Bug 29118 – libgsf 1.8.2 breaks nautilus

• Product: Gentoo Linux
• Status: RESOLVED INVALID

[Description André Lemos 2003-09-19 06:30:24 UTC]

With libgsf 1.8.2:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 16384 (LWP 12992)]
0x4021e945 in open_path () from /usr/lib/libgsf-1.so.1


With libgsf 1.8.1 runs fine.


Portage 2.0.49-r4 (default-x86-1.4, gcc-3.3.1, glibc-2.3.2-r1, 2.6.0-test5-mm2)
=================================================================
System uname: 2.6.0-test5-mm2 i686 Intel(R) Celeron(R) CPU 1.70GHz
distcc 2.10 i686-pc-linux-gnu (protocols 1 and 2) (default port 3632) [disabled]
ACCEPT_KEYWORDS="x86 ~x86"
AUTOCLEAN="yes"
CFLAGS="-mfpmath=sse -fomit-frame-pointer -mmmx -msse -msse2 -march=pentium3
-mcpu=pentium4 -O2 -fno-strict-aliasing -fstack-protector -finline-functions
-falign-jumps=5 -falign-loops=5 -falign-functions=64 -pipe"
CHOST="i686-pc-linux-gnu"
COMPILER="gcc3"
CONFIG_PROTECT="/etc /var/qmail/control /usr/kde/2/share/config
/usr/kde/3/share/config /usr/X11R6/lib/X11/xkb
/usr/share/texmf/tex/generic/config/ /usr/share/texmf/tex/platex/config/
/usr/share/config"
CONFIG_PROTECT_MASK="/etc/gconf /etc/env.d"
CXXFLAGS="-mfpmath=sse -fomit-frame-pointer -mmmx -msse -msse2 -march=pentium3
-mcpu=pentium4 -O2 -fno-strict-aliasing -fstack-protector -finline-functions
-falign-jumps=5 -falign-loops=5 -falign-functions=64 -pipe"
DISTDIR="/usr/portage/distfiles"
FEATURES="ccache autoaddcvs sandbox userpriv"
GENTOO_MIRRORS="http://www.mirror.ac.uk/sites/www.ibiblio.org/gentoo
http://gentoo.oregonstate.edu
http://distro.ibiblio.org/pub/Linux/distributions/gentoo"
MAKEOPTS="-j3"
PKGDIR="/usr/portage/packages"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
PORTDIR_OVERLAY="/usr/local/portage"
SYNC="rsync://rsync.europe.gentoo.org/gentoo-portage"
USE="x86 arts avi crypt cups encode foomaticdb gif jpeg libg++ mad mikmod mmx
mpeg ncurses png quicktime spell truetype xml2 xmms xv zlib gtkhtml alsa gdbm
slang readline tetex aalib bonobo svga ggi tcltk java guile mysql X sdl gpm tcpd
pam libwww ssl perl python esd imlib oggvorbis gnome qt motif opengl mozilla cdr
apache2 dvd gtk2 ipv6 maildir sasl sse tiff usb wmf xml -oss -apm -nls -pdflib
-berkdb gtk -ldap joystick oss -kde"

[Comment 1 foser (RETIRED) 2003-09-19 07:22:49 UTC]

rule 1 of gentoo bugreporting, lower your CFLAGS

afaik nautilus shouldn't even link to libgsf, it is only used by gnumeric.

[Comment 2 André Lemos 2003-09-19 07:30:12 UTC]

I did lowered my flags to rebuild nautilus down to "-O2 -march=pentium3 -pipe" ....


AFAIK, gdb told me that nautilus does use libgsf, as you can see. And if I don't even use gnumeric, what is it doing on my system? ;)

[Comment 3 Spider (RETIRED) 2003-09-19 07:35:29 UTC]

and what flags did you build libgsf with? 

[Comment 4 foser (RETIRED) 2003-09-19 07:39:40 UTC]

'only used' was 'only used that i know of', but it still isn't used by nautilus. There are other things that may pull it in. But i found your probably dep, you use librsvg which can use it and librsvg is used by nautilus if you use svg icons. Try switching to a non-svg theme. If you do backtraces, why don't you include them ?

Still i don't see this problem and your pretty high mad-o-meter level of CFLAGS makes me suspect that in some way it is related.

lowering CFLAGS on 1 build only may not be enough

[Comment 5 André Lemos 2003-09-19 07:40:29 UTC]

To be honest, I haven't striped down the flags for libgsf for neither version. So maybe 1.8.2 has issues with this flags. I'll give it a go, and then I'll let you know what the outcome was.

[Comment 6 André Lemos 2003-09-19 07:48:30 UTC]

I've rebuild libgsf with "-O2 -march=pentium3 -pipe" and did a kill -HUP on nautilus, everything looks fine.

So I guess this particular version doesn't handle quite well this flags (I know they aren't very ortodox).

[Comment 7 Roberto Nibali 2003-09-21 02:46:31 UTC]

This is the open_path() bug I've been chasing down the last 24 hours. It has been extensively documented in bug #29208. If you're interested, please have a look:

http://bugs.gentoo.org/show_bug.cgi?id=29208

A rebuild of the defect library will fix those segfault issues in open_path(), as the reporter correctly stated previously.

[Comment 8 foser (RETIRED) 2003-09-21 03:50:58 UTC]

closing, non-reproducable

as far as the comment in #7 goes, it looks like the same problem only there is no answer to why this happens. Just fixing symptoms is not really solving the bug.