Bug 104173

Summary: xmms (and other applications) crash with message libgcc_s.so.1 must be installed for pthread_cancel to work
Product: [Retired] Red Hat Linux Beta Reporter: Keith Lea <keith>
Component: glibcAssignee: Jakub Jelinek <jakub>
Status: CLOSED NOTABUG QA Contact:
Severity: high Docs Contact:
Priority: medium    
Version: beta1CC: fweimer, notting
Target Milestone: ---   
Target Release: ---   
Hardware: i686   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2003-09-11 07:39:52 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On:    
Bug Blocks: 100643    

Description Keith Lea 2003-09-10 18:40:36 UTC
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5b) Gecko/20030909
Firebird/0.6.1+

Description of problem:
I'm not sure if glibc is the right component for this, but the only other bug
involving pthread_cancel is under glibc.

XMMS crashes (usually only when switching songs) and prints "libgcc_s.so.1 must
be installed for pthread_cancel to work" to the console. Other applications have
done this as well but I don't remember what they were now. XMMS does it after
almost every song.

I am using kernel-2.4.22-1.2039.nptl and I do have libgcc_s.so.1:

lrwxrwxrwx    1 root     root           28 Sep 10 00:23 /lib/libgcc_s.so.1 ->
libgcc_s-3.3.1-20030904.so.1*
-rwxr-xr-x    1 root     root        29196 Sep  8 13:53
/lib/libgcc_s-3.3.1-20030904.so.1*

This happens with XMMS 1.2.7 and 1.2.8.


Version-Release number of selected component (if applicable):
glibc-2.3.2-82

How reproducible:
Always

Steps to Reproduce:
1. Run XMMS
2. Switch songs or let a song finish so the next song is played

Additional info:

Comment 1 Jakub Jelinek 2003-09-10 21:43:08 UTC
Are you sure you don't have a different (older) libgcc_s.so.1 lurking somewhere
in your library search path (if LD_LIBRARY_PATH is used there, if not,
in ld.so.cache)?
pthread_cancel is trying to dlopen libgcc_s.so.1, which has:
_Unwind_Resume
__gcc_personality_v0
_Unwind_ForcedUnwind
_Unwind_GetCFA
symbols exported from it.
E.g. strace output (or LD_DEBUG=all dump) could make it clearer what's going on.

Comment 2 Keith Lea 2003-09-11 00:06:32 UTC
It looks like I did have another libgcc_s.so.1 in /usr/local/lib, and removing
it fixes the crash. Removing /usr/local/lib/libstdc++.* also fixes my GLIBC
errors in other applications. I don't know if I'm supposed to resolve the bug,
so I won't, but I'm all set now. Thanks for the help. (Any idea why this would
happen? This is sort of an old system (I started with RH7.3, so it's not thta
old), but I don't remember ever manually installing glibc or gcc or anything.)

Comment 3 Jakub Jelinek 2003-09-11 07:39:52 UTC
That really looks like manually installed gcc.

Comment 4 Simon 2004-05-03 18:57:28 UTC
Worth looking at this other bug:

http://bugzilla.redhat.com/bugzilla/long_list.cgi?buglist=104689

Similar symptoms that boiled down to a problem with NPTL. Solution:

export LD_ASSUME_KERNEL=2.4.19 

... before running xmms. Works wonders for me, too! :)