Re: [bug report] fstests generic/774 hang

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

On 05/11/2025 10:39, John Garry wrote:
> 
> +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.

This issue goes away when I stop using lfsr, i.e. the test passes.

The problem is that lfsr init in fio does not have enough "blocks", and 
this comes from how the fio bs is same as the increment aw_io_inc, both 
1M in my case. I think that aw_io_inc needs to be much lager than bs.

BTW, I think that the random number gen fio param is only relevant in 
fio write mode. It seems to be even set in 774 for verify read.

Thanks,
John