Bug 924354
| Summary: | collect2: ld terminated with signal 11 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Product: | Red Hat Enterprise Linux 5 | Reporter: | Orion Poplawski <orion> | ||||||||
| Component: | binutils | Assignee: | Jeff Law <law> | ||||||||
| Status: | CLOSED ERRATA | QA Contact: | Martin Cermak <mcermak> | ||||||||
| Severity: | medium | Docs Contact: | |||||||||
| Priority: | unspecified | ||||||||||
| Version: | 5.10 | CC: | law, mcermak, mfranc, ohudlick, pfrankli | ||||||||
| Target Milestone: | rc | ||||||||||
| Target Release: | --- | ||||||||||
| Hardware: | ppc | ||||||||||
| OS: | Linux | ||||||||||
| Whiteboard: | |||||||||||
| Fixed In Version: | binutils-2.17.50.0.6-25.el5 | Doc Type: | Bug Fix | ||||||||
| Doc Text: |
Cause: The PPC linker did not verify certain pointers were validly non-NULL prior to dereferencing those pointers.
Consequence: As a result, under certain circumstances the PPC linker could get a segmentation fault/bus error.
Fix: The PPC linker code has been changed to properly check for NULL pointers and take appropriate action.
Result: Links no longer segfault or bus error.
|
Story Points: | --- | ||||||||
| Clone Of: | Environment: | ||||||||||
| Last Closed: | 2013-09-30 22:11:42 UTC | Type: | Bug | ||||||||
| Regression: | --- | Mount Type: | --- | ||||||||
| Documentation: | --- | CRM: | |||||||||
| Verified Versions: | Category: | --- | |||||||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||||||
| Embargoed: | |||||||||||
| Bug Depends On: | |||||||||||
| Bug Blocks: | 928849 | ||||||||||
| Attachments: |
|
||||||||||
|
Description
Orion Poplawski
2013-03-21 15:34:53 UTC
This could well end up being the same bug as 855163. Both are linker faults that are specific to ppc/ppc64 builds -- and both show up building components of the mozilla/thunderbird suites. Orion, We need a reasonable way to be able to test this. I've tried without success to set up an environment where I can build thunderbird-lightning on RHEL 5. At the minimum this environment needs to allow for installing an updated binutils linker. That would allow us to test if the fix for 855163 will fix this bug as well. If you've got a suitable srpm and a koji instance where I can throw builds, that would be a huge step forward. Jeff, I'm afraid I don't have any special access to ppc machines, I'm just using the standard Fedora/EPEL builders. If you have access to the Fedora builders you should be able submit build there. I've added ExcludeArch: ppc for now to the EL5 build, so you'd need to remove that before building. You should be able to check out the el5 branch from the Fedora git repo for thunderbird-lightning. When I try to fire off a ppc koji build, it claims not to know about the epel target. [law@stumpy thunderbird-lightning]$ ppc-koji build --scratch el5 thunderbird-lightning-1.9.1-1.el5.src.rpm Usage: koji build [options] target <srpm path or scm url> (Specify the --help global option for a list of other help options) koji: error: Unknown build target: el5 [law@stumpy thunderbird-lightning]$ ppc-koji list-targets Name Buildroot Destination --------------------------------------------------------------------------------------------- dist-f12 dist-f12-build dist-f12 dist-f12-updates-candidate dist-f12-build dist-f12-updates-candidate dist-f13-updates-candidate dist-f13-build dist-f13-updates-candidate dist-f14 dist-f14-build dist-f14 dist-f15 dist-f15-build dist-f15 dist-f15-updates-candidate dist-f15-build dist-f15-updates-candidate dist-f16 f16-build f16 f17 f17-build f17 f17-candidate f17-build f17-updates-candidate f18 f18-build f18-updates-candidate f18-candidate f18-build f18-updates-candidate f19 f19-build f19-updates-candidate f19-candidate f19-build f19-updates-candidate f19-rebuild f19-build f19-rebuild f20 f20-build f20 f20-candidate f20-build f20 rawhide f20-build f20 rawhide-repo-holder rawhide rawhide How precisely are you throwing ppc builds into koji? I would use koji, not ppc-koji koji build --scratch --arch-override=ppc dist-5E-epel ... I got it to fail after some hacking late last night. Right now it appears to be a different problem than was fixed for 855163. I'll be debugging today. Hopefully it's something already fixed upstream with an associated testcase :-) Created attachment 734934 [details]
Self-contained test
Extract on a ppc box.
sh doit
If the test segfaults, then it failed. Anything else is a pass.
Upstream fix identified. Doing a full build of lightning with a patched up binutils. Created attachment 735109 [details]
Patch from upstream
Created attachment 735963 [details]
Testcase for related ppc64 issue
Extract on a ppc64 machine and run ./doit
Segfault is a failure, anything else is a success.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHBA-2013-1306.html |