On 06/29/2011 01:10 PM, Sunil Mushran wrote:
> On 06/29/2011 12:40 AM, Christoph Hellwig wrote:
>> On Wed, Jun 29, 2011 at 04:53:07PM +1000, Dave Chinner wrote:
>>> On Tue, Jun 28, 2011 at 11:33:19AM -0400, Josef Bacik wrote:
>>>> This is a test to make sure seek_data/seek_hole is acting like it
>>>> does on
>>>> Solaris. It will check to see if the fs supports finding a hole or
>>>> not and will
>>>> adjust as necessary.
>>> So I just looked at this with an eye to validating an XFS
>>> implementation, and I came up with this list of stuff that the test
>>> does not cover that I'd need to test in some way:
>>> - files with clean unwritten extents. Are they a hole or
>>> data? What's SEEK_DATA supposed to return on layout like
>>> hole-unwritten-data? i.e. needs to add fallocate to the
>>> - files with dirty unwritten extents (i.e. dirty in memory,
>>> not on disk). They are most definitely data, and most
>>> filesystems will need a separate lookup path to detect
>>> dirty unwritten ranges because the state is kept
>>> separately (page cache vs extent cache). Plenty of scope
>>> for filesystem specific bugs here so needs a roubust test.
>> The discussion leading up to the resurrection of SEEK_HOLE/SEEK_DATA
>> was pretty much about that point. The conclusion based on the Sun
>> documentation and common sense was that SEEK_DATA may only consider
>> unwritten extents as hole if the filesystem has a way to distinguish
>> plain unwritten extents and those that have been dirtied. Else it
>> should be considered data.
>> Testing for making sure dirty preallocated areas aren't wrongly
>> reported sounds relatively easy, the rest falls into implementation
>> details, which imho is fine. Not reporting preallocated extents
>> as holes just is a quality of implementation issue and not a bug.
> I agree. And if I might add my 2 cents that it would be much easier
> if we added another test that created files with all the worrisome boundary
> conditions and used SEEK_DATA/HOLE to copy the files and compared
> using md5sum. This would be far easier than one that expects a certain
> pos for each operation.
That's a great point, I think I will rig something like that up. Thanks,