Re: [RFA] MIPS: MIPS_LAST_FP_ARG_REGNUM, MIPS_LAST_ARG_REGNUM changes

[Kevin Buettner <kevinb at redhat dot com>]

On May 21,  2:06pm, Andrew Cagney wrote:

> > On May 21, 12:17pm, Andrew Cagney wrote:
> > 
> > 
> >> > This is patch 2 of many more to come.  It depends upon
> >> > http://sources.redhat.com/ml/gdb-patches/2003-05/msg00268.html.
> >> > 
> >> > Okay?
> > 
> >> 
> >> Per my earlier comment, I don't think this one is right.  I don't think 
> >> things like LAST_ARG_REGNUM belong in that cooked/raw reg structure. 
> >> Instead, they should exist out side it.
> >> 
> >> If the code really wants to differentiate between the raw and cooked 
> >> register number, why not use the more explicit:
> >> 
> >> 	rawnum->gp0_regnum + tdep->last_arg_regnum
> > 
> > 
> > So, last_arg_regnum represents a count of the number of argument
> > registers?
> 
> It's an offset from register 0.

That doesn't help either for the same reasons that I gave earlier.  (I
left it quoted below.)

> However, I think if the first patch is 
> sorted out, this will fallout.

I don't recall seeing a response to my first patch.  I'll check the
archives...

> Andrew

Kevin

> > That's fine (at the moment anyway) for the GPRs, but it doesn't work
> > for the FPRs.  For o32, I have things arranged so that there are a
> > total of 16 cooked FPRs and 32 raw FPRs.  Therefore, argument register
> > counts will be different between cooked vs raw.  IMO, it really does
> > make sense to put these values into the cooked/raw structure.  I'm
> > having difficulty understanding why you're objecting to this layout.
> > 
> > Kevin
> > 
> 
>-- End of excerpt from Andrew Cagney