This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
RE: stabs vs. dwarf-2 clarification
- From: Jeff Baker <jbaker at qnx dot com>
- To: 'Nick Clifton' <nickc at redhat dot com>, Jeff Baker <jbaker at qnx dot com>
- Cc: "'binutils at sourceware dot org'" <binutils at sourceware dot org>
- Date: Tue, 5 Apr 2005 12:49:59 -0400
- Subject: RE: stabs vs. dwarf-2 clarification
Yes, it seems to be related to our antique compiler (2.95.3). Thanks for
looking.
-----Original Message-----
From: Nick Clifton [mailto:nickc@redhat.com]
Sent: Monday, April 04, 2005 1:44 PM
To: Jeff Baker
Cc: 'binutils@sourceware.org'
Subject: Re: stabs vs. dwarf-2 clarification
Hi Jeff,
> I have a (hopefully) quick question about the behaviour of stabs as
compared
> to dwarf-2. I have a small sample app that I build with stabs debugging
and
> link against a static lib that was built with no debugging. If I use
> addr2line to get the line number for anything from inside the static lib
> with stabs debugging I will get the line number of whatever the previous
> stabs entry is, no matter how far back it has to go. With dwarf-2 I get
> "??:0". Is this the way stabs is expected to work?
>
> I've attached a testcase that shows what I'm talking about. Modify the
> Makefile to fit your host system, make it and then ask for a line number
> from inside func1. You should get appprofiler_looper.c:126.
I tried to reproduce this but failed(*). I was however using the latest
binutils sources from the CVS repository and I know that Fred Fish has
been checking patches to fix things in this area, so perhaps that is why.
Cheers
Nick
(*) failed as in when I was compiling with "-gstabs" I still got "??:0"
as the lien associated with the address-known-to-be-inside-func1.