Al Viro
8f7b0ba1c8
Fix inotify watch removal/umount races ...
Inotify watch removals suck violently.
To kick the watch out we need (in this order) inode->inotify_mutex and
ih->mutex. That's fine if we have a hold on inode; however, for all
other cases we need to make damn sure we don't race with umount. We can
*NOT* just grab a reference to a watch - inotify_unmount_inodes() will
happily sail past it and we'll end with reference to inode potentially
outliving its superblock.
Ideally we just want to grab an active reference to superblock if we
can; that will make sure we won't go into inotify_umount_inodes() until
we are done. Cleanup is just deactivate_super().
However, that leaves a messy case - what if we *are* racing with
umount() and active references to superblock can't be acquired anymore?
We can bump ->s_count, grab ->s_umount, which will almost certainly wait
until the superblock is shut down and the watch in question is pining
for fjords. That's fine, but there is a problem - we might have hit the
window between ->s_active getting to 0 / ->s_count - below S_BIAS (i.e.
the moment when superblock is past the point of no return and is heading
for shutdown) and the moment when deactivate_super() acquires
->s_umount.
We could just do drop_super() yield() and retry, but that's rather
antisocial and this stuff is luser-triggerable. OTOH, having grabbed
->s_umount and having found that we'd got there first (i.e. that
->s_root is non-NULL) we know that we won't race with
inotify_umount_inodes().
So we could grab a reference to watch and do the rest as above, just
with drop_super() instead of deactivate_super(), right? Wrong. We had
to drop ih->mutex before we could grab ->s_umount. So the watch
could've been gone already.
That still can be dealt with - we need to save watch->wd, do idr_find()
and compare its result with our pointer. If they match, we either have
the damn thing still alive or we'd lost not one but two races at once,
the watch had been killed and a new one got created with the same ->wd
at the same address. That couldn't have happened in inotify_destroy(),
but inotify_rm_wd() could run into that. Still, "new one got created"
is not a problem - we have every right to kill it or leave it alone,
whatever's more convenient.
So we can use idr_find(...) == watch && watch->inode->i_sb == sb as
"grab it and kill it" check. If it's been our original watch, we are
fine, if it's a newcomer - nevermind, just pretend that we'd won the
race and kill the fscker anyway; we are safe since we know that its
superblock won't be going away.
And yes, this is far beyond mere "not very pretty"; so's the entire
concept of inotify to start with.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Greg KH <greg@kroah.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-11-15 12:26:44 -08:00
byteorder: provide swabb.h generically in asm/byteorder.h
2008-10-20 08:52:40 -07:00
V4L/DVB (9196): Add support for DSS delivery
2008-10-17 17:15:43 -03:00
mfd: twl4030 IRQ handling update
2008-10-22 01:19:37 +02:00
rtc: rtc-wm8350: add support for WM8350 RTC
2008-11-12 17:17:18 -08:00
mlx4_core: Multiple port type support
2008-10-22 15:38:42 -07:00
mmc: struct device - replace bus_id with dev_name(), dev_set_name()
2008-11-08 21:37:46 +01:00
[MTD] [NOR] Fix cfi_send_gen_cmd handling of x16 devices in x8 mode (v4)
2008-11-05 14:40:25 +01:00
netfilter: ctnetlink: remove bogus module dependency between ctnetlink and nf_nat
2008-10-14 11:58:31 -07:00
netfilter: xtables: move extension arguments into compound structure (5/6)
2008-10-08 11:35:19 +02:00
md: use sysfs_notify_dirent to notify changes to md/dev-xxx/state
2008-10-21 13:25:28 +11:00
regulator: Fix typo
2008-10-13 21:51:51 +01:00
spi: fix compile error
2008-10-30 11:38:47 -07:00
ssb: Fix DMA-API compilation for non-PCI systems
2008-11-10 13:50:19 -08:00
Merge branch 'next'
2008-10-15 15:54:56 -04:00
Merge branch 'master' into for-upstream
2008-10-20 16:07:19 +01:00
Merge branch 'video' into release
2008-11-11 21:15:50 -05:00
PCI: fix AER capability check
2008-10-20 11:01:52 -07:00
Configure out AIO support
2008-10-16 11:21:51 -07:00
ide: lba_capacity_is_ok() -> ata_id_is_lba_capacity_ok()
2008-10-10 22:39:31 +02:00
autofs4: add miscellaneous device for ioctls
2008-10-16 11:21:39 -07:00
autofs4: add miscellaneous device for ioctls
2008-10-16 11:21:39 -07:00
vmscan: split LRU lists into anon & file sets
2008-10-20 08:50:25 -07:00
include/linux/bcd.h: remove comments
2008-10-20 08:52:42 -07:00
Allow recursion in binfmt_script and binfmt_misc
2008-10-16 11:21:38 -07:00
bio: define __BIOVEC_PHYS_MERGEABLE
2008-11-06 08:41:55 +01:00
Merge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/dvrabel/uwb
2008-10-23 08:20:34 -07:00
[PATCH] end of methods switch: remove the old ones
2008-10-21 07:48:52 -04:00
blktrace: add support for driver data
2008-10-17 08:46:57 +02:00
fs: buffer lock use lock bitops
2008-10-20 08:52:32 -07:00
Add c2 port support
2008-11-12 17:17:18 -08:00
[PATCH] switch cdrom_{open,release,ioctl} to sane APIs
2008-10-21 07:47:22 -04:00
cgroups: tiny cleanups
2008-10-30 11:38:45 -07:00
container freezer: implement freezer cgroup subsystem
2008-10-20 08:52:34 -07:00
include/linux/clk.h: fix comment
2008-10-16 11:21:32 -07:00
clarify usage expectations for cnt32_to_63()
2008-11-09 11:17:33 -08:00
compat: generic compat get/settimeofday
2008-10-16 11:21:33 -07:00
ftrace: move notrace to compiler.h
2008-10-14 10:35:08 +02:00
Revert "vt: fix background color on line feed"
2008-10-14 12:12:02 -07:00
[CPUFREQ][1/6] cpufreq: Add cpu number parameter to __cpufreq_driver_getavg()
2008-10-09 13:52:43 -04:00
cpumask: introduce new API, without changing anything, v3
2008-11-09 21:09:54 +01:00
proc: move /proc/vmcore creation to fs/proc/vmcore.c
2008-10-23 18:51:22 +04:00
[PATCH vfs-2.6 2/6] vfs: add d_ancestor()
2008-10-23 05:13:16 -04:00
Merge git://git.kernel.org/pub/scm/linux/kernel/git/viro/bdev
2008-10-23 10:23:07 -07:00
i2c: Constify i2c_get_clientdata's parameter
2008-10-22 20:21:31 +02:00
Add an instance parameter devpts interfaces
2008-10-13 09:51:43 -07:00
dm raid1: separate region_hash interface part1
2008-10-21 17:45:06 +01:00
intel-iommu: IA64 support
2008-10-18 14:29:15 +01:00
dmar: fix using early fixmap mapping for DMAR table parsing
2008-10-16 16:53:04 +02:00
Automatic MODULE_ALIAS() for DMI match tables.
2008-10-13 16:05:06 +01:00
MIPS: IP22/28: Switch over to RTC class driver
2008-10-15 12:46:51 +01:00
driver core: basic infrastructure for per-module dynamic debug messages
2008-10-16 09:24:47 -07:00
x86: Add UV EFI table entry v4
2008-10-16 16:53:13 +02:00
block: Add interface to abort queued requests
2008-10-09 08:56:13 +02:00
x86: ioperm user_regset
2008-10-12 12:05:55 +02:00
include: replace __FUNCTION__ with __func__
2008-10-16 11:21:30 -07:00
ext3: add an option to control error handling on file data
2008-10-20 08:52:37 -07:00
include: replace __FUNCTION__ with __func__
2008-10-16 11:21:30 -07:00
fb: convert lock/unlock_kernel() into local fb mutex
2008-10-20 08:52:36 -07:00
floppy: support arbitrary first-sector numbers
2008-10-09 08:56:19 +02:00
[PATCH] introduce fmode_t, do annotations
2008-10-21 07:47:06 -04:00
firewire: Add more documentation to firewire-cdev.h
2008-10-15 22:21:11 +02:00
freezer_cg: use thaw_process() in unfreeze_cgroup()
2008-10-30 11:38:45 -07:00
fs: remove prepare_write/commit_write
2008-10-30 11:38:45 -07:00
gianfar: Fix race in TBI/SerDes configuration
2008-10-31 00:59:46 -04:00
[PATCH] introduce fmode_t, do annotations
2008-10-21 07:47:06 -04:00
ftrace: remove ftrace hash
2008-10-23 16:00:24 +02:00
fuse: implement nonseekable open
2008-10-16 16:08:57 +02:00
proc: move /proc/diskstats boilerplate to block/genhd.c
2008-10-23 17:57:37 +04:00
gpio_free might sleep, generic part
2008-10-16 11:21:40 -07:00
HID: fix oops during suspend of unbound HID devices
2008-10-27 15:06:01 +01:00
HID: hiddev.h: Fix example code.
2008-10-14 23:51:02 +02:00
hrtimer: clean up unused callback modes
2008-11-12 09:54:40 +01:00
proc: switch /proc/meminfo to seq_file
2008-10-23 13:52:40 +04:00
i2c-algo-pcf: Add adapter hooks around xfer begin and end
2008-10-22 20:21:30 +02:00
V4L/DVB (9240): saa7127: Fix two typos
2008-10-17 17:25:11 -03:00
i2c: Add info->archdata field
2008-10-22 20:21:33 +02:00
i2o: Fix 32/64bit DMA locking
2008-10-16 11:21:38 -07:00
i7300_idle: Disable ioat channel only on platforms where ile driver can load
2008-10-24 12:54:18 -04:00
IPV6: Fix default gateway criteria wrt. HIGH/LOW preference radv option
2008-10-15 16:03:01 -07:00
[PATCH] switch ide_disk_ops ->ioctl() to sane prototype
2008-10-21 07:47:30 -04:00
net: Rationalise email address: Network Specific Parts
2008-10-13 19:01:08 -07:00
net: Rationalise email address: Network Specific Parts
2008-10-13 19:01:08 -07:00
net: Rationalise email address: Network Specific Parts
2008-10-13 19:01:08 -07:00
gre: Add netlink interface
2008-10-09 11:59:55 -07:00
net: fix packet socket delivery in rx irq handler
2008-11-04 14:49:57 -08:00
net: Rationalise email address: Network Specific Parts
2008-10-13 19:01:08 -07:00
adjust init section definitions
2008-10-29 22:02:09 +01:00
Fix inotify watch removal/umount races
2008-11-15 12:26:44 -08:00
intel-iommu: IA64 support
2008-10-18 14:29:15 +01:00
Merge branch 'genirq-v28-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip
2008-10-20 13:23:01 -07:00
io mapping: clean up #ifdefs
2008-11-03 18:21:45 +01:00
introduce generic iommu_num_pages function
2008-10-16 11:21:33 -07:00
Merge branch 'core-v28-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip
2008-10-16 15:17:40 -07:00
VT-d: Changes to support KVM
2008-10-15 14:24:08 +02:00
genirq: improve include files
2008-10-18 14:05:18 +02:00
genirq: fix off by one and coding style
2008-10-21 15:54:40 +02:00
V4L/DVB (8959): include <linux/videodev2.h> into linux/ivtv.h
2008-10-12 09:37:01 -02:00
jbd: fix error handling for checkpoint io
2008-10-23 08:55:01 -07:00
ext4: Replace hackish ext4_mb_poll_new_transaction with commit callback
2008-10-16 20:00:24 -04:00
Remove Andrew Morton's old email accounts
2008-10-16 11:21:32 -07:00
vsprintf: use new vsprintf symbolic function pointer format
2008-10-16 11:21:31 -07:00
byteorder: add new headers for make headers-install
2008-10-20 08:52:40 -07:00
Add 'pr_fmt()' format modifier to pr_xyz macros.
2008-11-15 11:43:37 -08:00
Merge branch 'genirq-v28-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip
2008-10-20 13:23:01 -07:00
add key_revoke() dummy for KEYS=n
2008-10-12 11:44:37 -07:00
klist: don't iterate over deleted entries
2008-10-09 08:56:04 +02:00
pm: rework disabling of user mode helpers during suspend/hibernation
2008-10-16 11:21:29 -07:00
ftrace: ignore functions that cannot be kprobe-ed
2008-10-14 10:34:22 +02:00
KVM: Future-proof device assignment ABI
2008-10-28 14:22:15 +02:00
KVM: Fix guest shared interrupt with in-kernel irqchip
2008-10-28 14:21:34 +02:00
leds: Make default trigger fields const
2008-10-20 22:34:12 +01:00
libata: revert convert-to-block-tagging patches
2008-11-10 08:04:47 -08:00
ftrace: move notrace to compiler.h
2008-10-14 10:35:08 +02:00
integrity: special fs magic
2008-10-13 09:47:43 +11:00
block: implement extended dev numbers
2008-10-09 08:56:06 +02:00
Input: move map_to_7segment.h to include/linux
2008-10-10 23:16:16 -04:00
markers: fix synchronize marker unregister static inline
2008-10-14 10:39:28 +02:00
memcg: allocate all page_cgroup at boot
2008-10-20 08:52:39 -07:00
vmscan: move isolate_lru_page() to vmscan.c
2008-10-20 08:50:25 -07:00
mlock: mlocked pages are unevictable
2008-10-20 08:52:30 -07:00
vmscan: kill unused lru functions
2008-10-20 08:52:31 -07:00
memcg: allocate all page_cgroup at boot
2008-10-20 08:52:39 -07:00
mmiotrace: remove left-over marker cruft
2008-10-14 10:37:17 +02:00
memcg: allocate all page_cgroup at boot
2008-10-20 08:52:39 -07:00
HID: fix hid_device_id for cross compiling
2008-10-29 00:51:25 +01:00
param: Fix duplicate module prefixes
2008-10-22 10:00:23 +11:00
core_param() for genuinely core kernel parameters
2008-10-22 10:00:23 +11:00
include/linux/mount.h: remove CVS keyword
2008-10-16 11:21:30 -07:00
fat: Cleanup FAT attribute stuff
2008-11-06 15:41:21 -08:00
[PATCH vfs-2.6 6/6] vfs: add LOOKUP_RENAME_TARGET intent
2008-10-23 05:13:20 -04:00
net: delete excess kernel-doc notation
2008-10-30 23:54:35 -07:00
Make nfs_file_cred more robust.
2008-10-17 13:06:45 -04:00
NFS: Fix the resolution problem with nfs_inode_attrs_need_update()
2008-10-14 19:23:17 -04:00
of: Add new helper of_parse_phandles_with_args()
2008-10-13 10:55:47 +11:00
powerpc/oprofile: Fix mutex locking for cell spu-oprofile
2008-10-21 15:17:48 +11:00
mlock: mlocked pages are unevictable
2008-10-20 08:52:30 -07:00
memcg: fix page_cgroup allocation
2008-10-23 08:55:02 -07:00
mm: page lock use lock bitops
2008-10-20 08:52:32 -07:00
Unevictable LRU Infrastructure
2008-10-20 08:50:26 -07:00
parport: remove CVS keywords
2008-10-16 11:21:49 -07:00
vfs: Use const for kernel parser table
2008-10-13 10:10:37 -07:00
PCI: annotate return value of pci_ioremap_bar with __iomem
2008-11-03 14:31:18 -08:00
PCI: Hotplug core: remove 'name'
2008-10-22 16:42:43 -07:00
Merge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6
2008-10-23 10:20:36 -07:00
PCI: add support for function level reset
2008-10-22 16:42:35 -07:00
Phonet: do not reply to indication reset packets
2008-10-26 23:07:25 -07:00
phylib: add mdiobus_{read,write}
2008-10-08 16:38:41 -07:00
pid_ns: kill the now unused task_child_reaper()
2008-10-16 11:21:48 -07:00
platform: add new device registration helper
2008-10-16 09:24:51 -07:00
include: replace __FUNCTION__ with __func__
2008-10-16 11:21:30 -07:00
Merge branch 'linus' into test
2008-10-23 00:11:07 -04:00
proc: remove fs/proc/proc_misc.c
2008-10-23 18:54:05 +04:00
profiling: fix up CONFIG_PROC_FS=n build
2008-10-23 08:55:01 -07:00
make ptrace_untrace() static
2008-10-20 08:52:39 -07:00
quota: remove CVS keywords
2008-10-16 11:21:46 -07:00
quota: remove CVS keywords
2008-10-16 11:21:46 -07:00
remove ratelimt()
2008-11-12 17:17:17 -08:00
include: replace __FUNCTION__ with __func__
2008-10-16 11:21:30 -07:00
[PATCH] remember mode of reiserfs journal
2008-10-21 07:49:04 -04:00
mm: increase the default mlock limit from 32k to 64k
2008-10-30 11:38:45 -07:00
ring_buffer: implement new locking
2008-10-14 10:39:05 +02:00
make mm/rmap.c:anon_vma_cachep static
2008-10-20 08:52:40 -07:00
include: replace __FUNCTION__ with __func__
2008-10-16 11:21:30 -07:00
fix for account_group_exec_runtime(), make sure ->signal can't be freed under rq->lock
2008-11-11 08:01:43 +01:00
nfsd: fix vm overcommit crash
2008-10-30 11:38:47 -07:00
seq_file: add seq_cpumask_list(), seq_nodemask_list()
2008-10-20 08:52:39 -07:00
tty: Cris has a nice RS485 ioctl so we should steal it
2008-10-13 09:51:40 -07:00
serial: sh-sci: fix cannot work SH7723 SCIFA
2008-11-12 12:29:56 +09:00
slab: document SLAB_DESTROY_BY_RCU
2008-11-13 20:49:02 +02:00
smc911x: Add IRQ polarity configuration
2008-10-22 07:00:38 -04:00
cpumask: introduce new API, without changing anything
2008-11-06 09:05:33 +01:00
linux/string.h: fix comment typo
2008-11-02 10:15:07 -08:00
byteorder: use generic C version for value byteswapping
2008-10-20 08:52:40 -07:00
vmscan: unevictable LRU scan sysctl
2008-10-20 08:52:31 -07:00
introduce generic header file for the software IO/TLB
2008-10-16 11:21:28 -07:00
sysctl: simplify ->strategy
2008-10-16 11:21:47 -07:00
x86: sysfs: kill owner field from attribute
2008-10-20 08:52:42 -07:00
Remove Andrew Morton's old email accounts
2008-10-16 11:21:32 -07:00
telephony: trivial: fix up email address
2008-11-11 09:30:23 -08:00
tty: Add termiox
2008-10-13 09:51:40 -07:00
NOHZ: unify the nohz function calls in irq_enter()
2008-10-17 18:13:38 +02:00
Merge branch 'timers/range-hrtimers' into v28-range-hrtimers-for-linus-v2
2008-10-22 09:48:06 +02:00
Add round_jiffies_up and related routines
2008-11-06 08:42:48 +01:00
sched: fine-tune SD_SIBLING_INIT
2008-11-07 16:09:23 +01:00
tracepoints: synchronize unregister static inline
2008-10-14 10:39:31 +02:00
tty: some ICANON magic is in the wrong places
2008-10-13 09:51:44 -07:00
Add an instance parameter devpts interfaces
2008-10-13 09:51:43 -07:00
[PATCH] introduce fmode_t, do annotations
2008-10-21 07:47:06 -04:00
USB: don't register endpoints for interfaces that are going away
2008-11-13 14:45:00 -08:00
V4L/DVB (9320): v4l2: Add 10-bit RAW Bayer formats
2008-10-21 14:31:15 -02:00
proc: move /proc/vmallocinfo to mm/vmalloc.c
2008-10-23 15:48:28 +04:00
proc: move /proc/zoneinfo boilerplate to mm/vmstat.c
2008-10-23 17:35:04 +04:00
tty: Remove more special casing and out of place code
2008-10-13 09:51:42 -07:00
wait: kill is_sync_wait()
2008-10-16 11:21:31 -07:00
cpumask: introduce new API, without changing anything
2008-11-06 09:05:33 +01:00
vfs: Add no_nrwrite_index_update writeback control flag
2008-10-16 10:09:17 -04:00
8f7b0ba1c8