On Sun, Oct 10, 2021 at 04:31:25PM +0800, Zorro Lang wrote: > On Fri, Oct 01, 2021 at 06:11:49PM +0100, Luis Henriques wrote: > > Hi! > > > > I'm seeing generic/095 failing both with ext4 and xfs (but not on btrfs). > > Here's what I'm getting: > > > > # cat results/generic/095.out.bad > > QA output created by 095 > > fio: io_u error on file /tmp/mnt/scratch/file1: Invalid argument: read offset=51200, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file1: Invalid argument: read offset=133120, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file2: Invalid argument: write offset=92160, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file2: Invalid argument: write offset=158720, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file2: Invalid argument: write offset=513024, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file2: Invalid argument: write offset=33792, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file2: Invalid argument: write offset=449536, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file1: Invalid argument: read offset=45056, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file1: Invalid argument: read offset=30720, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file1: Invalid argument: write offset=222208, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file1: Invalid argument: read offset=342016, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file1: Invalid argument: write offset=177152, buflen=1024 > > fio: pid=4090, got signal=11 > > fio: io_u error on file /tmp/mnt/scratch/file1: Invalid argument: write offset=43008, buflen=1024 > > fio: io_u error on file /tmp/mnt/scratch/file1: Invalid argument: write offset=308224, buflen=1024 > > fio: pid=4086, got signal=11 > > I *guess* it might be caused by a striped storage(fs) of you? I'm not sure I understand what you mean by that. As I mentioned somewhere else in this thread, I'm using zram devices for testing. If I change the test to use file-backed storage device instead, it passes without any issues. Cheers, -- Luís > > Thanks, > Zorro > > > Silence is golden > > > > This was for ext4 but the output for xfs is similar. Since fio is getting > > killed I suspect this may be an issue with this tool. Thus I'm sending > > this to both fio and fstests mailing-lists. > > > > I'm using a 5.15-rc3 kernel and fio-3.28. Is anyone else seeing something > > similar? > > > > Cheers, > > -- > > Luís > > >
