This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: Add support for RedBoot SWIs to ARM Simulator
- From: Richard Earnshaw <rearnsha at arm dot com>
- To: Nick Clifton <nickc at cambridge dot redhat dot com>
- Cc: Richard dot Earnshaw at arm dot com, gdb-patches at sources dot redhat dot com
- Date: Thu, 09 May 2002 12:11:21 +0100
- Subject: Re: Add support for RedBoot SWIs to ARM Simulator
- Organization: ARM Ltd.
- Reply-to: Richard dot Earnshaw at arm dot com
> Hi Richard,
>
> > This may trample on the toes of somebody using that SWI as a real
> > trap into an OS.
>
> Well it would if they were then trying to run code compiled for that
> OS on the simulator.
>
> > We probably should have a way of configuring the simulator at
> > run-time to allow emulation of a set of SWI values. Either from a
> > command-line switch (for the arm-...-run command) or a gdb hook (for use
> > within gdb).
>
> > Something like:
> >
> > arm-elf-run -emul-swi={redboot|angel|demon|none}
>
> Hmm, well that could be done. Personally I would hold off doing it
> until there is actually a situation where it is needed. At the moment
> support all three SWI interfaces simultaneously is not a problem. So
> unless someone comes up with a real situation where it causes problems
> I would leave things as they are. (Just my 2 cents worth).
>
> Cheers
> Nick
>
Actually, we already have. Look at the mess that is involved with
testing for XScale -- saying that XScale doesn't support Demon swi's is
nonsence -- especially when we default to enabling XScale extensions for
some simulations. I ran into something like this only the other day --
the simulator was turning on XScale (despite the fact that I wasn't using
any of those extensions), and thus disabling the SWIs I wanted.
Note the comment at the top of ARMul_OSHandleSWI which really gives the
game away:
/* Intel do not want DEMON SWI support. */
if (state->is_XScale)
...
R.