Re: inconsistent breakpoints with Sun Workshop 6.2

[Michael Mueller]

Fabien,

You can find unofficial patches for this and a couple of other issues on
my website: http://www.mm.kay-mueller.de/gdb_for_sun_cc.html. (They are
for gdb 6.2. Haven't looked into 6.2.1 yet.) I'm using gdb for Sun cc
compiled programs myself a lot and am willing to help if you run into
more problems.

Mark,

This is still open, see thread

  "patch 1 for Sun C compiled target" on gdb-patches,
  http://sources.redhat.com/ml/gdb-patches/2004-07/msg00106.html

I was busy with other things lately but I am willing to come back to my
Sun/gdb problems (see list on above website). Also still open:

  patch #2 for Sun C compiled target programs
  patch #3 for Sun C compiled target programs

Michael


Fabien COUTANT wrote:
> Hi,
> I have to debug a scientific app written in C/C++/Fortran built with
> Sun Workshop 6.2 (Solaris 2.8 on sun4u), and I would like to use gdb
> (well insight actually) to debug the C/C++ part.
> I tried two recent gdb versions (6.2.1 and insight 6.1) and both show
> the same bug on the little sample program below:
>
> File "hello.c":
> 1    #include <stdlib.h>
> 2    #include <stdlib.h>
> 3
> 4    extern int main (int argc, char **argv)
> 5    {
> 6        printf ("Hello, world !\n") ;
> 7        return EXIT_SUCCESS ;
> 8    }
>
> In gdb session, I first enter "b main", it replies breakpoint is set
> on line 7 !  When I "r"un it executes the printf before stopping :(
> If instead I enter "b 6" gdb dies on obscure error (bus error? illegal
> instruction?  I don't remember) when I try to "r"un the program.
>
> The program was compiled and linked with "-xs -xildoff" flags as I
> have found in other articles (otherwise gdb doesn't find symbols).
> Also tried adding "-xarch=generic" or "-xarch=v8" with same result.
> Compiling the program with gcc (2.95 or 3.3) works fine, but
> unfortunately I can't use gcc :(
>
> I spent some time searching for this problem on the web and didn't
> find anything.  This must be a new bug so I report it here.  I hope
> someone is able to fix it... or tell me a workaround.