Summary: | app-text/ghostscript-gpl-9.04-r1 fails to build on ~x86 without SSE2 support | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Andrew Savchenko <bircoph> |
Component: | [OLD] Printing | Assignee: | Printing Team <printing> |
Status: | RESOLVED INVALID | ||
Severity: | normal | CC: | bircoph |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
URL: | http://bugs.ghostscript.com/show_bug.cgi?id=692464 | ||
Whiteboard: | tracking upstream | ||
Package list: | Runtime testing required: | --- | |
Attachments: |
build.log
emerge --info ghostscript-gpl-9.04-sse2.patch ghostscript-gpl-9.04-r1.ebuild.patch |
Description
Andrew Savchenko
![]() Created attachment 283673 [details]
build.log
Created attachment 283675 [details]
emerge --info
MAKEOPTS="-j1" FEATURES="-distcc -ccache" emerge ghostscript-gpl fails in the same way. Created attachment 283723 [details, diff]
ghostscript-gpl-9.04-sse2.patch
-DHAVE_SSE2 was mistakenly added to the compilation flags, because of malfunctioned sse2 check in configure script. Compilation test must be used instead of linking test.
Proposed patch fixes this bug.
Created attachment 283727 [details, diff]
ghostscript-gpl-9.04-r1.ebuild.patch
SSE2 patch must be applied before eautoreconf.
(In reply to comment #4) > Proposed patch fixes this bug. I've created an upstream bug, see URL. (In reply to comment #2) > emerge --info Please confirm that it also fails when using sane(er) CLFAGS/LDFLAGS like: CFLAGS="-march=native -O2 -pipe" CXXFLAGS="${CFLAGS}" LDFLAGS="-Wl,-O1 -Wl,--as-needed" Hmm, it compiles with my default CFLAGS/CXXFLAGS and simplified LDFLAGS="-Wl,-O1,--as-needed". Though, original LDFLAGS are formally valid. (In reply to comment #8) > it compiles with my default CFLAGS/CXXFLAGS and simplified > LDFLAGS="-Wl,-O1,--as-needed". Though, original LDFLAGS are formally valid. Well I can't judge on this since your flags are quite.. huge and beyond my knowledge. But shouldn't at least -march be the same for CFLAGS and LDFLAGS? And at least my Atom CPU supports SSE2 hence perhaps the problem? If you find out the problematic flag and if it's really not something unsupported or the like, please solve it upstream and reopen this bug. |