| Summary: | LSB builds fail in kernel sanity test with xmalloc failure | ||
|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] OE-Core | Reporter: | Ross Burton <ross.burton> |
| Component: | core | Assignee: | Aníbal Limón <anibal.limon> |
| Status: | VERIFIED FIXED | QA Contact: | Lucian Musat <georgex.l.musat> |
| Severity: | normal | ||
| Priority: | High | CC: | alexandru.c.georgescu, anibal.limon, benjamin.esquivel, mariano.lopez, markus.lehtonen, meta.mr.watcher, meta.watcher, richard.purdie |
| Version: | 1.9 | ||
| Target Milestone: | 2.0 | ||
| Hardware: | All | ||
| OS: | Multiple | ||
| Whiteboard: | |||
| OS type for building Yocto: | --- | Type of Regression: | Regression (Used to work) |
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |
|
Description
Ross Burton
2015-10-07 10:05:46 UTC
One of the things that look strange here is that the test is not triggering the logic for gathering logs to see the host/target context (mem, proc, dmesg, etc). Mariano did an intentional failure of the test and it *should* be presenting such logs. The first thing that occurs to us is to check what is going on with the host memory at the time of failure so we'll send a forced build with a patch that writes periodically the memory status to a file in the disc so we can come later an analyze it. One of the theories is that if the Ram is not being sufficient then the process for gathering the logs cannot start as well. As I've mentioned elsewhere, if you download one of the broken images and boot it locally, you can reproduce this the first time you run "make scripts" inside an image. I've run top in an image whilst showing the bug and I didn't see any significant resource problem. If you google the error, there are comments about PAM. We enable PAM in poky-lsb. We're also seeing this bug on x86, x86-64, arm and ppc so it looks like its something strucutural in our images, not something kernel/arch specific. I tested with arm lsb sdk image that fails and the problem is caused by bash custom malloc. I found it because i disabled bash custom malloc passing --without-bash-malloc to configure and all works good. Also of note is that the latest master runs on the AB don't fail like this... I found that all the distr(In reply to comment #3) > I tested with arm lsb sdk image that fails and the problem is caused by bash > custom malloc. > > I found it because i disabled bash custom malloc passing > --without-bash-malloc to configure and all works good. I found that common distros don't use it, for example: Debian & Ubuntu: http://bazaar.launchpad.net/~doko/+junk/pkg-bash-debian/view/head:/rules#L89 Fedora 20 (extract and see spec): ftp://rpmfind.net/linux/fedora-secondary/development/rawhide/source/SRPMS/b/bash-4.2.45-4.fc20.src.rpm LFS: http://www.linuxfromscratch.org/lfs/view/development/chapter06/bash.html So there is no reason for maintain enabled. Merge at, http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=05594f86366d97975501ffd84886d37f0c72e238 Having been seeing this of late so setting this as verified (on commit fc45deac89ef63ca1c44e763c38ced7dfd72cbe1) |