[Top] [All Lists]

Re: [PATCH v2] xfstests: Add test case to test xfs projid32bit functiona

To: Dave Chinner <david@xxxxxxxxxxxxx>
Subject: Re: [PATCH v2] xfstests: Add test case to test xfs projid32bit functionality a bit more extensively.
From: Eric Sandeen <sandeen@xxxxxxxxxxx>
Date: Wed, 05 Sep 2012 19:40:38 -0500
Cc: Boris Ranto <ranto.boris@xxxxxxxxx>, xfs-oss <xfs@xxxxxxxxxxx>
In-reply-to: <20120906003150.GM15292@dastard>
References: <CAFZPdfg+5kRfUuVOtRk1pqKrRai6CZBxisvXnuPUr-_SNAmGTg@xxxxxxxxxxxxxx> <20120906003150.GM15292@dastard>
User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:15.0) Gecko/20120824 Thunderbird/15.0
On 9/5/12 7:31 PM, Dave Chinner wrote:
> On Tue, Aug 28, 2012 at 02:59:42PM +0200, Boris Ranto wrote:
>>  The test covers several areas including enabling projid32bit
>> functionality dynamically by xfs_admin, dumping, restoring, quota
>> reporting and xfs_db projid values reporting.
>>  This test case hits two bugs: one for broken xfsdump/xfsrestore
>> functionality and one for enabling projid32bit functionality with
>> xfs_admin on a LVM device (SCRATCH_DEV must be an LVM device to hit
>> this).
> The LVM problem is incidental - if there's a problem with LVM
> devices then xfs_db should show the same problem, as should all
> other xfs_admin commands that use xfs_db. Hence I don't think that
> there is any point in mentioning it here.

I agree.  And FWIW, though I tried to bisect my way to the root cause,
in the end I realized that upstream it's no longer a problem, so TBH
I stopped looking ...


>> +mkdir -p $dir
>> +touch $dir/{16,32}less
> "less" is a weird filename suffix. What's it mean? Why not just
> something standard like "test"?

16 or less (fewer) bits, 32 or less bits.

>> +# Now, we can examine the file systems with xfs_db
>> +# These two should report the same values
>> +echo "These two values of 16bit project quota ids shall be the same"
> No need to output this into the out file. The comment says it all,
> and we don't need the text in the output file to test for test
> failure.

Although the comments don't end up in the output file; if it fails with an out 
file diff like:

 core.projid_lo = 24853
 core.projid_hi = 32401
 core.projid_lo = 24853
-core.projid_hi = 32401
+core.projid_hi = 0
 core.projid_lo = 24853
 core.projid_hi = 32401

it's sometimes tedious to work your way back to where the numbers came from and 
what was being tested at the time; IOW comment-style landmarks in the output 
file like this are sometimes helpful, in my experience.


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