This is the mail archive of the
gdb-prs@sources.redhat.com
mailing list for the GDB project.
Re: gdb/1250 [regression] bad backtrace when function that calls 'abort' at end
- From: Mark Kettenis <kettenis at chello dot nl>
- To: cagney at redhat dot com
- Cc: gdb-prs at sources dot redhat dot com,
- Date: 10 Aug 2003 19:58:00 -0000
- Subject: Re: gdb/1250 [regression] bad backtrace when function that calls 'abort' at end
- Reply-to: Mark Kettenis <kettenis at chello dot nl>
The following reply was made to PR backtrace/1250; it has been noted by GNATS.
From: Mark Kettenis <kettenis@chello.nl>
To: gdb-gnats@sources.redhat.com
Cc:
Subject: Re: gdb/1250 [regression] bad backtrace when function that calls 'abort' at end
Date: Sun, 10 Aug 2003 21:51:36 +0200 (CEST)
Michael,
Using GCC 3.2.3, things work even without "-g" for me. Things only
break for me if I completely strip symbols from the binary (by using
strip). That's understandable. Without symbol information it's
impossible to determine where a function starts, which means we cannot
do any prologue analysis. Since we now assume that functions are
frameless by default, backtraces for functions with a frame will fail.
We can't get this right in all circumstances, and I still think that
assuming a function is frameless is a sensible thing to do if prologue
analysis fails. However, it might be a good idea to assume that code
has a frame if we fail to do any prologue analysis at all because we
cannot find the start of a function. What do you think about that?
Mark