xfs
[Top] [All Lists]

Re: [PATCH V2] xfstests generic/306 fixes and updates

To: Dave Chinner <david@xxxxxxxxxxxxx>
Subject: Re: [PATCH V2] xfstests generic/306 fixes and updates
From: Eric Sandeen <sandeen@xxxxxxxxxxx>
Date: Tue, 30 Apr 2013 09:16:59 -0500
Cc: Eric Sandeen <sandeen@xxxxxxxxxx>, xfs-oss <xfs@xxxxxxxxxxx>
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <20130430074052.GJ23072@dastard>
References: <517AD562.3060208@xxxxxxxxxx> <517E9B52.7030008@xxxxxxxxxxx> <20130430074052.GJ23072@dastard>
User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:17.0) Gecko/20130328 Thunderbird/17.0.5
On 4/30/13 2:40 AM, Dave Chinner wrote:
> On Mon, Apr 29, 2013 at 11:09:54AM -0500, Eric Sandeen wrote:
>> older xfs_io refused to write to /dev/null because it's
>> not a file on an xfs filesystem.  So add -F.
> 
> You could just do:
> 
> export XFS_IO_PROG="$XFS_IO_PROG -F"
> 
> and not have to change so much of the test. And with a comment it
> would make it obvious to a future ready why the -F flag is used ;)
> 
> Cheers,
> 
> Dave.
> 

Well, it's not just this test; it's any non-xfs-specific test
which uses xfs_io for anything at all.  We have -F's sprinkled
through many tests for that reason, even though -F is deprecated upstream.

There are 15 non-xfs-specific tests which use XFS_IO_PROG:

[sandeen@sandeen xfstests]$ grep -l XFS_IO_PROG tests/generic/* tests/shared/* 
| wc -l
15

and in most cases -F is explicitly added:

[sandeen@sandeen xfstests]$ grep XFS_IO_PROG tests/generic/* tests/shared/* | 
grep "\-F" | wc -l
42

but in this case it wasn't; ditto for 298.

So I guess it's a question of remembering to fix each individual test (and 
every new
test) or somehow making it automatic.  It'll get missed often, since this really
only comes up when testing a distro with an old xfsprogs.

-Eric

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