xfs
[Top] [All Lists]

Re: [PATCH] xfstests: xfs directory unbalance assert test

To: Mark Tinguely <tinguely@xxxxxxx>
Subject: Re: [PATCH] xfstests: xfs directory unbalance assert test
From: Eric Sandeen <sandeen@xxxxxxxxxxx>
Date: Tue, 17 Sep 2013 11:06:50 -0500
Cc: xfs@xxxxxxxxxxx
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <52387C1B.6090203@xxxxxxx>
References: <20130917145946.124195107@xxxxxxx> <20130917145959.333796933@xxxxxxx> <52387831.5010205@xxxxxxxxxxx> <52387C1B.6090203@xxxxxxx>
User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:17.0) Gecko/20130801 Thunderbird/17.0.8
On 9/17/13 10:58 AM, Mark Tinguely wrote:
> On 09/17/13 10:41, Eric Sandeen wrote:
>> On 9/17/13 9:59 AM, Mark Tinguely wrote:
>>> This tests triggers an assert in the XFS directory unbalance code.
>>> This test originally written by Brian Foster and suggestions
>>> from Micheal Semon.
>>
>> cool, thanks.  Comments below.
>>
>>> Signed-off-by: Mark Tinguely<tinguely@xxxxxxx>
>>> ---
>>>   tests/generic/319     |   62 
>>> ++++++++++++++++++++++++++++++++++++++++++++++++++
>>>   tests/generic/319.out |    2 +
>>>   tests/generic/group   |    1
>>>   3 files changed, 65 insertions(+)
>>>
>>> Index: b/tests/generic/319
>>> ===================================================================
>>> --- /dev/null
>>> +++ b/tests/generic/319
>>> @@ -0,0 +1,62 @@
>>> +#! /bin/bash
>>> +# FS QA Test No. 319
>>> +#
>>> +# Test directory code correctly handles fsstress filling the filesystem
>>> +#
>>> +#-----------------------------------------------------------------------
>>> +# Copyright (c) 2013 SGI.  All Rights Reserved.
>>> +#
>>> +# This program is free software; you can redistribute it and/or
>>> +# modify it under the terms of the GNU General Public License as
>>> +# published by the Free Software Foundation.
>>> +#
>>> +# This program is distributed in the hope that it would be useful,
>>> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
>>> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>>> +# GNU General Public License for more details.
>>> +#
>>> +# You should have received a copy of the GNU General Public License
>>> +# along with this program; if not, write the Free Software Foundation,
>>> +# Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
>>> +#-----------------------------------------------------------------------
>>> +#
>>> +
>>> +seq=`basename $0`
>>> +seqres=$RESULT_DIR/$seq
>>> +echo "QA output created by $seq"
>>> +
>>> +here=`pwd`
>>> +tmp=/tmp/$$
>>> +status=1    # failure is the default!
>>> +trap "_cleanup; exit \$status" 0 1 2 3 15
>>> +
>>> +_cleanup()
>>> +{
>>> +    cd /
>>> +}
>>
>> That seems pointless; usually it's done w/ rm -f $tmp.*
>> right after, but we have no tmpfile, so...
> 
> Yeah, no cleanup is needed.
> 
>>> +
>>> +# get standard environment, filters and checks
>>> +. ./common/rc
>>> +. ./common/filter
>>> +_require_scratch
>>> +
>>> +# real QA test starts here
>>> +
>>> +_supported_fs generic
>>> +_supported_os IRIX Linux
>>> +
>>> +_scratch_unmount>  /dev/null 2>&1
>>
>> Aside:
>>
>> I see this done both ways - is it required to unmount scratch at the 
>> beginning
>> of a test?  I don't think so (I know it's done in many tests, though, but
>> again, C&P&  cargo cult?  Or not?  I'm not sure :( )
>>
>> I guess it doesn't hurt, but at some point I'd like to get it straight
>> about who's required to umount scratch, and when (if at all).
> 
> Have to unmount for the mkfs, as noted by Eryu, it is already done. I
> would rather manually unmount it than be surprised when someone
> changes the common files.

If that happens, tons of tests will break.  I'd really just remove it
for clarity, but *shrug*

>>
>>> +_scratch_mkfs_sized 11g>>  $seqres.full 2>&1
>>
>> _scratch_mkfs_sized doesn't take units like this ('g'), so the above fails to
>> actually make an 11g fs:
>>
>> # Create fs of certain size on scratch device
>> # _scratch_mkfs_sized<size in bytes>  [optional blocksize]
>> _scratch_mkfs_sized()
>>
>> so we get this in 319.full:
>>
>> expr: non-numeric argument
>> ./common/rc: line 576: [: 11g: integer expression expected
>>
>> but then it seems like mkfs carries on anyway w/ defaults.  :(
>>
>> Apparently the mkfs 11g part isn't actually critical? ;)
> 
> it works on xfs because mkfs.xfs size can take those values, but yes it 
> breaks on other filesystem. my bad.

One other nitpick in this area, please remove $seqres.full before you start so 
it doesn't grow each time the test is run.

(hm maybe we should add that to ./check or something; so many tests miss this)



But - no, it doesn't work for xfs either, at least not in all cases, because
it doesn't do the device size check.  Here's xfs output on a < 11G device:

expr: non-numeric argument
./common/rc: line 582: [: 11g: integer expression expected
** mkfs failed with extra mkfs options added to "-bsize=4096" by test 319 **
** attempting to mkfs using only test 319 options: -d size=11g -b size=4096 **
size 11g specified for data subvolume is too large, maximum is 1048241 blocks
Usage: mkfs.xfs
/* blocksize */         [-b log=n|size=num]
...

and w/o error checking (2>&1 and no || _fail) the test just carries on w/o a
fresh mkfs, on whatever size it happens to be.

-Eric

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