| Summary: | x11-libs/gtk+-2.16.6 fails tests on ppc64 | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Alexis Ballier <aballier> |
| Component: | Current packages | Assignee: | Gentoo Linux Gnome Desktop Team <gnome> |
| Status: | RESOLVED FIXED | ||
| Severity: | minor | CC: | ppc64 |
| Priority: | High | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| URL: | https://bugzilla.gnome.org/show_bug.cgi?id=595746 | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
|
Description
Alexis Ballier
2009-09-20 13:36:45 UTC
* QA Notice: Package has poor programming practices which may compile * fine but exhibit random runtime failures. * gtkcellrendereraccel.c:464: warning: implicit declaration of function ‘gtk_grab_remove’ * gtkcellrendereraccel.c:602: warning: implicit declaration of function ‘gtk_grab_add’ ^^ These cause the failures. If during compilation of C libraries the declaration aren't explicit, the compiler doesn't have a chance to avoid a PLT jump for local symbols on some arches apparently, which glib/pango/gtk+ try to heavily avoid, as to even have a test for not having any PLT jumps of this kind in the library at all (it helps with startup time). Very interesting blog-worthy side-effect of implicit declaration QA warnings indeed. It might be special to glib/pango/gtk+ though, as they pull some gcc hacks to avoid these R_*_JMP_SLOT things for local symbols with most gcc versions - I think newer ones can do it automatically in some situations. I saw the failure on my x86 desktop as well, but interestingly not on amd64. I'll fix it in a bit. https://bugzilla.gnome.org/show_bug.cgi?id=448253#c5 has an upstream reference for the connection between pltcheck and implicit declarations. Anyhow, fixed in CVS. |