| Summary: | dev-libs/beecrypt-4.1.2-r1 failed | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Robert Golding <robert.golding> |
| Component: | Current packages | Assignee: | Sandro Bonazzola (RETIRED) <sanchan> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | amd64, caster, ispeters |
| Priority: | High | ||
| Version: | 2006.1 | ||
| Hardware: | AMD64 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | emerge output for beecrypt-4.1.2-r1 | ||
|
Description
Robert Golding
2006-12-14 01:28:09 UTC
*** Bug 158111 has been marked as a duplicate of this bug. *** (In reply to comment #1) > *** Bug 158111 has been marked as a duplicate of this bug. *** > This has happened three times now, I submit a bug and it duplicates itself .. why? I only click on the commit button once. Is it the browser, or my ISP? Created attachment 104009 [details]
emerge output for beecrypt-4.1.2-r1
Full failed emerge output for beecrypt-4.1.2-r1
Ok, tried emerge with CFLAGS="-march=i686 -O2 -pipe" and it emerged. I don't know what this will do to my general setup though, or if it will effect something else in turn. Will let you know if it busts anything else. I tried this because of the following warning in the ebuild; --> ewarn "The MMX assembler code presents TEXTREL issues. If you don't want them try using" ewarn "relaxed CFLAGS like -march=i686 instead of -march=pentium3 and so on." ewarn "This ebuild fails on multilib system with multilib-strict on AMD64." ewarn "Feel free to help upstream solving the above bugs." ------> What does that all mean? (In reply to comment #4) > Ok, tried emerge with CFLAGS="-march=i686 -O2 -pipe" and it emerged. Looking at your emerge log: i686-pc-linux-gnu-gcc -DHAVE_CONFIG_H -I. -I. -I. -I./.. -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -DOPTIMIZE_X86_64 -march=athlon64 -O2 -pipe -fomit-frame-pointer -DNDEBUG -fomit-frame-pointer -Wa,--noexecstack -c base64.c -fPIC -DPIC -o .libs/base64.o it seems that using -march=athlon64 with x86 arch exploit a bug in configure script. Using athlon64 let configure assume you're building for amd64 arch. > I tried this because of the following warning in the ebuild; > --> > ewarn "The MMX assembler code presents TEXTREL issues. If you don't want them > try using" > ewarn "relaxed CFLAGS like -march=i686 instead of -march=pentium3 and so on." > ewarn "This ebuild fails on multilib system with multilib-strict on AMD64." > ewarn "Feel free to help upstream solving the above bugs." > ------> > What does that all mean? that the ebuild could fail if you use FEATURES="stricter" as first issue. If you want to learn more about textrel see "HOWTO Locate and Fix .text Relocations (TEXTRELs)" http://www.gentoo.org/proj/en/hardened/pic-fix-guide.xml CCing amd64 people because this bug affect only -march=athlon64 even if only with x86 arch. I'll try to solve this next week, I'll be away until monday. Fixed in cvs. Commit message: Fix incorrect behaviour with CHOST="i686-..." and CFLAGS="-march=athlon64 ...", bug #158109. Thanks to Robert Golding <robert.golding@gmail.com> for reporting. Solution adopted: Filtering with-arch=athlon64 when CHOST doesn't begin with x86_64, reverting to CHOST as default. *** Bug 158724 has been marked as a duplicate of this bug. *** I(In reply to comment #8) > *** Bug 158724 has been marked as a duplicate of this bug. *** > Fixed in CVS. Fix incorrect behaviour with CHOST!="x86_64-..." and any -march flags related to x86_64, including athlon64 and k8 fixing also bug #158724. |
