This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [rfc] MSR and System regs for RedBoot target
- From: Daniel Jacobowitz <drow at mvista dot com>
- To: Andrew Cagney <ac131313 at ges dot redhat dot com>
- Cc: Mark Kettenis <kettenis at chello dot nl>, gdb-patches at sources dot redhat dot com
- Date: Thu, 19 Sep 2002 19:28:07 -0400
- Subject: Re: [rfc] MSR and System regs for RedBoot target
- References: <3D6D9B48.70407@ges.redhat.com> <86hehdp9dn.fsf@elgar.kettenis.dyndns.org> <3D8A5CB9.5040106@ges.redhat.com>
On Thu, Sep 19, 2002 at 07:24:41PM -0400, Andrew Cagney wrote:
> >Andrew Cagney <ac131313@ges.redhat.com> writes:
> >
> >
> >>Hello,
> >>
> >>The attached (a patch against my sysregs branch) based mostly by code
> >>previously written by Fernando Nasser, adds MSR and system register
> >>support for an i386 RedBoot target. They each get their own group.
> >>That way:
> >> info registers msr
> >>and
> >>info registers system
> >>works (but MSR and SYSTEM registers are not displayed by ``info
> >>registers''.).
> >
> >
> >Those system registers seem like a good idea to me. I'm not so sure
> >about those MSRs.
>
> I don't know either here. I'm going through old lost changes.
>
> >>The patch (apart from demonstrating that reggroups really do work :-)
> >>identifies a number of issues:
> >>
> >>- The patch makes RedBoot the default i386 abi -- if nothing else hits,
> >>this gets to be it. Its done by brute force. This goes back to the
> >>default discussed earlier for the ``set osabi'' command. Better re-read
> >>the thread ...
> >
> >
> >Does the OS/ABI have to be named "RedBoot"? I think most of this
> >stuff could just as well be added to the generic i386 target.
>
> It depends.
>
> The MSR registers are implemented in a RedBoot specific way - it uses
> target_query() and a qMSR packet. The qMSR packet came about because
> there are potentially ~4gb of MSR registers and the remote protocol
> doesn't support sparse register numbers.
>
> There are several possible paths here:
> - leave qMSR as something RedBoot specific
> - formalize it and make it part of the protocol
> - provide a mechanism for handling sparse remote protocol register
> numbers so that [Pp] packets can be used.
I'm all for that last one... but it's tied in with all the
defining-remote-register-mapping discussions that never seem to get
implemented.
--
Daniel Jacobowitz
MontaVista Software Debian GNU/Linux Developer