS. Move it into vfs_quota_sync and add a wait parameter
to ->quota_sync to tell if we need it or not.
My audit of the GFS2 code says it's also not needed given the way GFS2
implements quotas, but I'd be happy if this can get a detailed review.
Currently Q_XQUOTASYNC calls into the quota_sync method, but XFS does something
entirely different in it than the rest of the filesystems. xfs_quota which
calls Q_XQUOTASYNC expects an asynchronous data writeout to flush delayed
allocations, while the "VFS" quota support wants to flush changes to the quota
file.
So make Q_XQUOTASYNC call into the writeback code directly and make the
quota_sync method optional as XFS doesn't need in the sense expected by the
rest of the quota code.
GFS2 was using limited XFS-style quota and has a quota_sync method fitting
neither the style used by vfs_quota_sync nor xfs_fs_quota_sync. I left it
in for now as per discussion with Steve it expects to be called from the
sync path this way.
Stop having complicated different routines for checking permissions for
XQM vs "VFS" quotas. Instead do the checks for having sb->s_qcop and
a valid type directly in do_quotactl, and munge the *quotactl_valid functions
into a check_quotactl_permission helper that only checks for permissions.
The Q_SYNC command can be called without the path to a device, in which case
it iterates over all superblocks. Special case this variant directly in
sys_quotactl so that the other code always gets a superblock and doesn't
need to deal with this case.
Jan Kara [Tue, 9 Feb 2010 17:20:39 +0000 (18:20 +0100)]
quota: Fix warning when a delayed write happens before quota is enabled
If a delayed-allocation write happens before quota is enabled, the
kernel spits out a warning:
WARNING: at fs/quota/dquot.c:988 dquot_claim_space+0x77/0x112()
because the fact that user has some delayed allocation is not recorded
in quota structure.
Make dquot_initialize() update amount of reserved space for user if it sees
inode has some space reserved. Also make sure that reserved quota space does
not go negative and we warn about the filesystem bug just once.
Dmitry Monakhov [Tue, 9 Feb 2010 16:53:36 +0000 (17:53 +0100)]
quota: manage reserved space when quota is not active [v2]
Since we implemented generic reserved space management interface,
then it is possible to account reserved space even when quota
is not active (similar to i_blocks/i_bytes).
Without this patch following testcase result in massive comlain from
WARN_ON in dquot_claim_space()
Christoph Egger [Fri, 5 Feb 2010 13:13:33 +0000 (14:13 +0100)]
jbd[2]: remove references to BUFFER_DEBUG
CONFIG_BUFFER_DEBUG seems to have been removed from the documentation
somewhere around 2.4.15 and seemingly hasn't been available even
longer. It is, however, still referenced at one place from the jbd
code (one is a copy of the other header). Time to clean it up
Dmitry Monakhov [Tue, 2 Feb 2010 13:05:53 +0000 (16:05 +0300)]
ext3: trivial quota cleanup
The patch is aimed to reorganize and simplify quota code a bit.
Quota code is itself complex enouth, but we can make it more readable
in some places:
- Move quota option parsing to separate functions.
- Simplify old-quota and journaled-quota mix check.
Jan Kara [Wed, 6 Jan 2010 20:58:48 +0000 (21:58 +0100)]
ext3: Use bitops to read/modify EXT3_I(inode)->i_state
At several places we modify EXT3_I(inode)->i_state without holding i_mutex
(ext3_release_file, ext3_bmap, ext3_journalled_writepage, ext3_do_update_inode,
...). These modifications are racy and we can lose updates to i_state. So
convert handling of i_state to use bitops which are atomic.
Jan Kara [Wed, 6 Jan 2010 16:20:35 +0000 (17:20 +0100)]
quota: Cleanup S_NOQUOTA handling
Cleanup handling of S_NOQUOTA inode flag and document it a bit. The flag
does not have to be set under dqptr_sem. Only functions modifying inode's
dquot pointers have to check the flag under dqptr_sem before going forward
with the modification. This way we are sure that we cannot add new dquot
pointers to the inode which is just becoming a quota file.
The good thing about this cleanup is that there are no more places in quota
code which enforce i_mutex vs. dqptr_sem lock ordering (in particular that
dqptr_sem -> i_mutex of quota file). This should silence some (false) lockdep
warnings with ext4 + quota and generally make life of some filesystems easier.
Linus Torvalds [Thu, 4 Mar 2010 16:26:08 +0000 (08:26 -0800)]
Merge branch 'for-linus' of git://git.open-osd.org/linux-open-osd
* 'for-linus' of git://git.open-osd.org/linux-open-osd:
exofs: groups support
exofs: Prepare for groups
exofs: Error recovery if object is missing from storage
exofs: convert io_state to use pages array instead of bio at input
exofs: RAID0 support
exofs: Define on-disk per-inode optional layout attribute
exofs: unindent exofs_sbi_read
exofs: Move layout related members to a layout structure
exofs: Recover in the case of read-passed-end-of-file
exofs: Micro-optimize exofs_i_info
exofs: debug print even less
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
sparc64: Make prom entry spinlock NMI safe.
sparc64: Kill off old sys_perfctr system call and state.
sparc: Update defconfigs.
sparc: Provide io{read,write}{16,32}be().
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/ide-next-2.6: (49 commits)
drivers/ide: Fix continuation line formats
ide: fixed section mismatch warning in cmd640.c
ide: ide_timing_compute() fixup
ide: make ide_get_best_pio_mode() static
via82cxxx: use ->pio_mode value to determine pair device speed
tx493xide: use ->pio_mode value to determine pair device speed
siimage: use ->pio_mode value to determine pair device speed
palm_bk3710: use ->pio_mode value to determine pair device speed
it821x: use ->pio_mode value to determine pair device speed
cs5536: use ->pio_mode value to determine pair device speed
cs5535: use ->pio_mode value to determine pair device speed
cmd64x: fix handling of address setup timings
amd74xx: use ->pio_mode value to determine pair device speed
alim15x3: fix handling of UDMA enable bit
alim15x3: fix handling of DMA timings
alim15x3: fix handling of command timings
alim15x3: fix handling of address setup timings
ide-timings: use ->pio_mode value to determine fastest PIO speed
ide: change ->set_dma_mode method parameters
ide: change ->set_pio_mode method parameters
...
Linus Torvalds [Thu, 4 Mar 2010 16:15:33 +0000 (08:15 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6: (52 commits)
init: Open /dev/console from rootfs
mqueue: fix typo "failues" -> "failures"
mqueue: only set error codes if they are really necessary
mqueue: simplify do_open() error handling
mqueue: apply mathematics distributivity on mq_bytes calculation
mqueue: remove unneeded info->messages initialization
mqueue: fix mq_open() file descriptor leak on user-space processes
fix race in d_splice_alias()
set S_DEAD on unlink() and non-directory rename() victims
vfs: add NOFOLLOW flag to umount(2)
get rid of ->mnt_parent in tomoyo/realpath
hppfs can use existing proc_mnt, no need for do_kern_mount() in there
Mirror MS_KERNMOUNT in ->mnt_flags
get rid of useless vfsmount_lock use in put_mnt_ns()
Take vfsmount_lock to fs/internal.h
get rid of insanity with namespace roots in tomoyo
take check for new events in namespace (guts of mounts_poll()) to namespace.c
Don't mess with generic_permission() under ->d_lock in hpfs
sanitize const/signedness for udf
nilfs: sanitize const/signedness in dealing with ->d_name.name
...
Fix up fairly trivial (famous last words...) conflicts in
drivers/infiniband/core/uverbs_main.c and security/tomoyo/realpath.c
Linus Torvalds [Thu, 4 Mar 2010 16:04:22 +0000 (08:04 -0800)]
Merge git://git.infradead.org/battery-2.6
* git://git.infradead.org/battery-2.6:
power_supply: bq27x00: fix voltage and current units
power_supply: bq27x00: add status and time properties
power_supply: bq27x00: add BQ27500 support
power_supply: bq27x00: fix temperature conversion
power_supply: bq27x00: remove unused struct fields
power_supply: bq27x00: remove double endian swap
da9030_battery: fix spelling in comment
wm97xx_battery: Clean up some warnings
Linus Torvalds [Thu, 4 Mar 2010 15:52:06 +0000 (07:52 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lrg/voltage-2.6
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lrg/voltage-2.6: (27 commits)
Regulators: wm8400 - cleanup platform driver data handling
Regulators: wm8994 - clean up driver data after removal
Regulators: wm831x-xxx - clean up driver data after removal
Regulators: pcap-regulator - clean up driver data after removal
Regulators: max8660 - annotate probe and remove methods
Regulators: max1586 - annotate probe and remove methods
Regulators: lp3971 - fail if platform data was not supplied
Regulators: tps6507x-regulator - mark probe method as __devinit
Regulators: tps65023-regulator - mark probe method as __devinit
Regulators: twl-regulator - mark probe function as __devinit
Regulators: fixed - annotate probe and remove methods
Regulators: ab3100 - fix probe and remove annotations
Regulators: virtual - use sysfs attribute groups
twl6030: regulator: Configure STATE register instead of REMAP
regulator: Provide optional dummy regulator for consumers
regulator: Assume regulators are enabled if they don't report anything
regulator: Convert fixed voltage regulator to use enable_time()
regulator: Add WM8994 regulator support
regulator: enable max8649 regulator driver
regulator: trivial: fix typos in user-visible Kconfig text
...
Linus Torvalds [Thu, 4 Mar 2010 15:49:37 +0000 (07:49 -0800)]
Merge branch 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6
* 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (151 commits)
vga_switcheroo: disable default y by new rules.
drm/nouveau: fix *staging* driver build with switcheroo off.
drm/radeon: fix typo in Makefile
vga_switcheroo: fix build on platforms with no ACPI
drm/radeon: Fix printf type warning in 64bit system.
drm/radeon/kms: bump the KMS version number for square tiling support.
vga_switcheroo: initial implementation (v15)
drm/radeon/kms: do not disable audio engine twice
Revert "drm/radeon/kms: disable HDMI audio for now on rv710/rv730"
drm/radeon/kms: do not preset audio stuff and start timer when not using audio
drm/radeon: r100/r200 ums: block ability for userspace app to trash 0 page and beyond
drm/ttm: fix function prototype to match implementation
drm/radeon: use ALIGN instead of open coding it
drm/radeon/kms: initialize set_surface_reg reg for rs600 asic
drm/i915: Use a dmi quirk to skip a broken SDVO TV output.
drm/i915: enable/disable LVDS port at DPMS time
drm/i915: check for multiple write domains in pin_and_relocate
drm/i915: clean-up i915_gem_flush_gpu_write_domain
drm/i915: reuse i915_gpu_idle helper
drm/i915: ensure lru ordering of fence_list
...
Fixed trivial conflicts in drivers/gpu/vga/Kconfig
The slab tree adds a percpu variable usage case (commit 9dfc6e68bfe6ee452efb1a4e9ca26a9007f2b864 "SLUB: Use this_cpu operations in
slub"), but the percpu tree removes the prefixing of percpu variables (commit dd17c8f72993f9461e9c19250e3f155d6d99df22 "percpu: remove per_cpu__ prefix"),
thus causing the following compilation error:
CC mm/slub.o
mm/slub.c: In function ‘alloc_kmem_cache_cpus’:
mm/slub.c:2078: error: implicit declaration of function ‘per_cpu_var’
mm/slub.c:2078: warning: assignment makes pointer from integer without a cast
make[1]: *** [mm/slub.o] Error 1
Akira Fujita [Thu, 4 Mar 2010 05:34:58 +0000 (00:34 -0500)]
ext4: Fix the NULL reference in double_down_write_data_sem()
If EXT4_IOC_MOVE_EXT ioctl is called with NULL donor_fd, fget() in
ext4_ioctl() gets inappropriate file structure for donor; so we need
to do this check earlier, before calling double_down_write_data_sem().
Akira Fujita [Thu, 4 Mar 2010 05:31:06 +0000 (00:31 -0500)]
ext4: Fix insertion point of extent in mext_insert_across_blocks()
If the leaf node has 2 extent space or fewer and EXT4_IOC_MOVE_EXT
ioctl is called with the file offset where after the 2nd extent
covers, mext_insert_across_blocks() always tries to insert extent into
the first extent. As a result, the file gets corrupted because of
wrong extent order. The patch fixes this problem.
padata: Allocate the cpumask for the padata instance
The cpumask of the padata instance was used without allocated.
This caused boot crashes if CONFIG_CPUMASK_OFFSTACK is enabled.
This patch fixes this by doing proper allocation for this cpumask.
Akinobu Mita [Thu, 4 Mar 2010 04:55:01 +0000 (23:55 -0500)]
ext4: consolidate in_range() definitions
There are duplicate macro definitions of in_range() in mballoc.h and
balloc.c. This consolidates these two definitions into ext4.h, and
changes extents.c to use in_range() as well.
Akinobu Mita [Thu, 4 Mar 2010 04:53:39 +0000 (23:53 -0500)]
ext4: cleanup to use ext4_group_first_block_no()
This is a cleanup and simplification patch which takes some open-coded
calculations to calculate the first block number of a group and
converts them to use the (already defined) ext4_group_first_block_no()
function.
Dan Williams [Thu, 4 Mar 2010 04:21:13 +0000 (21:21 -0700)]
ioat: cleanup ->timer_fn() and ->cleanup_fn() prototypes
If the calling convention of ->timer_fn() and ->cleanup_fn() are unified
across hardware versions we can drop parameters to ioat_init_channel() and
unify ioat_is_dma_complete() implementations.
Both ->timer_fn() and ->cleanup_fn() are modified to expect a struct
dma_chan pointer.
Dan Williams [Thu, 4 Mar 2010 04:21:13 +0000 (21:21 -0700)]
ioat3: interrupt coalescing
The hardware automatically disables further interrupts after each event
until rearmed. This allows a delay to be injected between the occurence
of the interrupt and the running of the cleanup routine. The delay is
scaled by the descriptor backlog and then written to the INTRDELAY
register which specifies the number of microseconds to hold off
interrupt delivery after an interrupt event occurs. According to
powertop this reduces the interrupt rate from ~5000 intr/s to ~150
intr/s per without affecting throughput (simple dd to a raid6 array).
Dan Williams [Thu, 4 Mar 2010 04:21:10 +0000 (21:21 -0700)]
ioat: close potential BUG_ON race in the descriptor cleanup path
Since ioat_cleanup_preamble() and the update of the last completed
descriptor are not synchronized there is a chance that two cleanup threads
can see descriptors to clean. If the first cleans up all pending
descriptors then the second will trigger the BUG_ON.
Linus Torvalds [Tue, 2 Mar 2010 16:36:46 +0000 (08:36 -0800)]
Prioritize synchronous signals over 'normal' signals
This makes sure that we pick the synchronous signals caused by a
processor fault over any pending regular asynchronous signals sent to
use by [t]kill().
This is not strictly required semantics, but it makes it _much_ easier
for programs like Wine that expect to find the fault information in the
signal stack.
Without this, if a non-synchronous signal gets picked first, the delayed
asynchronous signal will have its signal context pointing to the new
signal invocation, rather than the instruction that caused the SIGSEGV
or SIGBUS in the first place.
This is not all that pretty, and we're discussing making the synchronous
signals more explicit rather than have these kinds of implicit
preferences of SIGSEGV and friends. See for example
http://bugzilla.kernel.org/show_bug.cgi?id=15395
for some of the discussion. But in the meantime this is a simple and
fairly straightforward work-around, and the whole
if (x & Y)
x &= Y;
thing can be compiled into (and gcc does do it) just three instructions:
movq %rdx, %rax
andl $Y, %eax
cmovne %rax, %rdx
so it is at least a simple solution to a subtle issue.
Jan Kara [Wed, 3 Mar 2010 21:19:32 +0000 (16:19 -0500)]
ext4: Release page references acquired in ext4_da_block_invalidatepages
We forget to release page references we acquire in
ext4_da_block_invalidatepages. Luckily, this function gets called only if we
are not able to allocate blocks for delay-allocated data so that function
should better never be called.
To avoid potential problems with an empty /dev open /dev/console
from rootfs instead of waiting to mount our root filesystem and
mounting it there. This effectively guarantees that there will
be a device node, and it won't be on a filesystem that we will
ever unmount, so there are no issues with leaving /dev/console
open and pinning the filesystem.
This is actually more effective than automatically mounting
devtmpfs on /dev because it removes removes the occasionally
problematic assumption that /dev/console exists from the boot
code.
With this patch I was able to throw busybox on my /boot partition
(which has no /dev directory) and boot into userspace without
problems.
The only possible negative consequence I can think of is that
someone out there deliberately used did not use a character device
that is major 5 minor 2 for /dev/console. Does anyone know of a
situation in which that could make sense?
Al Viro [Wed, 3 Mar 2010 19:13:08 +0000 (14:13 -0500)]
fix race in d_splice_alias()
rehashing the negative placeholder opens a race with d_lookup();
we unhash it almost immediately (by d_move()), but the race
window is there. Since d_move() doesn't rely on target being
hashed, we don't need that d_rehash() at all.
Miklos Szeredi [Wed, 10 Feb 2010 11:15:53 +0000 (12:15 +0100)]
vfs: add NOFOLLOW flag to umount(2)
Add a new UMOUNT_NOFOLLOW flag to umount(2). This is needed to prevent
symlink attacks in unprivileged unmounts (fuse, samba, ncpfs).
Additionally, return -EINVAL if an unknown flag is used (and specify
an explicitly unused flag: UMOUNT_UNUSED). This makes it possible for
the caller to determine if a flag is supported or not.
Al Viro [Fri, 5 Feb 2010 07:21:06 +0000 (02:21 -0500)]
get rid of useless vfsmount_lock use in put_mnt_ns()
It hadn't been needed since we'd sanitized the logics in
mark_mounts_for_expiry() (which, in turn, used to be a
rudiment of bad old times when namespace_sem was per-ns).
Al Viro [Fri, 5 Feb 2010 06:41:33 +0000 (01:41 -0500)]
get rid of insanity with namespace roots in tomoyo
passing *any* namespace root to __d_path() as root is equivalent
to just passing it {NULL, NULL}; no need to bother with finding
the root of our namespace in there.
Al Viro [Sat, 30 Jan 2010 20:27:27 +0000 (15:27 -0500)]
fix mnt_mountpoint abuse in smack
(mnt,mnt_mountpoint) pair is conceptually wrong; if you want
to use it for generating pathname and for nothing else *and*
if you know that vfsmount tree is unchanging, you can get
away with that, but the right solution for that is (mnt,mnt_root).
Valerie Aurora [Tue, 26 Jan 2010 19:20:47 +0000 (14:20 -0500)]
VFS: Clean up shared mount flag propagation
The handling of mount flags in set_mnt_shared() got a little tangled
up during previous cleanups, with the following problems:
* MNT_PNODE_MASK is defined as a literal constant when it should be a
bitwise xor of other MNT_* flags
* set_mnt_shared() clears and then sets MNT_SHARED (part of MNT_PNODE_MASK)
* MNT_PNODE_MASK could use a comment in mount.h
* MNT_PNODE_MASK is a terrible name, change to MNT_SHARED_MASK
No one is calling this anymore as everyone has switched to
invalidate_mapping_pages long time ago. Also update a few
references to it in comments. nfs has two more, but I can't
easily figure what they are actually referring to, so I left
them as-is.
Al Viro [Sun, 24 Jan 2010 01:10:29 +0000 (20:10 -0500)]
Get rid of mnt_mountpoint abuses in ext4
path to mnt/mnt->mnt_root is no worse than that to
mnt->mnt_parent/mnt->mnt_mountpoint *and* needs no
pinning the sucker down (mnt is not going away and
mnt->mnt_root won't change)
Dan Williams [Wed, 3 Mar 2010 18:47:43 +0000 (11:47 -0700)]
ioat2: kill pending flag
The pending == 2 case no longer exists in the driver so, we can use
ioat2_ring_pending() outside the lock to determine if there might be any
descriptors in the ring that the hardware has not seen.
Al Viro [Sat, 16 Jan 2010 19:10:21 +0000 (14:10 -0500)]
add several pieces to shared subtree documentation
* document locking
* add the missing part of data structure invariants (relationship
between mnt_share and mnt_slave lists in case of a peer group
among slaves).
First of all, get_source() never results in CL_PROPAGATION
alone. We either get CL_MAKE_SHARED (for the continuation
of peer group) or CL_SLAVE (slave that is not shared) or both
(beginning of peer group among slaves). Massage the code to
make that explicit, kill CL_PROPAGATION test in clone_mnt()
(nothing sets CL_MAKE_SHARED without CL_PROPAGATION and in
clone_mnt() we are checking CL_PROPAGATION after we'd found
that there's no CL_SLAVE, so the check for CL_MAKE_SHARED
would do just as well).
Nick Piggin [Tue, 22 Dec 2009 00:28:53 +0000 (16:28 -0800)]
fs: improve remount,ro vs buffercache coherency
Invalidate sb->s_bdev on remount,ro.
Fixes a problem reported by Jorge Boncompte who is seeing corruption
trying to snapshot a minix filesystem image. Some filesystems modify
their metadata via a path other than the bdev buffer cache (eg. they may
use a private linear mapping for their metadata, or implement directories
in pagecache, etc). Also, file data modifications usually go to the bdev
via their own mappings.
These updates are not coherent with buffercache IO (eg. via /dev/bdev)
and never have been. However there could be a reasonable expectation that
after a mount -oremount,ro operation then the buffercache should
subsequently be coherent with previous filesystem modifications.
So invalidate the bdev mappings on a remount,ro operation to provide a
coherency point.
The problem was exposed when we switched the old rd to brd because old rd
didn't really function like a normal block device and updates to rd via
mappings other than the buffercache would still end up going into its
buffercache. But the same problem has always affected other "normal"
block devices, including loop.
Alas, ->d_revalidate() can't rely on ->lookup() finishing what
it's started; if d_alloc() in do_lookup() fails, we are not going
to call ->lookup() at all.
projects / linux.git / log