linux-ext4 - Re: [PATCHv2 7/9] generic/620: Use _mkfs_dev

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Sun, 8 Aug 2021 21:32:48 +0800
From:   Eryu Guan <guan@...u.me>
To:     Ritesh Harjani <riteshh@...ux.ibm.com>
Cc:     fstests@...r.kernel.org, linux-ext4@...r.kernel.org,
        Theodore Ts'o <tytso@....edu>,
        "Darrick J . Wong" <djwong@...nel.org>
Subject: Re: [PATCHv2 7/9] generic/620: Use _mkfs_dev_blocksized to use 4k bs

On Tue, Aug 03, 2021 at 10:36:22AM +0530, Ritesh Harjani wrote:
> On 21/08/02 12:03AM, Eryu Guan wrote:
> > On Wed, Jul 21, 2021 at 10:58:00AM +0530, Ritesh Harjani wrote:
> > > ext4 with 64k blocksize (passed by user config) fails with below error for
> > > this given test which requires dmhugedisk. Since this test anyways only
> > > requires 4k bs, so use _mkfs_dev_blocksized() to fix this.

I don't see how this test always requires 4k blocksize, 1k blocksized
xfs also passes the test.

> > >
> > > <error log with 64k bs>
> > > mkfs.ext4: Input/output error while writing out and closing file system

Is this a bug in mkfs.ext4 or expected error (unsupported config)? If
it's an expected error, it'd be better to explain it in commit log as
well.

> > >
> > > Signed-off-by: Ritesh Harjani <riteshh@...ux.ibm.com>
> > > ---
> > >  tests/generic/620 | 4 +++-
> > >  1 file changed, 3 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/tests/generic/620 b/tests/generic/620
> > > index b052376f..444e682d 100755
> > > --- a/tests/generic/620
> > > +++ b/tests/generic/620
> > > @@ -42,7 +42,9 @@ sectors=$((2*1024*1024*1024*17))
> > >  chunk_size=128
> > >
> > >  _dmhugedisk_init $sectors $chunk_size
> > > -_mkfs_dev $DMHUGEDISK_DEV
> > > +
> > > +# Use 4k blocksize.
> > > +_mkfs_dev_blocksized 4096 $DMHUGEDISK_DEV
> >
> > We run the test by forcing 4k blocksize, which could be tested in 4k
> > blocksize setup. Maybe it's another case that should _notrun in 64k
> > blocksize setup.
> 
> So for testing that, first I should mkfs and mount a scratch device with the
> passed mount/mkfs options and then see if the blocksize passed is 64K, if yes
> I should _notrun this case.
> 
> Isn't the current approach of (_mkfs_dev_blocksized 4096) is better then above
> approach?

If the test always requires 4k blocksize, forcing creating a 4k
blocksize filesystem doesn't increase any test coverage, I don't see any
point introducing a new _mkfs_dev_blocksized helper just to do so.

And even if we decide to force 4k blocksize config, I think it'd be
better to update _scratch_mkfs_blocksized() to take device as argument,
like what _check_scratch_fs() does, so we don't duplicate all the code
to create fs with specified blocksize.

Thanks,
Eryu

> 
> -ritesh
> 
> > Thanks,
> > Eryu
> >
> > >  _mount $DMHUGEDISK_DEV $SCRATCH_MNT || _fail "mount failed for $DMHUGEDISK_DEV $SCRATCH_MNT"
> > >  testfile=$SCRATCH_MNT/testfile-$seq
> > >
> > > --
> > > 2.31.1