This is the mail archive of the
gcc-help@gcc.gnu.org
mailing list for the GCC project.
Re: breakpoints at bogus locations
- From: Allen Hopkins <allenh at eecs dot berkeley dot edu>
- To: Eljay Love-Jensen <eljay at adobe dot com>
- Cc: gcc-help at gcc dot gnu dot org
- Date: Mon, 31 Jan 2005 14:24:09 -0800
- Subject: Re: breakpoints at bogus locations
- References: <41FE9251.3020100@eecs.berkeley.edu><6.2.0.14.2.20050131142418.02277e40@iplan-mn.corp.adobe.com>
GDB was my first suspect (as for other weird problems in the past). The
problem is there with 6.0post-0.20040223.20rh, 6.1 and 6.3.
-Allen
Eljay Love-Jensen wrote:
Hi Allen,
What version of gdb are you using?
--Eljay
At 02:17 PM 1/31/2005, you wrote:
I'm debugging a large g++ program with gdb, and setting a breakpoint
at a particular line results in a breakpoint at some altogether
different place.
I've run "readelf -wl" to look at the DWARF-2 correspondence between
line numbers and addresses. The address corresponding to the line in
question is also associated with lines in other source files. So
something looks screwy with g++.
The problem doesn't occur with g++ 3.0 or before, and it does with
every version of g++ since 3.0.1. The program includes many files &
some large libraries, and I haven't been able to generate a simple
demonstration case.
Any suggestions about how to isolate this bug are welcome. So far,
I've only been able to determine that it's in the changes made between
g++ 3.0 and 3.0.1.
The problem occurs on GNU/Linux on an i686 machine, but not on a Sun
workstation running Solaris.
This is a show stopper for months of work, and I'm pretty much at a loss.
-Allen Hopkins