From: Eric Whitney Subject: [PATCH] ext4/271: improve mkfs and mount error handling Date: Thu, 2 Oct 2014 16:32:48 -0400 Message-ID: <20141002203248.GB2364@wallace> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4@vger.kernel.org To: fstests@vger.kernel.org Return-path: Received: from mail-qa0-f46.google.com ([209.85.216.46]:49009 "EHLO mail-qa0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751885AbaJBUcz (ORCPT ); Thu, 2 Oct 2014 16:32:55 -0400 Content-Disposition: inline Sender: linux-ext4-owner@vger.kernel.org List-ID: If 271's attempt to make a scratch device fails, it can inadvertently mount and run on a pre-existing scratch file system. This can lead to invalid test results unless the mkfs failure is separately noted in 271.full. Noted while debugging an ext4 inline data test failure. Small ext4 file systems, as used by this test, use small inodes that are incompatible with inline data. Signed-off-by: Eric Whitney --- tests/ext4/271 | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/tests/ext4/271 b/tests/ext4/271 index e77ae57..44d9147 100755 --- a/tests/ext4/271 +++ b/tests/ext4/271 @@ -41,11 +41,14 @@ _supported_os Linux _need_to_be_root _require_scratch -_scratch_mkfs_sized $((128 * 1024 * 1024)) >> $seqres.full 2>&1 +rm -f $seqres.full + +_scratch_mkfs_sized $((128 * 1024 * 1024)) >> $seqres.full 2>&1 \ + || _fail "mkfs failed" # -onoload and EXT4_SYNC_FL on file is important becase result in # metadata sync writes inside ext4_handle_dirty_metadata() -_scratch_mount -onoload +_scratch_mount -onoload >> $seqres.full 2>&1 || _fail "mount failed" touch $SCRATCH_MNT/file chattr +S $SCRATCH_MNT/file # Create sparse file -- 1.9.1