Bug 201376

Summary:	sys-libs/glibc[profile]: programs linked with -lc_p always segfault early
Product:	Gentoo Linux	Reporter:	Jeffrey Graham <grahamspam>
Component:	[OLD] Core system	Assignee:	Gentoo Toolchain Maintainers <toolchain>
Status:	RESOLVED UPSTREAM
Severity:	major	CC:	nbowler, tanarrian
Priority:	High
Version:	unspecified
Hardware:	x86
OS:	Linux
URL:	https://sourceware.org/bugzilla/show_bug.cgi?id=9689
See Also:	https://sourceware.org/bugzilla/show_bug.cgi?id=9689
Whiteboard:
Package list:		Runtime testing required:	---

Description Jeffrey Graham 2007-12-05 16:43:26 UTC

Added use flag profile to /etc/make.conf and re-emerged glibc 2.6.1 to get libc_p.a

Built my app and linked it against libc_p.a like this:
gcc test.c -g -pg -O0 -o test -static -lc_p

The program segfaults. The debugger reports this:
Program received signal SIGSEGV, Segmentation fault.
0x08083aeb in __mcount_internal().


This bug has been reported to other distributions...
http://people.debian.org/~terpstra/message/20070917.132524.9ab97a81.en.html



Reproducible: Always

Comment 1 SpanKY gentoo-dev

2008-09-07 06:39:36 UTC

*** Bug 236524 has been marked as a duplicate of this bug. ***

Comment 2 SpanKY gentoo-dev

2008-12-27 10:32:58 UTC

still happens with glibc-2.9 :/

$ gcc test.c -g -pg -O0 -o test -static -lc_p
$ gdb ./test
(gdb) r
Starting program: /home/vapier/test

Program received signal SIGSEGV, Segmentation fault.
0x000000000042fecb in __mcount_internal ()
(gdb) bt
#0  0x000000000042fecb in __mcount_internal ()
#1  0x000000000040dc04 in mcount ()
#2  0x0000000000000000 in ?? ()
(gdb)

Comment 3 SpanKY gentoo-dev

2008-12-27 22:02:32 UTC

Debian bug report (not that it has any info):
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=506850

the crash seems to be because mcount uses atomic code which relies on the TLS descriptors being setup, but that doesnt happen until __libc_start_main.  but since __libc_start_main itself calls mcount, we crash.

at any rate, moved upstream