[ECOS] GDB's stack trace.

[ECOS] GDB's stack trace.

[Patrick O'Grady]

Hi, all--

Does anyone out there know what tells GDB that it's at the top of the
stack during a back-trace?  Whenever I 'bt' a stopped thread, I get
legtimate stack data followed by a bunch of junk...  somehow GDB gets told
when the stack ends--anyone know how this is done?  Thanks in advance..

-patrick
patrick@softprocess.com

Re: [ECOS] GDB's stack trace.

[Bob Koninckx]

Patrick O'Grady wrote:
> 
> Hi, all--
> 
> Does anyone out there know what tells GDB that it's at the top of the
> stack during a back-trace?  Whenever I 'bt' a stopped thread, I get
> legtimate stack data followed by a bunch of junk...  somehow GDB gets told
> when the stack ends--anyone know how this is done?  Thanks in advance..
> 
> -patrick
> patrick@softprocess.com
If I am not mistaking, GDB decides that it reached the top of the stack
as soon as it sees a "terminating stack record", that is, a stack record
that has no valid (0x0000000) backchain and return address anymore.

Are you sure that the stack is correctly initialised before entering the
new thread ?? I suppose that a terminating stack record must be present
before the call to the thread entry point. This way, the entry will be
seen as the last call on that stack.

Hope this helps,

Bob
-- 
----------------------------------------------------------------------- 
 ir. Bob Koninckx
 Katholieke Universiteit Leuven       
 Division Production Engineering,     tel.  +32 16 322535
 Machine Design and Automation        fax.  +32 16 322987
 Celestijnenlaan 300B                 bob.koninckx@mech.kuleuven.ac.be
 B-3001 Leuven Belgium
 http://www.mech.kuleuven.ac.be/pma.html
-----------------------------------------------------------------------

Re: [ECOS] GDB's stack trace.

[Patrick O'Grady]

Hi, Bob--

Your suggestion worked great: I modified the function which initialized
the stack to include a few extra 0's at the top of the stack...  now the
stack traces come out perfectly.  Thanks for your advice!

-patrick


On Fri, 21 Jan 2000, Bob Koninckx wrote:

> Patrick O'Grady wrote:
> > 
> > Hi, all--
> > 
> > Does anyone out there know what tells GDB that it's at the top of the
> > stack during a back-trace?  Whenever I 'bt' a stopped thread, I get
> > legtimate stack data followed by a bunch of junk...  somehow GDB gets told
> > when the stack ends--anyone know how this is done?  Thanks in advance..
> > 
> > -patrick
> > patrick@softprocess.com
> If I am not mistaking, GDB decides that it reached the top of the stack
> as soon as it sees a "terminating stack record", that is, a stack record
> that has no valid (0x0000000) backchain and return address anymore.
> 
> Are you sure that the stack is correctly initialised before entering the
> new thread ?? I suppose that a terminating stack record must be present
> before the call to the thread entry point. This way, the entry will be
> seen as the last call on that stack.
> 
> Hope this helps,
> 
> Bob
> -- 
> ----------------------------------------------------------------------- 
>  ir. Bob Koninckx
>  Katholieke Universiteit Leuven       
>  Division Production Engineering,     tel.  +32 16 322535
>  Machine Design and Automation        fax.  +32 16 322987
>  Celestijnenlaan 300B                 bob.koninckx@mech.kuleuven.ac.be
>  B-3001 Leuven Belgium
>  http://www.mech.kuleuven.ac.be/pma.html
> -----------------------------------------------------------------------
> 

Re: [ECOS] GDB's stack trace.

[Jonathan Larmour]

Patrick O'Grady wrote:
> 
> Does anyone out there know what tells GDB that it's at the top of the
> stack during a back-trace?  Whenever I 'bt' a stopped thread, I get
> legtimate stack data followed by a bunch of junk...  somehow GDB gets told
> when the stack ends--anyone know how this is done?  Thanks in advance..

We've had problems sometimes with this, mostly due to the compiler
optimizing the function prolog. Which target and tools version (gcc, gdb and
binutils) are you using?

Jifl