From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Petazzoni Date: Tue, 14 Aug 2018 22:47:17 +0200 Subject: [Buildroot] [PATCH v3 1/2] package/lvm2: Fix runtime crash when using uclibc In-Reply-To: <20180726171524.479-1-m.niestroj@grinn-global.com> References: <20180726171524.479-1-m.niestroj@grinn-global.com> Message-ID: <20180814224717.02d57e06@windsurf> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net Hello Marcin, Once again, thanks for this work! On Thu, 26 Jul 2018 19:15:23 +0200, Marcin Niestroj wrote: > When using uclibc libdevmapper.so was calling dm_task_get_info_base() > function recursively, leading to segmentation fault. This was > happening because uclibc linker loader just takes first existing > 'dm_task_get_info' (which is 'dm_task_get_info_base') symbol in elf > binary, instead of default version. > > Add upstreamable lvm2 patch [1], which introduces > --enable-symvers[=STYLE] switch. Use that switch to disable symbol > versions, as we do not plan to support binaries compiled against > old libdevmapper library. > > [1] https://www.redhat.com/archives/dm-devel/2018-July/msg00187.html > > Signed-off-by: Marcin Niestroj > --- > Changes v2 -> v3: none I have added a reference to the Buildroot bug report being fixed by this, and I've applied to master. Two questions: (1) Could you try to push this forward in terms of upstreaming acceptance ? I know you have submitted the patch, but I looked today and apparently you haven't received any feedback. (2) It would be nicer if the availability of symbol versioning could be auto-detected. Is there a compile-time test that can be done to verify if symbol versioning is available ? This would perhaps help make the patch even more acceptable upstream. Thanks! Thomas -- Thomas Petazzoni, CTO, Bootlin (formerly Free Electrons) Embedded Linux and Kernel engineering https://bootlin.com