From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mx2.suse.de ([195.135.220.15]:57460 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750728AbcCXMNT (ORCPT ); Thu, 24 Mar 2016 08:13:19 -0400 Date: Thu, 24 Mar 2016 13:13:47 +0100 From: Jan Kara Subject: Re: [PATCH] generic/338: Add mmap race test Message-ID: <20160324121347.GD4025@quack.suse.cz> References: <1458740555-3512-1-git-send-email-jack@suse.cz> <20160324114934.GU11419@eguan.usersys.redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20160324114934.GU11419@eguan.usersys.redhat.com> Sender: fstests-owner@vger.kernel.org To: Eryu Guan Cc: Jan Kara , fstests@vger.kernel.org, Toshi Kani List-ID: On Thu 24-03-16 19:49:34, Eryu Guan wrote: > On Wed, Mar 23, 2016 at 02:42:35PM +0100, Jan Kara wrote: > > Add test which spawns two threads racing to write to file via mmap and > > checks the result. This is mainly interesting to uncover races in DAX > > fault handling. > > > > Signed-off-by: Jan Kara > > --- > > src/Makefile | 5 +- > > src/holetest.c | 342 ++++++++++++++++++++++++++++++++++++++++++++++++++ > > tests/generic/338 | 57 +++++++++ > > tests/generic/338.out | 73 +++++++++++ > > tests/generic/group | 1 + > > 5 files changed, 476 insertions(+), 2 deletions(-) > > create mode 100644 src/holetest.c > > create mode 100755 tests/generic/338 > > create mode 100644 tests/generic/338.out > > 338 is taken, 340 is free now :) OK, will update. > [snip] > > + > > +seq=`basename $0` > > +seqres=$RESULT_DIR/$seq > > +echo "QA output created by $seq" > > +tmp=/tmp/$$ > > +status=1 # failure is the default! > > +trap "_cleanup; exit \$status" 0 1 2 3 15 > > + > > +_cleanup() > > +{ > > + cd / > > + rm -f $tmp.* > > +} > > + > > +# get standard environment and checks > > +. ./common/rc > > + > > +# real QA test starts here > > +_supported_fs generic > > +_supported_os Linux > > +_require_scratch > > Need to make sure "holetest" is built here, and there's a new helper to > do it now > > _require_test_program "holetest" OK, thanks. > And test fails when I tested on ext3/2/NFS/CIFS, because they have no > fallocate support, so need to test for fallocate support as well > > _require_xfs_io_command "falloc" Actually, I have modified the test program not to use fallocate(2) but instead just use posix_fallocate(3). That uses fallocate(2) when available and falls back to write when not which is what we wanted and using both was just duplicating effort needlessly. Thanks for review. Honza -- Jan Kara SUSE Labs, CR