btrfs: unlock newly allocated extent buffer after error
authorQu Wenruo <wqu@suse.com>
Tue, 14 Sep 2021 06:57:59 +0000 (14:57 +0800)
committerDavid Sterba <dsterba@suse.com>
Thu, 7 Oct 2021 20:04:20 +0000 (22:04 +0200)
[BUG]
There is a bug report that injected ENOMEM error could leave a tree
block locked while we return to user-space:

  BTRFS info (device loop0): enabling ssd optimizations
  FAULT_INJECTION: forcing a failure.
  name failslab, interval 1, probability 0, space 0, times 0
  CPU: 0 PID: 7579 Comm: syz-executor Not tainted 5.15.0-rc1 #16
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
  rel-1.12.0-59-gc9ba5276e321-prebuilt.qemu.org 04/01/2014
  Call Trace:
   __dump_stack lib/dump_stack.c:88 [inline]
   dump_stack_lvl+0x8d/0xcf lib/dump_stack.c:106
   fail_dump lib/fault-inject.c:52 [inline]
   should_fail+0x13c/0x160 lib/fault-inject.c:146
   should_failslab+0x5/0x10 mm/slab_common.c:1328
   slab_pre_alloc_hook.constprop.99+0x4e/0xc0 mm/slab.h:494
   slab_alloc_node mm/slub.c:3120 [inline]
   slab_alloc mm/slub.c:3214 [inline]
   kmem_cache_alloc+0x44/0x280 mm/slub.c:3219
   btrfs_alloc_delayed_extent_op fs/btrfs/delayed-ref.h:299 [inline]
   btrfs_alloc_tree_block+0x38c/0x670 fs/btrfs/extent-tree.c:4833
   __btrfs_cow_block+0x16f/0x7d0 fs/btrfs/ctree.c:415
   btrfs_cow_block+0x12a/0x300 fs/btrfs/ctree.c:570
   btrfs_search_slot+0x6b0/0xee0 fs/btrfs/ctree.c:1768
   btrfs_insert_empty_items+0x80/0xf0 fs/btrfs/ctree.c:3905
   btrfs_new_inode+0x311/0xa60 fs/btrfs/inode.c:6530
   btrfs_create+0x12b/0x270 fs/btrfs/inode.c:6783
   lookup_open+0x660/0x780 fs/namei.c:3282
   open_last_lookups fs/namei.c:3352 [inline]
   path_openat+0x465/0xe20 fs/namei.c:3557
   do_filp_open+0xe3/0x170 fs/namei.c:3588
   do_sys_openat2+0x357/0x4a0 fs/open.c:1200
   do_sys_open+0x87/0xd0 fs/open.c:1216
   do_syscall_x64 arch/x86/entry/common.c:50 [inline]
   do_syscall_64+0x34/0xb0 arch/x86/entry/common.c:80
   entry_SYSCALL_64_after_hwframe+0x44/0xae
  RIP: 0033:0x46ae99
  Code: f7 d8 64 89 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 48 89 f8 48
  89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d
  01 f0 ff ff 73 01 c3 48 c7 c1 bc ff ff ff f7 d8 64 89 01 48
  RSP: 002b:00007f46711b9c48 EFLAGS: 00000246 ORIG_RAX: 0000000000000055
  RAX: ffffffffffffffda RBX: 000000000078c0a0 RCX: 000000000046ae99
  RDX: 0000000000000000 RSI: 00000000000000a1 RDI: 0000000020005800
  RBP: 00007f46711b9c80 R08: 0000000000000000 R09: 0000000000000000
  R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000017
  R13: 0000000000000000 R14: 000000000078c0a0 R15: 00007ffc129da6e0

  ================================================
  WARNING: lock held when returning to user space!
  5.15.0-rc1 #16 Not tainted
  ------------------------------------------------
  syz-executor/7579 is leaving the kernel with locks still held!
  1 lock held by syz-executor/7579:
   #0: ffff888104b73da8 (btrfs-tree-01/1){+.+.}-{3:3}, at:
  __btrfs_tree_lock+0x2e/0x1a0 fs/btrfs/locking.c:112

[CAUSE]
In btrfs_alloc_tree_block(), after btrfs_init_new_buffer(), the new
extent buffer @buf is locked, but if later operations like adding
delayed tree ref fail, we just free @buf without unlocking it,
resulting above warning.

[FIX]
Unlock @buf in out_free_buf: label.

Reported-by: Hao Sun <sunhao.th@gmail.com>
Link: https://lore.kernel.org/linux-btrfs/CACkBjsZ9O6Zr0KK1yGn=1rQi6Crh1yeCRdTSBxx9R99L4xdn-Q@mail.gmail.com/
CC: stable@vger.kernel.org # 5.4+
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
fs/btrfs/extent-tree.c

index fc3da7585fb786be71f0308541a4d21b4b96305f..0ab456cb4bf801fc9f6f89623100ea182964b8f4 100644 (file)
@@ -4859,6 +4859,7 @@ struct extent_buffer *btrfs_alloc_tree_block(struct btrfs_trans_handle *trans,
 out_free_delayed:
        btrfs_free_delayed_extent_op(extent_op);
 out_free_buf:
+       btrfs_tree_unlock(buf);
        free_extent_buffer(buf);
 out_free_reserved:
        btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 0);