Re: INFO: task hung in blk_queue_enter

[Tetsuo Handa]

On 2018/06/07 12:29, Ming Lei wrote:
>> The vfs_fsync() isn't necessary in loop_update_dio() since both
>> generic_file_write_iter() and generic_file_read_iter() can handle
>> buffered io vs dio well.
>>
>> I will send one patch to remove the vfs_sync() later.
> 
> Hi Tetsuo,
> 
> The issue might be fixed by removing this vfs_sync(), but I'd like to
> understand the idea behind since vfs_sync() shouldn't have caused
> any IO to this loop queue.
> 
> I also tried to do the test via the following c syzbot, but can't reproduce
> it yet after running it for several hours.
> 
> https://syzkaller.appspot.com/x/repro.c?id=4727023951937536
> 
> Could you share us how you reproduce it?

I just reported that syzbot started hitting vfs_sync() case.
I haven't succeeded reproducing vfs_sync() case in my environment.

But in general, I modify "any reproducer provided by syzbot that loops
forever" to call exit() if open() etc. failed, for continuing the loop
unlikely helps after hitting "Too many open files" error.

   res = syscall(__NR_memfd_create, 0x200002c0, 0);
   if (res != -1)
     r[1] = res;
+  else
+    exit(1);

Thus, I restart the reproducer in a loop from shell
rather than keep running for hours.

   while :; do echo -n .; ./a.out; done