Created attachment 903169 [details] ASAN report (x86_64) openrc-0.55 appears to unconditionally crash on big-endian systems, as well as the "rc-status" command. I observed this on ppc64 big-endian, sparc64, mips64, and hppa. The crashes are not reproducible on little-endian systems (including ppc64 little-endian). Boot log examples: INIT: version 3.09 booting OpenRC 0.55 is starting up Gentoo Linux (ppc64) * Mounting /proc ... [ ok ] free(): invalid pointer * Mounting /run ... [ ok ] * /run/openrc: creating directory * /run/lock: creating directory * /run/lock: correcting owner * Caching service dependencies ... [ ok ] free(): invalid pointer free(): invalid pointer INIT: Entering runlevel: 3 free(): invalid pointer INIT: version 3.09 booting OpenRC 0.55 is starting up Gentoo Linux (sparc64) * Mounting /proc ... [ 32.388529] random: crng init done [ ok ] corrupted size vs. prev_size * Mounting /run ... [ ok ] * /run/openrc: creating directory * /run/lock: creating directory * /run/lock: correcting owner * Caching service dependencies ... [ ok ] * Clock skew detected with `/etc/rc.conf' INIT: Entering runlevel: 3 corrupted size vs. prev_size Userspace tool examples: $ sudo rc-status * Caching service dependencies ... corrupted size vs. prev_size Aborted $ sudo rc-status * Caching service dependencies ... double free or corruption (!prev) Aborted These glibc abort messages vary, but the command always crashes. This is reproducible in a clean stable stage3 chroot using the "rc-status" command. awilfox was kindly able to gather ASAN reports which trigger on both ppc64 and amd64. It probably pretty soundly points to https://github.com/OpenRC/openrc/commit/0702a064a93af9a98e6954e035867882778f4268 as the culprit.
Not sure if this PR fixes the issue, but maybe?
The pull request was merged, could anyone with access to a big-endian machine test against master?
tested asan on musl (this is not reproducible on glibc), the report in rc_proc_getent got fixed by that patch, though other issues still exist i'll be going to fix those later today and ask williamh for a patch release
Confirmed this allows my BE machines to boot again.
Fixed in 0.55.1
commit 6d8e7b3d97f0b42b3280b5f31eac9a89f00ae172 Author: William Hubbs <williamh@gentoo.org> Date: Wed Sep 18 13:39:13 2024 -0500 sys-apps/openrc: add 0.55.1 Signed-off-by: William Hubbs <williamh@gentoo.org> commit b9d40c7a658101b4ec6de0fb8062cd72d02d3e9d Author: William Hubbs <williamh@gentoo.org> Date: Wed Sep 18 13:49:37 2024 -0500 sys-apps/openrc: drop 0.55 Signed-off-by: William Hubbs <williamh@gentoo.org>
