Anyone have test cases for SEEK_HOLE/SEEK_DATA?

Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Theodore Ts'o]

Does anyone have test cases for SEEK_HOLE/SEEK_DATA handy?  As near as I
can tell xfstests doesn't have any tests for SEEK_HOLE/SEEK_DATA
functionality, and I was hoping someone had some test programs and test
sets already written.

Thanks!!

                                                        - Ted

Re: Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Jie Liu]

Hi Ted,

On 09/18/12 10:04, Theodore Ts'o wrote:
> Does anyone have test cases for SEEK_HOLE/SEEK_DATA handy?  As near as I
> can tell xfstests doesn't have any tests for SEEK_HOLE/SEEK_DATA
> functionality, and I was hoping someone had some test programs and test
> sets already written.
I have submitted two test cases to xfstests for this tests.

xfstests: introduce 285 for SEEK_DATA/SEEK_HOLE sanity test
xfstests: introduce 286 for SEEK_DATA/SEEK_HOLE copy test


Thanks,
-Jeff

Re: Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Marco Stornelli]

2012/9/18 Theodore Ts'o <tytso@mit.edu>:
>
> Does anyone have test cases for SEEK_HOLE/SEEK_DATA handy?  As near as I
> can tell xfstests doesn't have any tests for SEEK_HOLE/SEEK_DATA
> functionality, and I was hoping someone had some test programs and test
> sets already written.
>
> Thanks!!
>
>                                                         - Ted


Hi,

under the folder "test" here
(http://pramfs.git.sourceforge.net/git/gitweb-index.cgi) you can find
the test lseek.c

Regards,

Marco

Re: Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Dave Chinner]

On Mon, Sep 17, 2012 at 10:04:34PM -0400, Theodore Ts'o wrote:
> 
> Does anyone have test cases for SEEK_HOLE/SEEK_DATA handy?  As near as I
> can tell xfstests doesn't have any tests for SEEK_HOLE/SEEK_DATA
> functionality, and I was hoping someone had some test programs and test
> sets already written.

Perhaps a cc to xfs@oss.sgi.com so the xfstests maintainers see the
question would be a sensible idea?

As it is, there are tests pending check-in that test
seek-data/seek-hole functionality - they were used to verify the
unwritten extent aware XFS implementation that recently went into
the kernel....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Mark Tinguely]

On 09/18/12 01:40, Dave Chinner wrote:
> On Mon, Sep 17, 2012 at 10:04:34PM -0400, Theodore Ts'o wrote:
>>
>> Does anyone have test cases for SEEK_HOLE/SEEK_DATA handy?  As near as I
>> can tell xfstests doesn't have any tests for SEEK_HOLE/SEEK_DATA
>> functionality, and I was hoping someone had some test programs and test
>> sets already written.
>
> Perhaps a cc to xfs@oss.sgi.com so the xfstests maintainers see the
> question would be a sensible idea?
>
> As it is, there are tests pending check-in that test
> seek-data/seek-hole functionality - they were used to verify the
> unwritten extent aware XFS implementation that recently went into
> the kernel....
>
> Cheers,
>
> Dave.

xfstest 285 and 286 (I believe these tests were only in the OSS version 
and not in the kernel.org developers version of xfstests).

There is rough version of another test that I got directly from Jeff 
Lui, the person who wrote the SEEK_HOLE/SEEK_DATA feature in XFS. This 
version tests the advance features of SEEK_HOLE/SEEK_DATA.

Ben asked me about this advanced test when he committed the advance 
features of SEEK_HOLE/SEEK_DATA to the XFS OSS tree. To the best of my 
knowledge, this test has not been submitted as an xfstest so we could 
not add it then.

--Mark Tinguely.

Re: Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Theodore Ts'o]

On Tue, Sep 18, 2012 at 09:43:36AM -0500, Mark Tinguely wrote:
> xfstest 285 and 286 (I believe these tests were only in the OSS
> version and not in the kernel.org developers version of xfstests).

Hi Mark,

Could you clarify what the relationship is between the OSS xfstests
tree and the one on kernel.org?  Is the OSS tree always ahead of the
kernel.org tree (i.e., are the commits in the OSS tree a superset of
the ones in the kernel.org tree)?

I had been under the impression that commits flowed from the
kernel.org tree to the OSS tree, but it sounds like I was misinformed
on that point?

Thanks,

						- Ted

Re: Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Eric Sandeen]

On 9/18/12 1:15 PM, Theodore Ts'o wrote:
> On Tue, Sep 18, 2012 at 09:43:36AM -0500, Mark Tinguely wrote:
>> xfstest 285 and 286 (I believe these tests were only in the OSS
>> version and not in the kernel.org developers version of xfstests).
> 
> Hi Mark,
> 
> Could you clarify what the relationship is between the OSS xfstests
> tree and the one on kernel.org?  Is the OSS tree always ahead of the
> kernel.org tree (i.e., are the commits in the OSS tree a superset of
> the ones in the kernel.org tree)?
> 
> I had been under the impression that commits flowed from the
> kernel.org tree to the OSS tree, but it sounds like I was misinformed
> on that point?

That was more or less correct at one point, but we recently decided to
ditch the kernel.org repos and let sgi manage a single repo on oss.sgi.com
going forward, just to simplify things.

We just reached that agreement and haven't really communicated it yet
I guess.

So going forward, stick with the one on oss.sgi.com.

Thanks,
-Eric

> Thanks,
> 
> 						- Ted
> 
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs
> 

Re: Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Ben Myers]

Hey Ted,

On Tue, Sep 18, 2012 at 02:15:56PM -0400, Theodore Ts'o wrote:
> On Tue, Sep 18, 2012 at 09:43:36AM -0500, Mark Tinguely wrote:
> > xfstest 285 and 286 (I believe these tests were only in the OSS
> > version and not in the kernel.org developers version of xfstests).
> 
> Could you clarify what the relationship is between the OSS xfstests
> tree and the one on kernel.org?  Is the OSS tree always ahead of the
> kernel.org tree (i.e., are the commits in the OSS tree a superset of
> the ones in the kernel.org tree)?

The trees on kernel.org are being retired.  I'm not sure what Christoph is
going to do with the kernel.org repos, but expect new changes to go directly
into the xfstests repo on oss.sgi.com.

> I had been under the impression that commits flowed from the
> kernel.org tree to the OSS tree, but it sounds like I was misinformed
> on that point?

Your impression is just a few days out of date.  See:

http://oss.sgi.com/archives/xfs/2012-09/msg00156.html

Regards,
	Ben

Re: Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Theodore Ts'o]

On Tue, Sep 18, 2012 at 01:28:41PM -0500, Ben Myers wrote:
> Your impression is just a few days out of date.  See:
> 
> http://oss.sgi.com/archives/xfs/2012-09/msg00156.html

On a unrelated note, has there been any thought about potentially
setting up a separate mailing list dedicated for xfstests?  I'm not
subscribed to the xfs mailing list, mainly because I was trying to
avoid adding yet another high volume mailing list to my inbox
(although to be fair, I guess it's only 25 messages per day even on
the busiest days), and so I've missed out on review of Dave's changes
to break out the tests into hierarchies, etc.  I wonder if there are
other xfstests users that aren't yet subscribed to the xfs mailing
list who would if there was a separate mailing list.

Not a big deal; if people really don't want a separate mailing list, I
guess I could subscribe to the xfs list as well...

      	      		       	   - Ted

Re: Anyone have test cases for SEEK_HOLE/SEEK_DATA?

[Ben Myers]

Hey Ted,

On Tue, Sep 18, 2012 at 02:42:01PM -0400, Theodore Ts'o wrote:
> On Tue, Sep 18, 2012 at 01:28:41PM -0500, Ben Myers wrote:
> > Your impression is just a few days out of date.  See:
> > 
> > http://oss.sgi.com/archives/xfs/2012-09/msg00156.html
> 
> On a unrelated note, has there been any thought about potentially
> setting up a separate mailing list dedicated for xfstests?

You might even be the first person to suggest that.  ;)

> I'm not
> subscribed to the xfs mailing list, mainly because I was trying to
> avoid adding yet another high volume mailing list to my inbox
> (although to be fair, I guess it's only 25 messages per day even on
> the busiest days), and so I've missed out on review of Dave's changes
> to break out the tests into hierarchies, etc.  I wonder if there are
> other xfstests users that aren't yet subscribed to the xfs mailing
> list who would if there was a separate mailing list.
> 
> Not a big deal; if people really don't want a separate mailing list, I
> guess I could subscribe to the xfs list as well...

I think it's worth discussing and I'm not resistant to a separate list.  But
like you said, xfs@oss.sgi.com isn't extremely high volume.  Maybe a filter for
xfstests in the subject would be good enough for your purposes?

Regards,
	Ben