Bug 125169 – glibc-2.3.6 fails with nptl -nptlonly on ppc

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description Joe Jezak (RETIRED) 2006-03-05 20:54:34 UTC]

With gcc-4.1.0, initfini.s needs to be compiled with -fno_unit_at_a_time, the correct patch will be added to the bug.

[Comment 1 Joe Jezak (RETIRED) 2006-03-05 20:55:07 UTC]

Created attachment 81486 [details, diff]
Correct patch #6100

[Comment 2 SpanKY 2006-03-05 21:11:22 UTC]

why ?  upstream glibc doesnt do it ... the current 6100 patch is what current glibc trunk is doing

[Comment 3 Chris White (RETIRED) 2006-03-05 21:34:01 UTC]

http://sourceware.org/ml/libc-alpha/2005-08/msg00002.html

this is what was found to describe the situation and why the patch needs to be as per the attachment.

[Comment 4 Joe Jezak (RETIRED) 2006-03-05 22:11:38 UTC]

Hmm, actually, the current patch does use += instead of =, so it should be passing the flag up instead of overriding it.  But I'm still getting the same errors as that original bug when compiling linuxthreads.

[Comment 5 SpanKY 2006-03-06 15:04:16 UTC]

... and if you go down through that thread you'll see that they end up at the same patch that we already have in cvs

$(fno-unit-at-a-time) is added in csu/Makefile for all initfini.s targets

[Comment 6 Joe Jezak (RETIRED) 2006-03-06 18:51:54 UTC]

It's still broken on linuxthreads for me with gcc-4.1.0 with the exact same error.  nptl compiles fine. 

[Comment 7 SpanKY 2006-03-11 01:06:34 UTC]

*** Bug 125731 has been marked as a duplicate of this bug. ***

[Comment 8 Joe Jezak (RETIRED) 2006-03-14 11:36:18 UTC]

Created attachment 82127 [details, diff]
Fixes linuxthreads compile issue on ppc and probably ppc64 as well

[Comment 9 SpanKY 2006-03-14 15:52:06 UTC]

thanks ... sent upstream and added to cvs ... this will be in glibc-2.3.6-r4