Re: [PATCH v8 2/2] xfs: add support for FALLOC_FL_WRITE_ZEROES

[Pankaj Raghav <pankaj.raghav@linux.dev>]

>> diff --git a/fs/xfs/xfs_bmap_util.c b/fs/xfs/xfs_bmap_util.c
>> index 855602cb35e8..e52ad4c25b66 100644
>> --- a/fs/xfs/xfs_bmap_util.c
>> +++ b/fs/xfs/xfs_bmap_util.c
>> @@ -710,6 +710,13 @@ xfs_alloc_file_space(
>>         imapp = &imaps[0];
>>         startoffset_fsb = XFS_B_TO_FSBT(mp, offset);
>>         endoffset_fsb = XFS_B_TO_FSB(mp, offset + count);
>> +
>> +       if (mode == XFS_ALLOC_FILE_SPACE_WRITE_ZEROES &&
>> +           xfs_inode_has_bigrtalloc(ip)) {
>> +               startoffset_fsb = xfs_fileoff_rounddown_rtx(mp, startoffset_fsb);
>> +               endoffset_fsb = xfs_fileoff_roundup_rtx(mp, endoffset_fsb);
>> +       }
>> +
>>         allocatesize_fsb = endoffset_fsb - startoffset_fsb;
>>
>>         /*
>>
>> With that "fixed", I ran into WARN_ON_ONCE(folio_pos > i_size) in iomap_zero_iter,
>> via xfs_falloc_setsize -> xfs_setattr_size -> xfs_zero_range. The
>> rt-aligned alloc now leaves written zeroed blocks past a non-rt-aligned
>> EOF. So WRITE_ZEROES on rtvol breaks "no written blocks past EOF".
> 
> Ah, right, because you can't have a dirty folio totally beyond EOF.
> I think you have to leave the blocks beyond EOF unwritten, then.
> 

Yeah. I was thinking something along those lines. In the xfs_alloc_file_space,
do a written->unwritten conversion at the end for bigrtalloc configurations.

>> I am not sure what would be the most clean way of solving this issue for rtvol.
> 
> Or I guess you could EOPNOTSUPP :P
> 

I can honestly do this for now and add a TODO for bigrtalloc so that we can add
it later. At least we can get the base patches upstream for FALLOC_FL_WRITE_ZEROES.

I will also send the generic test I have added to test these boundary conditions, so that
when we do add bigrtalloc support, there is already a test for this edge case (even though
generic/363 also triggers this edge case almost all the time).

Let me know what you think.

--
Pankaj