Re: [openrisc] or32-uclinux-gdb : 'Found a function without a prologue at ....'

[Marko Mlinar <markom@opencores.org>]

On Wednesday 13 November 2002 10:37, Sung-taek Lim wrote:
> From: Marko Mlinar <markom@opencores.org>
>
> > > (gdb) info stack
> > > #0  uart_send_char (ch=83 'S') at uartfunc.c:127
> > > Found a function without a prologue at 0x00002448
> > > Frame pc was at 0x0000245c
> > > #1  0x245c in uart_send_char (ch=-62 '&#51680;) at uartfunc.c:126
> > >
> > > 'or32-uclinux-gdb' seems to use some pattern searching method
> > > to find the function prologue which doesn't match the way
> > > 'or32-uclinux-gcc' generates the prologue.
> > > Has anyone tried this before?
> >
> >How did you compile your program? It should be with -g and without -Ox
> >(frame support does not work for frameless functions)
>
> Exactly with -g and without -Ox. (Though I didn't know -Ox makes functions
> frameless... then is it impossible to trace back the stack of optimized
> code like uClinux kernel?)
I think Simon used mostly or1ksim executed log, which gives you more detailed 
low level info for such debugging.

Honestly we haven't got enough high level applications to test the full gdb, 
so it is probably out of sync.

Marko

--
To unsubscribe from openrisc mailing list please visit http://www.opencores.org/mailinglists.shtml