Summary: | coreutils 5.0-r5 majorly broken : cp mv cat ls rm etc... | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Alexandre Rostovtsev (RETIRED) <tetromino> |
Component: | New packages | Assignee: | Seemant Kulleen (RETIRED) <seemant> |
Status: | RESOLVED FIXED | ||
Severity: | critical | ||
Priority: | High | ||
Version: | unspecified | ||
Hardware: | x86 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Alexandre Rostovtsev (RETIRED)
2003-12-08 20:01:07 UTC
OK, I figured out what went wrong. First, it's compiler flag -malign-double (more specifically, "-march=athlon-tbird -O2 -pipe -frename-registers -fomit-frame-pointer -malign-double" fails, while "-march=athlon-tbird -O2 -pipe -frename-registers -fomit-frame-pointer" and "-march=i686 -O2 -pipe -frename-registers -fomit-frame-pointer" both succeed. Second, coreutils 5.0.91-r2 also has the same problem with -malign-double. Maybe someone could change the ebuilds to check CFLAGS ? I've added the filter to coreutils, though we don't take responsibility for a user's own sense of adventure. That's honestly their own problem, not ours. Ahhh...too late... Even though, as already mentioned in the previous comment, it's highly adventurous (read: not recommended) to put `-malign-double' in CFLAGS, it's one of those flags that affect structure sizes and binary compatibility and has to be used either nowhere or everywhere (incl. binary-only packages). Filtering only in a few selected ebuilds may produce some hard to reproduce bug reports. in that case, I'm pulling it from the coreutils ebuilds. if you put malign double in your flags you're on your er, malignment, not ours |