Comment # 2 on bug 1047152 from Aleksa Sarai

I also forgot to mention that I've tested this both on our kernels as well as
the latest upstream kernel (4.12). I've also built the upstream kernel with
lockdep and all of the BTRFS knobs enabled:

CONFIG_BTRFS_FS=m
CONFIG_BTRFS_FS_POSIX_ACL=y
CONFIG_BTRFS_FS_CHECK_INTEGRITY=y
CONFIG_BTRFS_FS_RUN_SANITY_TESTS=y
CONFIG_BTRFS_DEBUG=y
CONFIG_BTRFS_ASSERT=y

And there's no warnings or errors in dmesg. I can also attach the output of
"btrfs inspect-internal dump-tree" if you like.

> causing wait_for_completion(&fs_info->qgroup_rescan_completion) to fail

I meant causing it to wait indefinitely in an uninterruptible sleep. I am a bit
 confused why this eventually causes every other btrfs operation to block as
well, but that's what I see.