Bug 613170 – sys-libs/glibc-2.26-r2 fails sanity tests on ia64

• Status: RESOLVED FIXED
• Product: Gentoo Linux

[Description Matt Turner 2017-03-19 06:24:05 UTC]

Created attachment 467488 [details]
build.log.gz

A few excerpts:

In file included from ../include/stdio.h:3:0,
                 from test-signgam-finite-c99.c:26:
../libio/stdio.h:316:6: error: unknown type name ‘_IO_cookie_io_functions_t’
      _IO_cookie_io_functions_t __io_funcs) __THROW __wur;
      ^


In file included from test-signgam-uchar.c:3:0:
test-signgam-main.c:32:6: error: conflicting types for ‘signgam’
 TYPE signgam;
      ^
In file included from ../include/math.h:3:0,
                 from test-signgam-main.c:24,
                 from test-signgam-uchar.c:3:
../math/math.h:168:12: note: previous declaration of ‘signgam’ was here
 extern int signgam;
            ^

[Comment 1 Andreas K. Hüttel 2017-09-11 09:40:27 UTC]

Please try again with glibc-2.26-r1 (the -r1 is important).

[Comment 2 Andreas K. Hüttel 2017-10-04 09:20:36 UTC]

(In reply to Andreas K. Hüttel from comment #1)
> Please try again with glibc-2.26-r1 (the -r1 is important).

[Comment 3 Matt Turner 2017-10-04 18:41:26 UTC]

On 2.26-r1:

> *   simple run test (/usr/bin/cal) failed

:(

Similarly running ebuild test fails with

../localedata/gen-locale.sh: line 29: 12476 Bus error               ${localedef_before_env} ${run_program_env} I18NPATH=../localedata ${localedef_after_env} --quiet -c -f $charmap -i $input ${common_objpfx}localedata/$out

So, stuff's broken. Not sure if the original bug is gone, or if libc being totally busted prevents us from getting to the compilation errors...

[Comment 4 Andreas K. Hüttel 2017-10-04 19:43:41 UTC]

(In reply to Matt Turner from comment #3)
> On 2.26-r1:
> 
> > *   simple run test (/usr/bin/cal) failed
> 
> :(
> 
> Similarly running ebuild test fails with
> 
> ../localedata/gen-locale.sh: line 29: 12476 Bus error              
> ${localedef_before_env} ${run_program_env} I18NPATH=../localedata
> ${localedef_after_env} --quiet -c -f $charmap -i $input
> ${common_objpfx}localedata/$out
> 
> So, stuff's broken. Not sure if the original bug is gone, or if libc being
> totally busted prevents us from getting to the compilation errors...

Please try 2.26-r2 (bumped yesterday); that contains a LOT of upstream fixes. If that still doesnt work we have to investigate in detail.

[Comment 5 Matt Turner 2017-10-04 20:21:04 UTC]

Same with -r2 :(

[Comment 6 Andreas K. Hüttel 2017-10-04 21:09:14 UTC]

(In reply to Matt Turner from comment #5)
> Same with -r2 :(

:(

Best thing to start with would be a backtrace of a failing binaries. The sanity test just runs some basic system binaries with the newly built glibc to make sure nothing fundamental is broken.

Sergei wrote some brief instructions in https://bugs.gentoo.org/632604#c4

[Comment 7 Andreas K. Hüttel 2017-10-04 21:26:58 UTC]

Maybe https://sourceware.org/bugzilla/show_bug.cgi?id=21908

[Comment 8 Andreas K. Hüttel 2017-10-05 09:25:37 UTC]

Please also try the following patch on top of 2.26-r2 : 

https://github.com/gentoo/glibc/commit/80f9a652cbe373cd20347c03da1401bd6d3102eb.patch

[Comment 9 Andreas K. Hüttel 2018-01-17 22:34:55 UTC]

(In reply to Andreas K. Hüttel from comment #8)
> Please also try the following patch on top of 2.26-r2 : 
> 
> https://github.com/gentoo/glibc/commit/
> 80f9a652cbe373cd20347c03da1401bd6d3102eb.patch

Any luck with this or any other approach?

[Comment 10 Andreas K. Hüttel 2018-02-08 21:20:43 UTC]

(In reply to Andreas K. Hüttel from comment #9)
> (In reply to Andreas K. Hüttel from comment #8)
> > Please also try the following patch on top of 2.26-r2 : 
> > 
> > https://github.com/gentoo/glibc/commit/
> > 80f9a652cbe373cd20347c03da1401bd6d3102eb.patch
> 
> Any luck with this or any other approach?

ping

[Comment 11 Sergei Trofimovich (RETIRED) 2018-02-08 22:19:18 UTC]

2.26-r5 builds fine and passes most tests