This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: backtrace() on ix86?
- From: Jakub Jelinek <jakub at redhat dot com>
- To: Thorsten Kukuk <kukuk at suse dot de>
- Cc: libc-alpha at sources dot redhat dot com
- Date: Mon, 19 Jul 2004 11:39:53 +0200
- Subject: Re: backtrace() on ix86?
- References: <20040719114017.GA14038@suse.de>
- Reply-to: Jakub Jelinek <jakub at redhat dot com>
On Mon, Jul 19, 2004 at 01:40:17PM +0200, Thorsten Kukuk wrote:
>
> Hi,
>
> It seems, that the backtrace() function on ix86 always misses the
> last function. A sample test case is included. bar() is never printed
> on ix86:
backtrace() on x86 ATM uses the chain of saved %ebp values
instead of dwarf2 unwind info (-fasynchronous-unwind-tables is not the
default on x86, unlike e.g. x86-64).
As such it has no knowledge about signal frames etc.
See http://sources.redhat.com/ml/libc-hacker/2003-12/msg00054.html
for i386 implementation of backtrace(3) using both .eh_frame
(as long as it is available) and with fallback to %ebp chains.
Your testcase would need to be compiled with -fasynchronous-unwind-tables
to work though.
Jakub