xfstests generic/285 fails reliably on Trond's linux-next branch

xfstests generic/285 fails reliably on Trond's linux-next branch

[Christoph Hellwig]

Now that 207 is fixed the next test starts failing..  285 is
the SEEK_HOLE/DATA sanity check, which might help with figuring out
which of the recent changes introduced it.  

Re: xfstests generic/285 fails reliably on Trond's linux-next branch

[Christoph Hellwig]

[-- Attachment #1: Type: text/plain, Size: 409 bytes --]

On Thu, Jul 21, 2016 at 01:18:02AM -0700, Christoph Hellwig wrote:
> Now that 207 is fixed the next test starts failing..  285 is
> the SEEK_HOLE/DATA sanity check, which might help with figuring out
> which of the recent changes introduced it.

A week later this still fails with the linux-next branch. Plain NFSv4.2
against a Linux/XFS server.  The 285.full file documenting the failures
is attached below:

[-- Attachment #2: 285.full --]
[-- Type: text/plain, Size: 6236 bytes --]

File system magic#: 0x6969
Allocation size: 524288

01. Test empty file                                   
01.01 SEEK_DATA expected -1 with errno -6, got -6.                succ
01.02 SEEK_HOLE expected -1 with errno -6, got -6.                succ
01.03 SEEK_HOLE expected -1 with errno -6, got -6.                succ

02. Test a tiny full file                             
02.01 SEEK_HOLE expected 8 or 8, got 8.                           succ
02.02 SEEK_DATA expected 0 or 0, got 0.                           succ
02.03 SEEK_DATA expected 1 or 1, got 1.                           succ
02.04 SEEK_HOLE expected 8 or 8, got 8.                           succ
02.05 SEEK_DATA expected 7 or 7, got 7.                           succ
02.06 SEEK_HOLE expected -1 with errno -6, got -6.                succ
02.07 SEEK_DATA expected -1 with errno -6, got -6.                succ
02.08 SEEK_HOLE expected -1 with errno -6, got -6.                succ
02.09 SEEK_DATA expected -1 with errno -6, got -6.                succ

03. Test a larger full file                           
03.01 SEEK_HOLE expected 1048676 or 1048676, got 1048676.         succ
03.02 SEEK_HOLE expected 1048676 or 1048676, got 1048676.         succ
03.03 SEEK_DATA expected 0 or 0, got 0.                           succ
03.04 SEEK_DATA expected 1 or 1, got 1.                           succ
03.05 SEEK_HOLE expected 1048676 or 1048676, got 1048676.         succ
03.06 SEEK_DATA expected 1048675 or 1048675, got 1048675.         succ
03.07 SEEK_HOLE expected -1 with errno -6, got -6.                succ
03.08 SEEK_DATA expected -1 with errno -6, got -6.                succ
03.09 SEEK_HOLE expected -1 with errno -6, got -6.                succ
03.10 SEEK_DATA expected -1 with errno -6, got -6.                succ

04. Test file hole at beg, data at end                
04.01 SEEK_HOLE expected 0 or 1048584, got 0.                     succ
04.02 SEEK_HOLE expected 1 or 1048584, got 1.                     succ
04.03 SEEK_DATA expected 1048576 or 1048576, got 1048576.         succ
04.04 SEEK_DATA expected 1048576 or 1048576, got 1048576.         succ
04.05 SEEK_HOLE expected 1048575 or 1048584, got 1048575.         succ
04.06 SEEK_DATA expected 1048576 or 1048576, got 1048576.         succ
04.07 SEEK_HOLE expected 1048584 or 1048584, got 1048584.         succ
04.08 SEEK_DATA expected 1048576 or 1048576, got 1048576.         succ
04.09 SEEK_HOLE expected 1048584 or 1048584, got 1048584.         succ
04.10 SEEK_DATA expected 1048577 or 1048577, got 1048577.         succ
04.11 SEEK_HOLE expected 1048584 or 1048584, got 1048584.         succ
04.12 SEEK_DATA expected 1048583 or 1048583, got 1048583.         succ
04.13 SEEK_HOLE expected -1 with errno -6, got -6.                succ
04.14 SEEK_DATA expected -1 with errno -6, got -6.                succ
04.15 SEEK_HOLE expected -1 with errno -6, got -6.                succ
04.16 SEEK_DATA expected -1 with errno -6, got -6.                succ

05. Test file data at beg, hole at end                
05.01 SEEK_HOLE expected 524288 or 2097152, got 524288.           succ
05.02 SEEK_HOLE expected 524288 or 2097152, got 524288.           succ
05.03 SEEK_DATA expected 0 or 0, got 0.                           succ
05.04 SEEK_DATA expected 1 or 1, got 1.                           succ
05.05 SEEK_HOLE expected 524288 or 2097152, got 524288.           succ
05.06 SEEK_DATA expected 524287 or 524287, got 524287.            succ
05.07 SEEK_HOLE expected 524288 or 2097152, got 524288.           succ
05.08 SEEK_DATA expected -1 with errno -6, got -6.                succ
05.09 SEEK_HOLE expected 524289 or 2097152, got 524289.           succ
05.10 SEEK_DATA expected -1 with errno -6, got -6.                succ
05.11 SEEK_HOLE expected 2097151 or 2097152, got 2097151.         succ
05.12 SEEK_DATA expected -1 with errno -6, got -6.                succ
05.13 SEEK_HOLE expected -1 with errno -6, got -6.                succ
05.14 SEEK_DATA expected -1 with errno -6, got -6.                succ
05.15 SEEK_HOLE expected -1 with errno -6, got -6.                succ
05.16 SEEK_DATA expected -1 with errno -6, got -6.                succ

06. Test file hole data hole data                     
06.01 SEEK_HOLE expected 0 or 2097152, got 0.                     succ
06.02 SEEK_HOLE expected 1 or 2097152, got 1.                     succ
06.03 SEEK_DATA expected 524288 or 524288, got 524288.            succ
06.04 SEEK_DATA expected 524288 or 524288, got 524288.            succ
06.05 SEEK_HOLE expected 524287 or 2097152, got 524287.           succ
06.06 SEEK_DATA expected 524288 or 524288, got 524288.            succ
06.07 SEEK_HOLE expected 1048576 or 2097152, got 1507328.         FAIL
06.08 SEEK_DATA expected 524288 or 524288, got 524288.            succ
06.09 SEEK_HOLE expected 1048576 or 2097152, got 1507328.         FAIL
06.10 SEEK_DATA expected 524289 or 524289, got 524289.            succ
06.11 SEEK_HOLE expected 1048576 or 2097152, got 1507328.         FAIL
06.12 SEEK_DATA expected 1048575 or 1048575, got 1048575.         succ
06.13 SEEK_HOLE expected 1048576 or 2097152, got 1507328.         FAIL
06.14 SEEK_DATA expected 1572864 or 1572864, got 1048576.         FAIL
06.15 SEEK_HOLE expected 1048577 or 2097152, got 1507328.         FAIL
06.16 SEEK_DATA expected 1572864 or 1572864, got 1048577.         FAIL
06.17 SEEK_HOLE expected 1572863 or 2097152, got 1572863.         succ
06.18 SEEK_DATA expected 1572864 or 1572864, got 1572864.         succ
06.19 SEEK_HOLE expected 2097152 or 2097152, got 2097152.         succ
06.20 SEEK_DATA expected 1572864 or 1572864, got 1572864.         succ
06.21 SEEK_HOLE expected 2097152 or 2097152, got 2097152.         succ
06.22 SEEK_DATA expected 1572865 or 1572865, got 1572865.         succ
06.23 SEEK_HOLE expected 2097152 or 2097152, got 2097152.         succ
06.24 SEEK_DATA expected 2097151 or 2097151, got 2097151.         succ
06.25 SEEK_HOLE expected -1 with errno -6, got -6.                succ
06.26 SEEK_DATA expected -1 with errno -6, got -6.                succ
06.27 SEEK_HOLE expected -1 with errno -6, got -6.                succ
06.28 SEEK_DATA expected -1 with errno -6, got -6.                succ

seek sanity check failed!

Re: xfstests generic/285 fails reliably on Trond's linux-next branch

[Anna Schumaker]

Hi Christoph,

On 07/28/2016 10:46 AM, Christoph Hellwig wrote:
> On Thu, Jul 21, 2016 at 01:18:02AM -0700, Christoph Hellwig wrote:
>> Now that 207 is fixed the next test starts failing..  285 is
>> the SEEK_HOLE/DATA sanity check, which might help with figuring out
>> which of the recent changes introduced it.
> 
> A week later this still fails with the linux-next branch. Plain NFSv4.2
> against a Linux/XFS server.  The 285.full file documenting the failures
> is attached below:
> 

What mount options are you using?  The test passes for me on xfs, btrfs, and ext4.

Thanks,
Anna

Re: xfstests generic/285 fails reliably on Trond's linux-next branch

[Christoph Hellwig]

On Thu, Jul 28, 2016 at 03:36:04PM -0400, Anna Schumaker wrote:
> What mount options are you using?  The test passes for me on xfs, btrfs, and ext4.

No mount options at all - just forcing 4.2 with a small patch to
super.c.

I've tried to bisect the issue, but I now see it with Linux 4.7-rc2 as
well.  I did upgrade xfstests in the meantime, so I'll see if that
caused the issue, or if it's something that's new in Linux 4.7-rc2 and
Tronds earlier tree was using an older baseline.