This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [RFA] remote-sim.c: Add support for multiple sim instances
- From: Tom Tromey <tromey at redhat dot com>
- To: Kevin Buettner <kevinb at redhat dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Thu, 08 Jul 2010 14:08:22 -0600
- Subject: Re: [RFA] remote-sim.c: Add support for multiple sim instances
- References: <20100702153017.4381070b@mesquite.lan>
>>>>> "Kevin" == Kevin Buettner <kevinb@redhat.com> writes:
Kevin> A very nice side effect of this patch is that it is now possible to
Kevin> have more than one runnable simulator instance.
Cool.
Kevin> It seems to me that, long term, the right way to do it is to create a
Kevin> new sim instance for each invocation of "target sim". That, however,
Kevin> is not what I did.
[...]
Kevin> But, back to the "right way" for a moment. It seems to me that,
Kevin> ideally, we'd want to be able to have one inferior connected to a
Kevin> remote target, another to a simulator, another to a different remote
Kevin> target, yet another to a sim using a different architecture variant,
Kevin> etc. In order for that to happen, the target vector will need to
Kevin> become a per-inferior data structure.
Yeah, this has come up a few times now. Somewhere, fairly recently,
Pedro mentioned his preferred approach to this.
Kevin> Comments? Okay to commit?
This isn't an area I am comfortable with, but I didn't want to let your
message go un-replied-to.
I think the idea of putting globals into a per-sim struct is great.
This seems unobjectionable to me.
For the rest, I don't really know, though it seems basically reasonable.
Tom