Summary: | sys-devel/binutils-2.19 fails tests due to pax patch | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Alexis Ballier <aballier> |
Component: | Current packages | Assignee: | Gentoo Toolchain Maintainers <toolchain> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | amd64, bircoph, brant, hardened, jfostiguy, kentnl, ole+gentoo, pageexec, patrick |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
ld.log selected failures
( Partial ) Replacement for Listed PAX Patch Misc emerge/paludis --info datas |
Description
Alexis Ballier
2008-11-01 10:50:12 UTC
the pax patch needs test updating ... i totally dont feel like doing it atm (In reply to comment #1) > the pax patch needs test updating ... i totally dont feel like doing it atm something like http://pax.grsecurity.net/binutils-2.19-pt-pax-flags-200811041810.patch ? Created attachment 173548 [details]
ld.log selected failures
Failures from LD.log 's test. ( after applying test-patches )
Created attachment 173550 [details]
( Partial ) Replacement for Listed PAX Patch
(In reply to comment #2) >> the pax patch needs test updating ... i totally dont feel like doing it atm > >something like >http://pax.grsecurity.net/binutils-2.19-pt-pax-flags-200811041810.patch ? It would appear the general gist of that patchset has already appeared in the codebase, however, the patches to the test have not been applied, so I hand replicated those changes and generated a patch ( attached ) that replicated those tests changes. However, they don't appear to have any bearing of the results of the tests. I have attached ( earlier ) a log from the test suite after applying the patch, and it would appear that there is much more wrong with it in things not affected by the patch. Created attachment 173552 [details]
Misc emerge/paludis --info datas
Attaching this diagnosis because I find it intruiging how:
1) We have similar processors
2) We have similar CFLAGS
3) We have similar LDFLAGS
( Although you'll note, env data from the build says the cflags are stripped back to sane value )
pax guys: that patch includes fix ups for x86 only ... in the past i had to fixup the tests for every arch the test changes in general should be much simpler than in the past as i changed upstream's regexes a bit to be more general ive fixed most x86_64 failures and tweak the i386 tests http://sources.gentoo.org/gentoo/src/patchsets/binutils/2.19/63_all_binutils-2.19-pt-pax-flags-20081101.patch (In reply to comment #8) > ive fixed most x86_64 failures and tweak the i386 tests > > http://sources.gentoo.org/gentoo/src/patchsets/binutils/2.19/63_all_binutils-2.19-pt-pax-flags-20081101.patch > With patch applied: # of expected passes 427 # of unexpected failures 20 # of expected failures 57 The similar problem here: 2.19.1-r1 fails with 43 unexpected failures on two ~x86 boxes with glibc-2.10.1. 2.19.1-r1 doesn't build with USE="vanilla", but 2.19.51.0.14 build OK with USE="vanilla" and passes all tests successfully. Moreover, it compiles OK =x11-drivers/xf86-video-intel-2.8.0, while ld from 2.19.1-r1 segfaults at linking of intel video driver. binutils-2.20-r1 and binutils-2.20.1 should pass all tests now. if you hit a failure after syncing, file a new report. |