Re: [RFC/RFA] Target vectors for native Linux targets

[Ulrich Weigand <uweigand at de dot ibm dot com>]

Daniel Jacobowitz wrote:

>I'm presenting an alternative patch, stealing bits from mine last
>December and other clever bits from Ulrich's.  His patch was much
>prettier than mine - but missed some important things that generated
>ugliness in my first patch (for instance non-FETCH_INFERIOR_REGISTERS
>targets).  I've come up with cleaner solutions than I did the first
>time round.

Thanks for looking into this again!

> On Fri, Aug 19, 2005 at 02:51:24AM +0200, Ulrich Weigand wrote:
> > One problem with the conversion is that I wouldn't want to have to
> > convert all the various Linux subtargets at the same time.  It's a
> > lot of work, and I'm unable to test most of those platforms.  Thus
> > I've thought of a way to stage the conversion:
> 
> I bit the bullet and did them all, since I had most of it lying around
> anyway.

Excellent ;-)

> This patch has only been tested on i686-pc-linux-gnu at the moment, but
> it's been mightily proofread, and I have high confidence in it.  It
> obsoletes about half your posted patch for S/390 (the watchpoint bits
> are an obvious follow-up).  Could you give this a spin on S/390 for me?

Your patch works fine with no regressions on s390-ibm-linux and
s390x-ibm-linux.  I'll be happy to do the watchpoint bits once 
your patch is committed ...

> +  /* FIXME drow/2005-09-04: The hardcoded use of register_addr should go
> +     away.  This requires disentangling the various definitions of it
> +     (particularly alpha-nat.c's).  */
> +#ifdef FETCH_INFERIOR_REGISTERS
> +  t = inf_ptrace_target ();
> +#else
> +  t = inf_ptrace_trad_target (register_addr);
> +#endif

This means FETCH_INFERIOR_REGISTERS needs to be kept around in various
nm-linux.h files for now?   I'd have hoped to get rid of at least the
s390 variant ...

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  Linux on zSeries Development
  Ulrich.Weigand@de.ibm.com