Hello,
On Sun, 19 Nov 2006 13:53:42 +0100, Josip Rodin
<[EMAIL PROTECTED]> wrote:
On Thu, Oct 12, 2006 at 02:35:30AM +0200, Jiří Paleček wrote:
Package: wmmount
Version: 1.0beta2-6.1
Severity: important
wmmount crashes every time I start it. The backtrace is just rubbish,
but
valgrind shows at least something, so I attached its output.
(Sorry about not seeing this earlier, my overzealous spam filters caught
the
binary attachment and stuffed it away to a place where I rarely look.)
It's going to take me a while to download valgrind and become acquainted
with it, so if you'd be so kind to summarize a bit so that I know what
I'm
aiming at? Also, since you know your way around debuggers :) you might
want
to fetch the source, build the binary and run the one with debugging
symbols, presuming that is the reason why the backtrace is useless.
Well, first of all, this is not the reason. Backtraces without debug info
look like the one in valgrind output. Rubbish I referred to was something
like
xxxxxx ??
000000 ??
etc. This usually means that phe program has corrupted its stack (eg. base
pointer, so gdb is unable to trace the stack) or somehow managed to jump
out of the code (or lose its instruction pointer).
To the crash. I have discovered it was caused by nonexistent fonts,
eg. those xserver claims to have, but for some reasons fails to
load (You can simulate that by renaming the file with the font wmmount
uses, but leaving fonts.dir intact). So it should be some checking
of return value somewhere omitted.
Regards
Jiri Palecek
