Btrfs: fix warning when replaying log after fsync of a tmpfile
authorFilipe Manana <fdmanana@suse.com>
Mon, 8 Oct 2018 10:12:55 +0000 (11:12 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Tue, 13 Nov 2018 19:08:58 +0000 (11:08 -0800)
commit55f21e169c84461bef80f1c3ab135f8b8b14785e
tree0a70f3ed9330b26d02ed344188325f013e4a047b
parent1d6d4a0339675b0f2edb421191dc6518d5720be2
Btrfs: fix warning when replaying log after fsync of a tmpfile

commit f2d72f42d5fa3bf33761d9e47201745f624fcff5 upstream.

When replaying a log which contains a tmpfile (which necessarily has a
link count of 0) we end up calling inc_nlink(), at
fs/btrfs/tree-log.c:replay_one_buffer(), which produces a warning like
the following:

  [195191.943673] WARNING: CPU: 0 PID: 6924 at fs/inode.c:342 inc_nlink+0x33/0x40
  [195191.943723] CPU: 0 PID: 6924 Comm: mount Not tainted 4.19.0-rc6-btrfs-next-38 #1
  [195191.943724] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.11.2-0-gf9626ccb91-prebuilt.qemu-project.org 04/01/2014
  [195191.943726] RIP: 0010:inc_nlink+0x33/0x40
  [195191.943728] RSP: 0018:ffffb96e425e3870 EFLAGS: 00010246
  [195191.943730] RAX: 0000000000000000 RBX: ffff8c0d1e6af4f0 RCX: 0000000000000006
  [195191.943731] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff8c0d1e6af4f0
  [195191.943731] RBP: 0000000000000097 R08: 0000000000000001 R09: 0000000000000000
  [195191.943732] R10: 0000000000000000 R11: 0000000000000000 R12: ffffb96e425e3a60
  [195191.943733] R13: ffff8c0d10cff0c8 R14: ffff8c0d0d515348 R15: ffff8c0d78a1b3f8
  [195191.943735] FS:  00007f570ee24480(0000) GS:ffff8c0dfb200000(0000) knlGS:0000000000000000
  [195191.943736] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  [195191.943737] CR2: 00005593286277c8 CR3: 00000000bb8f2006 CR4: 00000000003606f0
  [195191.943739] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
  [195191.943740] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
  [195191.943741] Call Trace:
  [195191.943778]  replay_one_buffer+0x797/0x7d0 [btrfs]
  [195191.943802]  walk_up_log_tree+0x1c1/0x250 [btrfs]
  [195191.943809]  ? rcu_read_lock_sched_held+0x3f/0x70
  [195191.943825]  walk_log_tree+0xae/0x1d0 [btrfs]
  [195191.943840]  btrfs_recover_log_trees+0x1d7/0x4d0 [btrfs]
  [195191.943856]  ? replay_dir_deletes+0x280/0x280 [btrfs]
  [195191.943870]  open_ctree+0x1c3b/0x22a0 [btrfs]
  [195191.943887]  btrfs_mount_root+0x6b4/0x800 [btrfs]
  [195191.943894]  ? rcu_read_lock_sched_held+0x3f/0x70
  [195191.943899]  ? pcpu_alloc+0x55b/0x7c0
  [195191.943906]  ? mount_fs+0x3b/0x140
  [195191.943908]  mount_fs+0x3b/0x140
  [195191.943912]  ? __init_waitqueue_head+0x36/0x50
  [195191.943916]  vfs_kern_mount+0x62/0x160
  [195191.943927]  btrfs_mount+0x134/0x890 [btrfs]
  [195191.943936]  ? rcu_read_lock_sched_held+0x3f/0x70
  [195191.943938]  ? pcpu_alloc+0x55b/0x7c0
  [195191.943943]  ? mount_fs+0x3b/0x140
  [195191.943952]  ? btrfs_remount+0x570/0x570 [btrfs]
  [195191.943954]  mount_fs+0x3b/0x140
  [195191.943956]  ? __init_waitqueue_head+0x36/0x50
  [195191.943960]  vfs_kern_mount+0x62/0x160
  [195191.943963]  do_mount+0x1f9/0xd40
  [195191.943967]  ? memdup_user+0x4b/0x70
  [195191.943971]  ksys_mount+0x7e/0xd0
  [195191.943974]  __x64_sys_mount+0x21/0x30
  [195191.943977]  do_syscall_64+0x60/0x1b0
  [195191.943980]  entry_SYSCALL_64_after_hwframe+0x49/0xbe
  [195191.943983] RIP: 0033:0x7f570e4e524a
  [195191.943986] RSP: 002b:00007ffd83589478 EFLAGS: 00000206 ORIG_RAX: 00000000000000a5
  [195191.943989] RAX: ffffffffffffffda RBX: 0000563f335b2060 RCX: 00007f570e4e524a
  [195191.943990] RDX: 0000563f335b2240 RSI: 0000563f335b2280 RDI: 0000563f335b2260
  [195191.943992] RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000020
  [195191.943993] R10: 00000000c0ed0000 R11: 0000000000000206 R12: 0000563f335b2260
  [195191.943994] R13: 0000563f335b2240 R14: 0000000000000000 R15: 00000000ffffffff
  [195191.944002] irq event stamp: 8688
  [195191.944010] hardirqs last  enabled at (8687): [<ffffffff9cb004c3>] console_unlock+0x503/0x640
  [195191.944012] hardirqs last disabled at (8688): [<ffffffff9ca037dd>] trace_hardirqs_off_thunk+0x1a/0x1c
  [195191.944018] softirqs last  enabled at (8638): [<ffffffff9cc0a5d1>] __set_page_dirty_nobuffers+0x101/0x150
  [195191.944020] softirqs last disabled at (8634): [<ffffffff9cc26bbe>] wb_wakeup_delayed+0x2e/0x60
  [195191.944022] ---[ end trace 5d6e873a9a0b811a ]---

This happens because the inode does not have the flag I_LINKABLE set,
which is a runtime only flag, not meant to be persisted, set when the
inode is created through open(2) if the flag O_EXCL is not passed to it.
Except for the warning, there are no other consequences (like corruptions
or metadata inconsistencies).

Since it's pointless to replay a tmpfile as it would be deleted in a
later phase of the log replay procedure (it has a link count of 0), fix
this by not logging tmpfiles and if a tmpfile is found in a log (created
by a kernel without this change), skip the replay of the inode.

A test case for fstests follows soon.

Fixes: 471d557afed1 ("Btrfs: fix loss of prealloc extents past i_size after fsync log replay")
CC: stable@vger.kernel.org # 4.18+
Reported-by: Martin Steigerwald <martin@lichtvoll.de>
Link: https://lore.kernel.org/linux-btrfs/3666619.NTnn27ZJZE@merkaba/
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
fs/btrfs/tree-log.c