[Top] [All Lists]

Re: [PATCH] xfstests 235: do smaller test IO

To: Dave Chinner <david@xxxxxxxxxxxxx>
Subject: Re: [PATCH] xfstests 235: do smaller test IO
From: Eric Sandeen <sandeen@xxxxxxxxxx>
Date: Mon, 26 Mar 2012 17:21:12 -0500
Cc: xfs-oss <xfs@xxxxxxxxxxx>, Eryu Guan <eguan@xxxxxxxxxx>
In-reply-to: <20120326220926.GO5091@dastard>
References: <4F70C3D9.80708@xxxxxxxxxx> <20120326220926.GO5091@dastard>
User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:11.0) Gecko/20120313 Thunderbird/11.0
On 3/26/12 5:09 PM, Dave Chinner wrote:
> On Mon, Mar 26, 2012 at 02:30:33PM -0500, Eric Sandeen wrote:
>> Test 235 fails on ext2/ext3 with 1024 fs block size because a
>> 16k write uses an extra metadata block.  If we do a smaller write
>> this won't happen.
> .....
>> -$XFS_IO_PROG -F -c 'pwrite 0 16k' -c 'fsync' \
>> +$XFS_IO_PROG -F -c 'pwrite 0 8k' -c 'fsync' \
> So now it fails only on 512 byte block size filesystems?

ext[23] don't do 512 byte block filesystems, so no.  ;)

> Perhaps rather than an exact match, a "within_tolerance" match could
> be done on the relevant fields?
> i.e something like:
> blksused=`{requota cmd} | awk '/fsgqa/ { print $3 }'`
> _within_tolerance "blocks used" $blksused 16 1
> If you wanted to get really fancy, the tolerance could be set
> depending on filesystem type and block size, but I don't think that
> is really necessary....

But I don't want to get really fancy, I spend enough time in xfstests
already.  ;)

TBH even the _within_tolerance will take more munging around to get right,
it doesn't seem worth it, but if it's required for a review, ok...


> Cheers,
> Dave.

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