Re: [RFA/ARM] Framificate the ARM port [3/3]

[Daniel Jacobowitz <drow@mvista.com>]

On Tue, Jul 01, 2003 at 07:24:38PM -0400, Andrew Cagney wrote:
> >Just in case you're wondering.
> >>
> >>This would break any architecture that didn't have an FP called "fp". 
> >>This is because, "$fp" defaults to get_frame_base() ... :-(
> >
> >
> >Argh!
> 
> :-)
> 
> >get_frame_base_address defaults to get_frame_base; I could update
> >value_of_builtin_frame_fp_reg safely, I think.  But I have no idea
> >where that would end up, so I'm going to defer to your judgement on
> >this if you've got a preference...
> 
> kfail arm-*-* gdb/497?
> 
> Having value_of_builtin_frame_fp_reg return get_frame_base_address is 
> certainly more correct.  The intent was for $fp to return the ABI's 
> virtual frame base register.
> 
> Try it.
> 
> PS: ARM $fp confusion fixed?
> http://sources.redhat.com/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gdb&pr=479

No, not fixed.  $fp will return the frame base now.  What it should do,
IMO, is return $r11.  Which isn't actually the frame pointer register
for Thumb mode, but that's the convention the assembler uses.  Which
may then re-break the test until we fix this properly...

Richard, does that sound right?  Andrew, have we got a standard
mechanism for register aliases, or would I need to create a
pseudo-register $fp that mapped onto $r11?

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer