Summary: | gnustep-base/gnustep-base-1.24.8: configure fails with MPROTECT on | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | lorem.ipsum.989 |
Component: | [OLD] Library | Assignee: | Gentoo Gnustep project <gnustep> |
Status: | RESOLVED INVALID | ||
Severity: | normal | CC: | alexander, stian |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
My emerge --info
config.log build.log |
Description
lorem.ipsum.989
2015-11-08 08:08:10 UTC
Created attachment 416270 [details]
My emerge --info
Created attachment 416272 [details]
config.log
Created attachment 416274 [details]
build.log
Thanks for the report! I know that all the libffi parts are very sensitive on PaX kernels, but nowadays it should work OK. Can you double-check all depending packages have been built correctly with USE=pax_kernel (especially libffi itself)? (In reply to Bernard Cafarelli from comment #4) > Thanks for the report! > > I know that all the libffi parts are very sensitive on PaX kernels, but > nowadays it should work OK. > Can you double-check all depending packages have been built correctly with > USE=pax_kernel (especially libffi itself)? Yes, I recompiled @world right after switching to hardened, gnustep-base is the only package that failed to build; libffi installs perfectly fine. A dirty solution I used to make it merge is simply disabling MPROTECT on conftest in the configure script: once installed, the library seems to work just fine (I need it for app-arch/unar, which I'm not experiencing any issues with). Oops, my bad. After I recompiled libffi again, everything installs just OK. NOTABUG. Same problem here, tried both version 1.24.6-r1 and version 1.24.8-r1 of gnustep-base. Recompiling libffi (3.0.13-r1 or 3.2.1) does not fix the problem, USE=pax_kernel is enabled. |