Re: [bug report] fstests generic/774 hang

[John Garry <john.g.garry@oracle.com>]

On 05/11/2025 08:52, John Garry wrote:
>> I don't think the disk supports atomic writes. It is just a regular 
>> TCMU device,
>> and its atomic write related sysfs attributes have value 0:
>>
>>    $ grep -rne . /sys/block/sdh/queue/ | grep atomic
>>    /sys/block/sdh/queue/atomic_write_unit_max_bytes:1:0
>>    /sys/block/sdh/queue/atomic_write_boundary_bytes:1:0
>>    /sys/block/sdh/queue/atomic_write_max_bytes:1:0
>>    /sys/block/sdh/queue/atomic_write_unit_min_bytes:1:0
>>
>> FYI, I attach the all sysfs queue attribute values of the device [2].
> 
> Yes, this would only be using software-based atomic writes.
> 
> Shinichiro, do the other atomic writes tests run ok, like 775, 767? You 
> can check group "atomicwrites" to know which tests they are.
> 
> 774 is the fio test.
> 
> Some things to try:
> - use a physical disk for the TEST_DEV
> - Don't set LOAD_FACTOR (if you were setting it). If not, bodge 774 to 
> reduce $threads to a low value, say, 2
> - trying turning on XFS_DEBUG config
> 
> BTW, Darrick has posted some xfs atomics fixes @ https://urldefense.com/ 
> v3/__https://lore.kernel.org/linux- 
> xfs/20251105001200.GV196370@frogsfrogsfrogs/T/*t__;Iw!!ACWV5N9M2RV99hQ! 
> IuEPY6yJ1ZEQu7dpfjUplkPJucOHMQ9cpPvIC4fiJhTi_X_7ImN0t6wGqxg9_GM6gWe4B1OBiBjEI8Gz_At0595tIQ$ . I doubt that they will help this, but worth trying.
> 
> I will try to recreate.

I tested this and the filesize which we try to write is huge, like 3.3G 
in my case. That seems excessive.

The calc comes from the following in 774:

filesize=$((aw_bsize * threads * 100))

aw_bsize for  me is 1M, and threads is 32

aw_bsize is large as XFS supports software-based atomics, which is 
generally going to be huge compared to anything which HW can support.

When I tried to run this test, it was not completing in a sane amount of 
time - it was taking many minutes before I gave up.

@shinichiro, please try this:

--- a/tests/generic/774
+++ b/tests/generic/774
@@ -29,7 +29,7 @@ aw_bsize=$(_max "$awu_min_write" "$((awu_max_write/4))")
  fsbsize=$(_get_block_size $SCRATCH_MNT)

  threads=$(_min "$(($(nproc) * 2 * LOAD_FACTOR))" "100")
-filesize=$((aw_bsize * threads * 100))
+filesize=$((aw_bsize * threads))
  depth=$threads
  aw_io_size=$((filesize / threads))
  aw_io_inc=$aw_io_size
ubuntu@jgarry-instance-20240626-1657-xfs-ubuntu:~/xfstests-dev$


Note, I ran with this change and the test now completes, but I get this:

+fio: failed initializing LFSR
     +fio: failed initializing LFSR
     +fio: failed initializing LFSR
     +fio: failed initializing LFSR
     +verify: bad magic header 0, wanted acca at file 
/home/ubuntu/mnt/scratch/test-file offset 0, length 1048576 (requested 
block: offset=0, length=1048576)
     +verify: bad magic header e3d6, wanted acca at file 
/home/ubuntu/mnt/scratch/test-file offset 8388608, length 1048576 
(requested block: offset=8388608, length=1048576)

I need to check that fio complaint.

Thanks,
John