openkylin
/
linux
mirror of https://gitee.com/openkylin/linux.git
9
0
Fork 0
Code Issues Projects Releases Wiki Activity
linux/fs
History
Linus Torvalds 2b047252d0 Fix TLB gather virtual address range invalidation corner cases 
Ben Tebulin reported:

 "Since v3.7.2 on two independent machines a very specific Git
  repository fails in 9/10 cases on git-fsck due to an SHA1/memory
  failures.  This only occurs on a very specific repository and can be
  reproduced stably on two independent laptops.  Git mailing list ran
  out of ideas and for me this looks like some very exotic kernel issue"

and bisected the failure to the backport of commit 53a59fc67f ("mm:
limit mmu_gather batching to fix soft lockups on !CONFIG_PREEMPT").

That commit itself is not actually buggy, but what it does is to make it
much more likely to hit the partial TLB invalidation case, since it
introduces a new case in tlb_next_batch() that previously only ever
happened when running out of memory.

The real bug is that the TLB gather virtual memory range setup is subtly
buggered.  It was introduced in commit 597e1c3580 ("mm/mmu_gather:
enable tlb flush range in generic mmu_gather"), and the range handling
was already fixed at least once in commit e6c495a96c ("mm: fix the TLB
range flushed when __tlb_remove_page() runs out of slots"), but that fix
was not complete.

The problem with the TLB gather virtual address range is that it isn't
set up by the initial tlb_gather_mmu() initialization (which didn't get
the TLB range information), but it is set up ad-hoc later by the
functions that actually flush the TLB.  And so any such case that forgot
to update the TLB range entries would potentially miss TLB invalidates.

Rather than try to figure out exactly which particular ad-hoc range
setup was missing (I personally suspect it's the hugetlb case in
zap_huge_pmd(), which didn't have the same logic as zap_pte_range()
did), this patch just gets rid of the problem at the source: make the
TLB range information available to tlb_gather_mmu(), and initialize it
when initializing all the other tlb gather fields.

This makes the patch larger, but conceptually much simpler.  And the end
result is much more understandable; even if you want to play games with
partial ranges when invalidating the TLB contents in chunks, now the
range information is always there, and anybody who doesn't want to
bother with it won't introduce subtle bugs.

Ben verified that this fixes his problem.

Reported-bisected-and-tested-by: Ben Tebulin <tebulin@googlemail.com>
Build-testing-by: Stephen Rothwell <sfr@canb.auug.org.au>
Build-testing-by: Richard Weinberger <richard.weinberger@gmail.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
 		2013-08-16 08:52:46 -07:00
..
9p Second round of 9p patches for the 3.11 merge window. 2013-07-11 10:21:23 -07:00
adfs Don't pass inode to ->d_hash() and ->d_compare() 2013-06-29 12:57:36 +04:00
affs Don't pass inode to ->d_hash() and ->d_compare() 2013-06-29 12:57:36 +04:00
afs Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2013-07-03 09:10:19 -07:00
autofs4 helper for reading ->d_count 2013-07-05 18:59:33 +04:00
befs [readdir] convert befs 2013-06-29 12:56:55 +04:00
bfs [readdir] convert bfs 2013-06-29 12:56:33 +04:00
btrfs btrfs: don't loop on large offsets in readdir 2013-08-09 19:34:56 -04:00
cachefiles mm: remove lru parameter from __pagevec_lru_add and remove parts of pagevec API 2013-07-03 16:07:31 -07:00
ceph Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client 2013-07-09 12:39:10 -07:00
cifs cifs: don't instantiate new dentries in readdir for inodes that need to be revalidated immediately 2013-08-07 10:57:06 -05:00
coda helper for reading ->d_count 2013-07-05 18:59:33 +04:00
configfs Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2013-07-14 11:42:26 -07:00
cramfs [readdir] convert f2fs 2013-06-29 12:56:46 +04:00
debugfs debugfs: debugfs_remove_recursive() must not rely on list_empty(d_subdirs) 2013-07-31 12:16:31 -04:00
devpts fs: Limit sys_mount to only request filesystem modules (Part 2). 2013-03-07 01:08:55 -08:00
dlm dlm: kill the unnecessary and wrong device_close()->recalc_sigpending() 2013-08-09 10:48:20 -07:00
ecryptfs Code cleanups and improved buffer handling during page crypto operations 2013-07-11 10:20:18 -07:00
efivarfs efivarfs: we can use simple_lookup() now 2013-07-14 17:48:35 +04:00
efs [readdir] convert efs 2013-06-29 12:56:31 +04:00
exofs Lots of bug fixes, cleanups and optimizations. In the bug fixes 2013-07-02 09:39:34 -07:00
exportfs [readdir] constify ->actor 2013-06-29 12:57:05 +04:00
ext2 [O_TMPFILE] it's still short a few helpers, but infrastructure should be OK now... 2013-06-29 12:57:10 +04:00
ext3 ext3: fix a BUG when opening a file with O_TMPFILE flag 2013-07-20 22:03:20 -04:00
ext4 ext4: flush the extent status cache during EXT4_IOC_SWAP_BOOT 2013-08-12 09:29:30 -04:00
f2fs f2fs: fix readdir incorrectness 2013-07-08 13:35:48 +04:00
fat fatfs: add FAT_IOCTL_GET_VOLUME_ID 2013-07-09 10:33:25 -07:00
freevxfs [readdir] convert freevxfs 2013-06-29 12:56:53 +04:00
fscache FS-Cache: Don't use spin_is_locked() in assertions 2013-06-19 14:16:47 +01:00
fuse fuse: readdirplus: cleanup 2013-07-17 14:53:54 +02:00
gfs2 Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2013-07-03 09:10:19 -07:00
hfs Don't pass inode to ->d_hash() and ->d_compare() 2013-06-29 12:57:36 +04:00
hfsplus Don't pass inode to ->d_hash() and ->d_compare() 2013-06-29 12:57:36 +04:00
hostfs [readdir] convert hostfs 2013-06-29 12:56:59 +04:00
hpfs Merge branch 'hpfs' from Mikulas Patocka 2013-07-04 11:22:55 -07:00
hppfs clean up scary strncpy(dst, src, strlen(src)) uses 2013-07-03 16:07:41 -07:00
hugetlbfs hugetlb: fix lockdep splat caused by pmd sharing 2013-08-13 17:57:48 -07:00
isofs Don't pass inode to ->d_hash() and ->d_compare() 2013-06-29 12:57:36 +04:00
jbd jbd: change journal_invalidatepage() to accept length 2013-05-21 23:26:36 -04:00
jbd2 jbd2: invalidate handle if jbd2_journal_restart() fails 2013-07-01 08:12:41 -04:00
jffs2 [readdir] convert jffs2 2013-06-29 12:56:47 +04:00
jfs A couple cleanups to JFS for 3.11 2013-07-11 10:19:34 -07:00
lockd LOCKD: Don't call utsname()->nodename from nlmclnt_setlockargs 2013-08-05 15:03:46 -04:00
logfs Lots of bug fixes, cleanups and optimizations. In the bug fixes 2013-07-02 09:39:34 -07:00
minix minix: bug widening a binary "not" operation 2013-06-29 12:57:35 +04:00
ncpfs ncpfs: fix error return code in ncp_parse_options() 2013-07-09 10:33:25 -07:00
nfs NFSv4: Fix up nfs4_proc_lookup_mountpoint 2013-08-07 20:47:26 -04:00
nfs_common
…
nfsd nfsd: Fix SP4_MACH_CRED negotiation in EXCHANGE_ID 2013-08-07 12:06:07 -04:00
nilfs2 helper for reading ->d_count 2013-07-05 18:59:33 +04:00
nls
…
notify fsnotify: update comments concerning locking scheme 2013-07-09 10:33:20 -07:00
ntfs Lots of bug fixes, cleanups and optimizations. In the bug fixes 2013-07-02 09:39:34 -07:00
ocfs2 ocfs2: fix null pointer dereference in ocfs2_dir_foreach_blk_id() 2013-08-13 17:57:49 -07:00
omfs [readdir] convert omfs 2013-06-29 12:56:37 +04:00
openpromfs [readdir] convert openpromfs 2013-06-29 12:56:32 +04:00
proc fs/proc/task_mmu.c: fix buffer overflow in add_page_map() 2013-08-13 17:57:50 -07:00
pstore Merge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc 2013-07-04 10:29:23 -07:00
qnx4 [readdir] convert qnx4 2013-06-29 12:56:38 +04:00
qnx6 [readdir] convert qnx6 2013-06-29 12:56:39 +04:00
quota quota: Convert use of typedef ctl_table to struct ctl_table 2013-07-04 19:22:55 +02:00
ramfs Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2013-02-26 20:16:07 -08:00
reiserfs reiserfs: fix deadlock in umount 2013-08-05 17:37:37 +04:00
romfs [readdir] convert romfs 2013-06-29 12:56:29 +04:00
squashfs [readdir] convert squashfs 2013-06-29 12:56:28 +04:00
sysfs sysfs: prevent warning when only using binary attributes 2013-07-16 10:57:36 -07:00
sysv Don't pass inode to ->d_hash() and ->d_compare() 2013-06-29 12:57:36 +04:00
ubifs Only a single patch which fixes a message. 2013-07-05 12:08:47 -07:00
udf udf: provide ->tmpfile() 2013-06-29 12:57:12 +04:00
ufs [readdir] simple local unixlike: switch to ->iterate() 2013-06-29 12:46:47 +04:00
xfs xfs: di_flushiter considered harmful 2013-07-25 10:41:42 -05:00
Kconfig efivarfs: Move to fs/efivarfs 2013-04-17 13:25:09 +01:00
Kconfig.binfmt fs: make binfmt support for #! scripts modular and removable 2013-04-30 17:04:04 -07:00
Makefile Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2013-05-01 17:51:54 -07:00
aio.c aio: fix wrong comment in aio_complete() 2013-07-03 16:08:06 -07:00
anon_inodes.c get_empty_filp()/alloc_file() leave both ->f_pos and ->f_version zero 2013-02-26 02:46:11 -05:00
attr.c
…
bad_inode.c [readdir] ->readdir() is gone 2013-06-29 12:57:04 +04:00
binfmt_aout.c mm: remove free_area_cache 2013-07-10 18:11:34 -07:00
binfmt_elf.c mm: remove free_area_cache 2013-07-10 18:11:34 -07:00
binfmt_elf_fdpic.c Merge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc 2013-05-02 10:16:16 -07:00
binfmt_em86.c
…
binfmt_flat.c new helper: read_code() 2013-04-29 15:40:23 -04:00
binfmt_misc.c binfmt_misc: reuse string_unescape_inplace() 2013-04-30 17:04:03 -07:00
binfmt_script.c
…
binfmt_som.c
…
bio-integrity.c bio-integrity: Add explicit field for owner of bip_buf 2013-03-23 14:26:34 -07:00
bio.c Merge branch 'for-3.10/core' of git://git.kernel.dk/linux-block 2013-05-08 10:13:35 -07:00
block_dev.c Merge branch 'for-3.11/core' of git://git.kernel.dk/linux-block 2013-07-11 13:03:24 -07:00
buffer.c mm: vmscan: take page buffers dirty and locked state into account 2013-07-03 16:07:29 -07:00
char_dev.c
…
compat.c [readdir] constify ->actor 2013-06-29 12:57:05 +04:00
compat_binfmt_elf.c
…
compat_ioctl.c compat.c: LOOP_CLR_FD is taken care of in loop.c itself... 2013-06-29 12:46:44 +04:00
coredump.c coredump: '% at the end' shouldn't bypass core_uses_pid logic 2013-07-03 16:08:02 -07:00
coredump.h
…
dcache.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2013-07-03 09:10:19 -07:00
dcookies.c consolidate compat lookup_dcookie() 2013-03-03 23:00:23 -05:00
direct-io.c Merge branch 'for-3.10/core' of git://git.kernel.dk/linux-block 2013-05-08 10:13:35 -07:00
drop_caches.c
…
eventfd.c
…
eventpoll.c Merge branch 'akpm' (updates from Andrew Morton) 2013-07-03 17:12:13 -07:00
exec.c Fix TLB gather virtual address range invalidation corner cases 2013-08-16 08:52:46 -07:00
fcntl.c vfs: add missing check for __O_TMPFILE in fcntl_init() 2013-08-05 18:25:32 +04:00
fhandle.c
…
file.c don't bother with deferred freeing of fdtables 2013-05-01 17:31:42 -04:00
file_table.c fput: turn "list_head delayed_fput_list" into llist_head 2013-07-13 13:29:10 +04:00
filesystems.c fs: Limit sys_mount to only request filesystem modules. 2013-03-03 19:36:31 -08:00
fs-writeback.c mm/writeback: don't check force_wait to handle bdi->work_list 2013-07-09 10:33:22 -07:00
fs_struct.c constify path_get/path_put and fs_struct.c stuff 2013-03-01 23:51:07 -05:00
generic_acl.c
…
inode.c allow the temp files created by open() to be linked to 2013-06-29 12:57:11 +04:00
internal.h constify rw_verify_area() 2013-06-29 12:57:34 +04:00
ioctl.c new helper: file_inode(file) 2013-02-22 23:31:31 -05:00
ioprio.c
…
libfs.c make simple_lookup() usable for filesystems that set ->s_d_op 2013-07-14 17:43:25 +04:00
locks.c locks: move file_lock_list to a set of percpu hlist_heads and convert file_lock_lock to an lglock 2013-07-08 13:36:42 +04:00
mbcache.c
…
mount.h get rid of full-hash scan on detaching vfsmounts 2013-04-09 14:12:52 -04:00
mpage.c
…
namei.c fs: Allow unprivileged linkat(..., AT_EMPTY_PATH) aka flink 2013-08-05 18:24:11 +04:00
namespace.c create_mnt_ns: unidiomatic use of list_add() 2013-05-04 15:18:53 -04:00
no-block.c
…
open.c fs: Fix file mode for O_TMPFILE 2013-08-05 18:24:10 +04:00
pipe.c aio: don't include aio.h in sched.h 2013-05-07 20:16:25 -07:00
pnode.c vfs: Fix invalid ida_remove() call 2013-05-31 15:16:33 -04:00
pnode.h Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2013-05-01 17:51:54 -07:00
posix_acl.c
…
proc_namespace.c
…
read_write.c vfs: export lseek_execute() to modules 2013-07-03 16:23:27 +04:00
readdir.c [readdir] constify ->actor 2013-06-29 12:57:05 +04:00
select.c net: rename include/net/ll_poll.h to include/net/busy_poll.h 2013-07-10 17:08:27 -07:00
seq_file.c seq_file: add seq_list_*_percpu helpers 2013-07-08 13:36:41 +04:00
signalfd.c switch signalfd{,4}() to COMPAT_SYSCALL_DEFINE 2013-03-03 22:58:46 -05:00
splice.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2013-07-03 09:10:19 -07:00
stack.c
…
stat.c switch vfs_getattr() to struct path 2013-02-26 02:46:08 -05:00
statfs.c
…
super.c livelock avoidance in sget() 2013-07-20 04:58:58 +04:00
sync.c teach SYSCALL_DEFINE<n> how to deal with long long/unsigned long long 2013-03-03 22:46:22 -05:00
timerfd.c timerfd: Add alarm timers 2013-05-29 12:57:34 -07:00
utimes.c
…
xattr.c
…
xattr_acl.c
…