Re: INFO: task hung in ext4_direct_IO

[Tetsuo Handa]

On 2018/05/14 23:36, syzbot wrote:
> Call Trace:
> Âcontext_switch kernel/sched/core.c:2859 [inline]
> Â__schedule+0x801/0x1e30 kernel/sched/core.c:3501
> Âschedule+0xef/0x430 kernel/sched/core.c:3545
> Â__rwsem_down_read_failed_common kernel/locking/rwsem-xadd.c:269 [inline]
> Ârwsem_down_read_failed+0x350/0x5e0 kernel/locking/rwsem-xadd.c:286
> Âcall_rwsem_down_read_failed+0x18/0x30 arch/x86/lib/rwsem.S:94
> Â__down_read arch/x86/include/asm/rwsem.h:83 [inline]
> Âdown_read+0xbd/0x1b0 kernel/locking/rwsem.c:26
> Âinode_lock_shared include/linux/fs.h:723 [inline]
> Âext4_direct_IO_read fs/ext4/inode.c:3828 [inline]
> Âext4_direct_IO+0x653/0x2110 fs/ext4/inode.c:3865
> Âgeneric_file_read_iter+0x510/0x2f00 mm/filemap.c:2341
> Âext4_file_read_iter+0x18b/0x3c0 fs/ext4/file.c:77
> Âcall_read_iter include/linux/fs.h:1778 [inline]
> Âgeneric_file_splice_read+0x552/0x910 fs/splice.c:307
> Âdo_splice_to+0x12e/0x190 fs/splice.c:880
> Âsplice_direct_to_actor+0x268/0x8d0 fs/splice.c:952
> Âdo_splice_direct+0x2cc/0x400 fs/splice.c:1061
> Âdo_sendfile+0x60f/0xe00 fs/read_write.c:1440
> Â__do_sys_sendfile64 fs/read_write.c:1495 [inline]
> Â__se_sys_sendfile64 fs/read_write.c:1487 [inline]
> Â__x64_sys_sendfile64+0x155/0x240 fs/read_write.c:1487
> Âdo_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
> Âentry_SYSCALL_64_after_hwframe+0x49/0xbe

This resembles

Call Trace:
 context_switch kernel/sched/core.c:2848 [inline]
 __schedule+0x801/0x1e30 kernel/sched/core.c:3490
 schedule+0xef/0x430 kernel/sched/core.c:3549
 __rwsem_down_write_failed_common+0x919/0x15d0 kernel/locking/rwsem-xadd.c:566
 rwsem_down_write_failed+0xe/0x10 kernel/locking/rwsem-xadd.c:595
 call_rwsem_down_write_failed+0x17/0x30 arch/x86/lib/rwsem.S:117
 __down_write arch/x86/include/asm/rwsem.h:142 [inline]
 down_write+0xa2/0x120 kernel/locking/rwsem.c:72
 inode_lock include/linux/fs.h:713 [inline]
 ext4_file_write_iter+0x303/0x1420 fs/ext4/file.c:235
 call_write_iter include/linux/fs.h:1784 [inline]
 new_sync_write fs/read_write.c:474 [inline]
 __vfs_write+0x64d/0x960 fs/read_write.c:487
 vfs_write+0x1f8/0x560 fs/read_write.c:549
 ksys_pwrite64+0x174/0x1a0 fs/read_write.c:652
 __do_sys_pwrite64 fs/read_write.c:662 [inline]
 __se_sys_pwrite64 fs/read_write.c:659 [inline]
 __x64_sys_pwrite64+0x97/0xf0 fs/read_write.c:659
 do_syscall_64+0x1b1/0x800 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

in the report below.

  INFO: task hung in ext4_file_write_iter
  https://syzkaller.appspot.com/bug?id=c7370ffdecb4c7c7a24a50e17acaf102850d43ab

But I have no idea why usual file write causes problems.

Since there is LOOP_CHANGE_FD request in the console output, maybe it is better to
try 170785a9cc72e8e1 ("loop: add recursion validation to LOOP_CHANGE_FD") now
and see whether it helps?