[PATCH] Create reasonable backtraces via gdb automatically

[Matthias Hopf]

On Oct 20, 10 03:14:31 +1100, Daniel Stone wrote:
> > I think this is something X could benefit from (especially for getting
> > better bug reports from users). Would this reasonable to commit?
> 
> I'm torn.  On one hand, glibc's backtrace() is a sad joke, and I'd love
> to see better debug info.  On the other hand, this seems like exactly
> the kind of 'let's work around our deficient OS' crap that we've spent
> the last six years deleting.

I think being able to call an external script doesn't exactly fit into
the same category. Though it's close.

Issue is that creating a reasonable backtrace is a complex task, and I
doubt you'll find much love in suggesting that this should be done by
glibc... ;-)

> Perhaps you could write the one coredump handler to rule them all (i.e.
> have the kernel invoke the script when a segfault is hit, not out of X's
> signal handler), and get that shipped everywhere instead? :)

*That* would probably a great thing. Out of my league ATM, though, would
take more time to dig into the code than I can reasonably spare.

Matthias

-- 
Matthias Hopf <mhopf at suse.de>      __        __   __
Maxfeldstr. 5 / 90409 Nuernberg   (_   | |  (_   |__          mat at mshopf.de
Phone +49-911-74053-715           __)  |_|  __)  |__  R & D   www.mshopf.de