Summary: | [fPIC] dev-util/valgrind-3.3.0 fails to compile for x86_64 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Juergen Rose <rose> |
Component: | Current packages | Assignee: | Michael Imhof (RETIRED) <tantive> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | amd64, chad.simmons, didier.fabert, griffon26, hp-cluster, jean.brefort, jsbronder, patrizio.bassi |
Priority: | High | ||
Version: | 2007.0 | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Juergen Rose
2007-12-18 11:53:05 UTC
See here for info: http://bugs.kde.org/show_bug.cgi?id=154240 Unfortunately I don't have time to work on this until about January 4. Can you please provide the output of: equery belongs /usr/lib64/libmpich.a I have the same error, output of "equery belongs /usr/lib/libmpich.a" is sys-cluster/mpich-1.2.7_p1 (/usr/lib/libmpich.a) Can someone explain why it is not linking against a shared object? If there is no libmpich.so, should libmpich.a be built with -fPIC? (In reply to comment #4) > Can someone explain why it is not linking against a shared object? > If there is no libmpich.so, should libmpich.a be built with -fPIC? > It appears that if you don't pass in --without-mpicc, valgrind is nice enough to scan your system and try to use it anyways. A quick fix until tantive gets some time to fix mpich (if that's possible) would probably be to pass --without-mpicc to econf. That worked for me against lam-mpi and openmpi. Could someone verify this also works with mpich? no change here since months? ping! I've added --without-mpicc to the valgrind ebuild... If you ever get to fix mpich to produce dynamic libs, I guess you can replace it with a use-dep and $(use_with mpi mpicc) ... but the fix is not in portage, i still cannot compile it... Olivier/Michael, can we do something for this? 3.3.1 is out too needs a bump About the bump: what do you mean? 3.3.1 is in the tree already. oh! for any reasons my emerge -uD world didn't show it. bumping to 3.3.1 fixed the issue, great. you can close this. (In reply to comment #12) > oh! for any reasons my emerge -uD world didn't show it. > > bumping to 3.3.1 fixed the issue, great. > > you can close this. > 3.3.1 works for me as well, closing as fixed. |