Summary: | media-gfx/imagemagick 6.2.4.2 fails to emerge | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | El Fabre <elfabre> |
Component: | New packages | Assignee: | Karol Wojtaszek (RETIRED) <sekretarz> |
Status: | RESOLVED DUPLICATE | ||
Severity: | normal | CC: | dirk.heinrichs.ext, elfabre, graphics+disabled, info, vapier |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
Complete building process output
emerge info output |
Description
El Fabre
2005-10-06 15:06:06 UTC
Created attachment 70017 [details]
Complete building process output
I get the same error on a K6 system for imagemagick, as well as for jpeg-6b-r6: ./libtool-wrap --mode=compile i586-pc-linux-gnu-gcc -march=k6-2 -O3 -fomit-frame-pointer -pipe -I. -c ./jdcolor.c i586-pc-linux-gnu-gcc -march=k6-2 -O3 -fomit-frame-pointer -pipe -I. -c ./jdcolor.c -fPIC -DPIC -o .libs/jdcolor.o {standard input}: Assembler messages: {standard input}:106: Error: value of ffffffffffffff7f too large for field of 1 bytes at 0000000000000156 make: *** [jdcolor.lo] Error 1 (In reply to comment #2) > I get the same error on a K6 system for imagemagick, as well as for jpeg-6b-r6: > Nope! I only get it on imagemagick. I just recompiled jpeg-6b-r6 again on purpose. It went through just fine. Can you pls post "emerge --info" and "emerge --verbose --pretend jpeg" so I can try to recompile it with your Useflag settings? Tx. El Fabre Created attachment 70338 [details]
emerge info output
jpeg doesn't have any use flags. However, emerge info output is attached. (In reply to comment #5) > jpeg doesn't have any use flags. However, emerge info output is attached. > You are right! no use flags. I see a huge gcc version difference. Wondering where you got that gcc 3.4.4. Explanation at http://bugs.gentoo.org/show_bug.cgi?id=104953 is a cryptic one. GCC 3.4.4 comes out of portage. Concerning the other bug: I've seen it before, but it doesn't have any useful information. I can confirm this happens on my machine too. Any workaround/fix? real easy workaround, dont use -march=k6-2 *** Bug 111674 has been marked as a duplicate of this bug. *** Obviously, this is not an ImageMagick problem but a much more general issue related to gcc optimizations on AMD K6. I got it on ImageMagick but Dirk gets it on jpeg-6b-r6 (I don't) and now I found that somebody had already noticed it under gimp (bug #30482) back in 2003! Aside from disabling the optimizations, as suggested (it does work), it looks like there is no solution for this, yet. *** This bug has been marked as a duplicate of 30482 *** Actually, looks like the resolution is coming up: GCC 4.1.x onward http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20653 |