xfs
[Top] [All Lists]

Re: [PATCH] Test to ensure that the EOFBLOCK_FL gets set/unset correctly

To: Dave Chinner <david@xxxxxxxxxxxxx>
Subject: Re: [PATCH] Test to ensure that the EOFBLOCK_FL gets set/unset correctly.
From: Eric Sandeen <sandeen@xxxxxxxxxxx>
Date: Fri, 27 Aug 2010 19:23:44 -0500
Cc: Akshay Lal <alal@xxxxxxxxxx>, xfs@xxxxxxxxxxx
In-reply-to: <20100828001703.GK705@dastard>
References: <1282941224-5805-1-git-send-email-alal@xxxxxxxxxx> <20100827233216.GJ705@dastard> <4C785254.2020708@xxxxxxxxxxx> <20100828001703.GK705@dastard>
User-agent: Thunderbird 2.0.0.24 (Macintosh/20100228)
Dave Chinner wrote:
> On Fri, Aug 27, 2010 at 07:03:32PM -0500, Eric Sandeen wrote:
>> Dave Chinner wrote:
>>
>>> I'm not sure this really is a generic test - it's testing an ext4
>>> specific bug. We've got other generic tests that exercise fallocate,
>>> and some filesystems (like XFS) don't have special bits to say there
>>> are extents beyond EOF and checking a filesystem repeated won't
>>> report any problems.  So perhaps if should be '_supported_fs ext4'
>>
>> Oops we're giving conflicting advice :)  I thought a test that
>> exercises blocks-past-eof-filling at various boundaries made
>> sense in general, even if the specific regression test is ext4-specific.
>>
>> Seems like at least ocfs2/btrfs might benefit from the basic exercise,
>> so I was recommending that it be generic.
> 
> Ok, that seems reasonable. If the bug results in filesystem
> corruption, then maybe just relying on the check at the end of the
> test to fail it would be appropriate?

That's fine by me, if e2fsck will squawk, that's enough.

-Eric

> Cheers,
> 
> Dave.

<Prev in Thread] Current Thread [Next in Thread>