From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alasdair G Kergon Date: Fri, 26 Mar 2010 18:31:39 +0000 Subject: [PATCH] Remove vgname check from _find_vg_rlocn and just return the rlocn slot. In-Reply-To: <1269625126-18152-1-git-send-email-dwysocha@redhat.com> References: <1269620108.2508.3.camel@f10-node1> <1269625126-18152-1-git-send-email-dwysocha@redhat.com> Message-ID: <20100326183139.GS5259@agk-dp.fab.redhat.com> List-Id: To: lvm-devel@redhat.com MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit On Fri, Mar 26, 2010 at 01:38:46PM -0400, Dave Wysochanski wrote: > Unclear what the purpose of this check was/is. Removing it allows for easier > refactoring, fixes at least one bug, and at least breaks no tests. If it > covers and important case, we should add a comment and/or testcase. The on-disk layout supports multiple mdas for different VGs stored on the same disk. But at the moment we only use one. [Create a test case: PV with 2 mdas, and different vgname in each...current code should ignore the 2nd one, maybe overwriting it with copy of first, I'm not sure.] > - /* FIXME Loop through rlocns two-at-a-time. List null-terminated. */ > - /* FIXME Ignore if checksum incorrect!!! */ FIXME removed, so does the patch now validate the checksum? I think the vgname validation is still needed on almost all occasions. Only the 'tell me which VG is in this mda' case doesn't want it. Alasdair