mirror_zfs/module/zfs
Richard Yao 6f9548c487 Fix deadlock in zfs_zget()
zfsonlinux/zfs#180 occurred because of a race between inode eviction and
zfs_zget(). zfsonlinux/zfs@36df284 tried to address it by making a call
to the VFS to learn whether an inode is being evicted.  If it was being
evicted the operation was retried after dropping and reacquiring the
relevant resources.  Unfortunately, this introduced another deadlock.

  INFO: task kworker/u24:6:891 blocked for more than 120 seconds.
        Tainted: P           O 3.13.6 #1
  "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
  kworker/u24:6   D ffff88107fcd2e80     0   891      2 0x00000000
  Workqueue: writeback bdi_writeback_workfn (flush-zfs-5)
   ffff8810370ff950 0000000000000002 ffff88103853d940 0000000000012e80
   ffff8810370fffd8 0000000000012e80 ffff88103853d940 ffff880f5c8be098
   ffff88107ffb6950 ffff8810370ff980 ffff88103a9a5b78 0000000000000000
  Call Trace:
   [<ffffffff813dd1d4>] schedule+0x24/0x70
   [<ffffffff8115fc09>] __wait_on_freeing_inode+0x99/0xc0
   [<ffffffff8115fdd8>] find_inode_fast+0x78/0xb0
   [<ffffffff811608c5>] ilookup+0x65/0xd0
   [<ffffffffa035c5ab>] zfs_zget+0xdb/0x260 [zfs]
   [<ffffffffa03589d6>] zfs_get_data+0x46/0x340 [zfs]
   [<ffffffffa035fee1>] zil_add_block+0xa31/0xc00 [zfs]
   [<ffffffffa0360642>] zil_commit+0x12/0x20 [zfs]
   [<ffffffffa036a6e4>] zpl_putpage+0x174/0x840 [zfs]
   [<ffffffff811071ec>] do_writepages+0x1c/0x40
   [<ffffffff8116df2b>] __writeback_single_inode+0x3b/0x2b0
   [<ffffffff8116ecf7>] writeback_sb_inodes+0x247/0x420
   [<ffffffff8116f5f3>] wb_writeback+0xe3/0x320
   [<ffffffff81170b8e>] bdi_writeback_workfn+0xfe/0x490
   [<ffffffff8106072c>] process_one_work+0x16c/0x490
   [<ffffffff810613f3>] worker_thread+0x113/0x390
   [<ffffffff81066edf>] kthread+0xdf/0x100

This patch implements the original fix in a slightly different manner in
order to avoid both deadlocks.  Instead of relying on a call to ilookup()
which can block in __wait_on_freeing_inode() the return value from igrab()
is used.  This gives us the information that ilookup() provided without
the risk of a deadlock.

Alternately, this race could be closed by registering an sops->drop_inode()
callback.  The callback would need to detect the active SA hold thereby
informing the VFS that this inode should not be evicted.

Signed-off-by: Richard Yao <ryao@gentoo.org>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #180
2014-04-04 09:11:54 -07:00
..
arc.c Illumos #4089 NULL pointer dereference in arc_read() 2014-03-24 11:06:57 -07:00
bplist.c Illumos #3464 2013-09-04 16:01:24 -07:00
bpobj.c Illumos #3603, #3604: bobj improvements 2013-10-31 14:57:51 -07:00
bptree.c 26126 panic system rather than corrupting pool if we hit bug 26100 2013-11-05 13:18:26 -08:00
dbuf_stats.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
dbuf.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
ddt_zap.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
ddt.c Add ddt, ddt_entry, and l2arc_hdr caches 2014-01-07 10:33:11 -08:00
dmu_diff.c Illumos #3598 2013-10-31 14:58:04 -07:00
dmu_object.c Illumos #3598 2013-10-31 14:58:04 -07:00
dmu_objset.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
dmu_send.c Add zfs_send_corrupt_data module option 2013-12-18 16:46:35 -08:00
dmu_traverse.c Illumos 4504 traverse_visitbp: visit group before user 2014-01-29 15:50:49 -08:00
dmu_tx.c Improve reporting of tx assignment wait times 2014-03-04 12:22:24 -08:00
dmu_zfetch.c Use enum type(zfetch_dirn_t) instead 2014-01-23 12:56:33 -08:00
dmu.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
dnode_sync.c Illumos #3742 2013-11-04 10:55:25 -08:00
dnode.c Illumos #4045 write throttle & i/o scheduler performance work 2013-12-06 09:32:43 -08:00
dsl_dataset.c Illumos #4574 get_clones_stat does not call zap_count in non-debug kernel 2014-03-04 11:50:13 -08:00
dsl_deadlist.c Illumos #3104: eliminate empty bpobjs 2013-01-08 10:35:43 -08:00
dsl_deleg.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
dsl_destroy.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
dsl_dir.c dmu_tx kstat cleanup 2014-03-04 12:22:24 -08:00
dsl_pool.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
dsl_prop.c Illumos #3742 2013-11-04 10:55:25 -08:00
dsl_scan.c Add erratum for issue #2094 2014-02-21 12:10:40 -08:00
dsl_synctask.c Export symbols dsl_sync_task{_nowait} 2014-03-07 10:01:36 -08:00
dsl_userhold.c Some nvlist allocations in hold processing need to use KM_PUSHPAGE. 2013-12-02 14:02:46 -08:00
fm.c Add zpool_events_seek() functionality 2014-03-31 16:10:57 -07:00
gzip.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
lz4.c Force LZ4_FORCE_SW_BITCOUNT for Sparc 2014-01-09 15:54:03 -08:00
lzjb.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
Makefile.in Add visibility in to cached dbufs 2013-10-25 13:59:40 -07:00
metaslab.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
refcount.c Illumos #3464 2013-09-04 16:01:24 -07:00
rrwlock.c Fix several new KM_SLEEP warnings 2013-09-25 15:44:22 -07:00
sa.c Properly handle updates of variably-sized SA entries. 2013-12-20 13:52:33 -08:00
sha256.c Add linux sha2 support 2010-08-31 13:41:59 -07:00
spa_boot.c Add linux kernel module support 2010-08-31 13:41:58 -07:00
spa_config.c Implement -t option to zpool import for temporary pool names 2014-03-20 12:05:30 -07:00
spa_errlog.c Illumos #3743 2013-11-04 10:55:25 -08:00
spa_history.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
spa_misc.c Use KM_PUSHPAGE in spa_add() for spa_label_features. 2014-03-10 09:09:30 -07:00
spa_stats.c replace nreserved with ndirty in txgs kstat 2014-03-04 12:22:24 -08:00
spa.c Add automatic hot spare functionality 2014-04-02 13:10:08 -07:00
space_map.c Illumos #3464 2013-09-04 16:01:24 -07:00
txg.c replace nreserved with ndirty in txgs kstat 2014-03-04 12:22:24 -08:00
uberblock.c Illumos #3598 2013-10-31 14:58:04 -07:00
unique.c Switch KM_SLEEP to KM_PUSHPAGE 2012-08-27 12:01:37 -07:00
vdev_cache.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
vdev_disk.c Invalidate Linux buffer cache on vdevs upon each flush 2014-03-04 12:22:03 -08:00
vdev_file.c vdev_file_io_start() to use taskq_dispatch(TQ_PUSHPAGE) 2014-01-23 09:58:07 -08:00
vdev_label.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
vdev_mirror.c Illumos #4045 write throttle & i/o scheduler performance work 2013-12-06 09:32:43 -08:00
vdev_missing.c Illumos #3598 2013-10-31 14:58:04 -07:00
vdev_queue.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
vdev_raidz.c Illumos #4045 write throttle & i/o scheduler performance work 2013-12-06 09:32:43 -08:00
vdev_root.c Illumos #3598 2013-10-31 14:58:04 -07:00
vdev.c Illumos #4045 write throttle & i/o scheduler performance work 2013-12-06 09:32:43 -08:00
zap_leaf.c Illumos #3598 2013-10-31 14:58:04 -07:00
zap_micro.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zap.c Illumos #3743 2013-11-04 10:55:25 -08:00
zfeature_common.c Illumos #3035 LZ4 compression support in ZFS and GRUB 2013-01-29 09:28:20 -08:00
zfeature.c Fix zap_lookup() in feature_is_supported(). 2014-03-04 11:44:44 -08:00
zfs_acl.c Allow chown/chgrp when no ACL SAs exist. 2014-01-23 11:07:29 -08:00
zfs_byteswap.c Add linux kernel module support 2010-08-31 13:41:58 -07:00
zfs_ctldir.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zfs_debug.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zfs_dir.c Illumos #4347 ZPL can use dmu_tx_assign(TXG_WAIT) 2013-12-06 09:30:51 -08:00
zfs_fm.c Add automatic hot spare functionality 2014-04-02 13:10:08 -07:00
zfs_fuid.c Illumos #3522 2013-10-30 14:51:27 -07:00
zfs_ioctl.c Clarify zpool_events_next() comment 2014-03-31 16:11:08 -07:00
zfs_log.c Only commit the ZIL once in zpl_writepages() (msync() case). 2013-11-23 15:08:29 -08:00
zfs_onexit.c Illumos #3598 2013-10-31 14:58:04 -07:00
zfs_replay.c Illumos #3598 2013-10-31 14:58:04 -07:00
zfs_rlock.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zfs_sa.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zfs_vfsops.c Propagate errors when registering "relatime" property callback. 2014-02-12 09:38:28 -08:00
zfs_vnops.c Fix zfs_getattr_fast types 2014-01-09 15:50:23 -08:00
zfs_znode.c Fix deadlock in zfs_zget() 2014-04-04 09:11:54 -07:00
zil.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zio_checksum.c Illumos #3598 2013-10-31 14:58:04 -07:00
zio_compress.c Illumos #3598 2013-10-31 14:58:04 -07:00
zio_inject.c Illumos #3598 2013-10-31 14:58:04 -07:00
zio.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zle.c Update core ZFS code from build 121 to build 141. 2010-05-28 13:45:14 -07:00
zpl_ctldir.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zpl_export.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zpl_file.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zpl_inode.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zpl_super.c Prune metadata from ghost lists in arc_adjust_meta 2014-02-21 16:10:49 -08:00
zpl_xattr.c cstyle: Resolve C style issues 2013-12-18 16:46:35 -08:00
zrlock.c Export ZFS symbols needed by Lustre. 2010-09-17 16:24:15 -07:00
zvol.c Use long holds in zvol_set_volsize() 2014-01-14 14:46:12 -08:00