FS#6367 - x86_64 idesk segfaults

Attached to Project: Arch Linux
Opened by Mikko Seppälä (Neverth) - Wednesday, 07 February 2007, 13:46 GMT
Last edited by Andreas Radke (AndyRTR) - Thursday, 26 April 2007, 04:39 GMT
Task Type Bug Report
Category Packages: Extra
Status Closed
Assigned To Andreas Radke (AndyRTR)
Architecture not specified
Severity Medium
Priority Normal
Reported Version 0.7.2 Gimmick
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

idesk segfaults, sometimes blinks icons for few milliseconds and crashes, doesn't report segementation fault on terminal when started from terminal, but segfault can be seen on strace output.
Also if something; eg terminal is placed so it is above all icons, it doesn't segfault, when terminal is drawn off, segfault follows.
strace and the behaviur suggest that the problem is within imlib2.
Other imlib (1 or 2) apps work as usual.
Also "borrowing" idesk binaries from debian amd64 results in samekinda behaviour. Cant figure out whats wrong with it.
This task depends upon

Closed by  Andreas Radke (AndyRTR)
Thursday, 26 April 2007, 04:39 GMT
Reason for closing:  Fixed
Additional comments about closing:  please ask to reopen if the problem resists
Comment by Fredrik (vEX) - Tuesday, 24 April 2007, 12:31 GMT
This is most likely not an idesk problem, but an imlib2 problem:debian-bugs-dist@lists.debian.org/msg265634.html"> http://www.mail-archive.com/debian-bugs-dist@lists.debian.org/msg265634.html.

Trying to run pypanel on Arch64 also leads to a segfault, using gdb I found it's the same problem as with idesk (according to debian mailing list):
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 47160048180000 (LWP 8168)]
0x00002ae44e8c0cbf in __imlib_amd64_blend_rgba_to_rgb_cmod ()
from /usr/lib/libImlib2.so.1

This can easily be fixed by rebuilding imlib2 with --disable-amd64 which turns off the faulty assembler code. I copied the PKGBUILD from abs and changed it (as well as moved to 1.3.0 from 1.2.2), I don't know if there are any problems with --enable-mmx since I don't have any i686 machine to try it on.


# $Id: PKGBUILD,v 1.8 2006/06/30 12:22:18 tpowa Exp $
# Maintainer: arjan <arjan@archlinux.org>
# Contributor: Tom Newsom <Jeepster@gmx.co.uk>
pkgname=imlib2
pkgver=1.3.0
pkgrel=1
pkgdesc="Imlib2 is the successor to Imlib. It is NOT a newer version -- Imlib 2 can be installed alongside Imlib 1.x"
arch=(i686 x86_64)
url="http://sourceforge.net/projects/enlightenment"
depends=('libtiff' 'libungif' 'bzip2' 'freetype2' 'libxext' 'libpng' 'libid3tag')
source=(http://download.sourceforge.net/sourceforge/enlightenment/$pkgname-$pkgver.tar.gz)
md5sums=('00b724fc6d2dcfa3045bb6a554bb2c8a')

build() {
cd $startdir/src/$pkgname-$pkgver

# Enable x86 MMX optimizations for i686 (32-bit assembler code not compatible with x86_64)
[ "${CARCH}" == "i686" ] && EXTRAOPTS="--enable-mmx"
# Disable AMD64 optimizations due to bug in assembler code
[ "${CARCH}" == "x86_64" ] && EXTRAOPTS="--disable-amd64"

./configure --prefix=/usr --sysconfdir=/etc/imlib2 ${EXTRAOPTS}
make || return 1
make DESTDIR=$startdir/pkg install
find $startdir/pkg -name '*.la' -exec rm {} \;
}

Loading...