Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Date:   Fri, 15 Apr 2022 11:10:07 -0400
From:   Eric Whitney <enwlinux@gmail.com>
To:     "Darrick J. Wong" <djwong@kernel.org>
Cc:     Eric Whitney <enwlinux@gmail.com>, fstests@vger.kernel.org,
        linux-ext4@vger.kernel.org
Subject: Re: [PATCH] common/filter: extend _filter_xfs_io to match -nan
Message-ID: <YlmKz6i8YXR0yk9P@debian-BULLSEYE-live-builder-AMD64>
References: <20220414142258.761835-1-enwlinux@gmail.com>
 <20220414152949.GA17014@magnolia>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220414152949.GA17014@magnolia>
Precedence: bulk

* Darrick J. Wong <djwong@kernel.org>:
> On Thu, Apr 14, 2022 at 10:22:58AM -0400, Eric Whitney wrote:
> > When run on ext4 with sufficiently fast x86_64 hardware, generic/130
> > sometimes fails because xfs_io can report rate values as -nan:
> > 0.000000 bytes, 0 ops; 0.0000 sec (-nan bytes/sec and -nan ops/sec)
> > 
> > _filter_xfs_io matches the strings 'inf' or 'nan', but not '-nan'.  In
> > that case it fails to convert the actual output to a normalized form
> > matching generic/130's golden output.  Extend the regular expression
> > used to match xfs_io's output to fix this.
> > 
> > Signed-off-by: Eric Whitney <enwlinux@gmail.com>
> > ---
> >  common/filter | 6 +++---
> >  1 file changed, 3 insertions(+), 3 deletions(-)
> > 
> > diff --git a/common/filter b/common/filter
> > index 5fe86756..5b20e848 100644
> > --- a/common/filter
> > +++ b/common/filter
> > @@ -168,9 +168,9 @@ common_line_filter()
> >  
> >  _filter_xfs_io()
> >  {
> > -    # Apart from standard numeric values, we also filter out 'inf' and 'nan'
> > -    # which can result from division in some cases
> > -    sed -e "s/[0-9/.]* [GMKiBbytes]*, [0-9]* ops\; [0-9/:. sec]* ([infa0-9/.]* [EPGMKiBbytes]*\/sec and [infa0-9/.]* ops\/sec)/XXX Bytes, X ops\; XX:XX:XX.X (XXX YYY\/sec and XXX ops\/sec)/"
> > +    # Apart from standard numeric values, we also filter out 'inf', 'nan', and
> > +    # '-nan' which can result from division in some cases
> > +    sed -e "s/[0-9/.]* [GMKiBbytes]*, [0-9]* ops\; [0-9/:. sec]* ([infa0-9/.-]* [EPGMKiBbytes]*\/sec and [infa0-9/.-]* ops\/sec)/XXX Bytes, X ops\; XX:XX:XX.X (XXX YYY\/sec and XXX ops\/sec)/"
> 
> /me squints at this regular expression and /thinks/ its ok.
> 
> Took me a while to figure out "infa" tho. :P

Hi Darrick:

Yeah, me too.  I initially thought that string would not match 'nan', but
then discovered bracket expressions after having not used sed in a very long
time.  It's a sloppy match - it'll recognize fan0 as well as 'inf' or 'nan',
etc. - and I guess the idea is that's good enough for filtering expected output
from xfs_io while being concise.  The sed documentation suggests that a '-' on
the beginning or end of the bracket expression will be treated literally, and
not as a metacharacter (part of a range), so we should be good there.

> 
> Reviewed-by: Darrick J. Wong <djwong@kernel.org>

Thanks very much for your review!

Eric

> 
> --D
> 
> >  }
> >  
> >  # Also filter out the offset part of xfs_io output
> > -- 
> > 2.30.2
> >