[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Simulavr-devel] initial patch for interrupt flag/enable bits
From: |
Theodore A. Roth |
Subject: |
Re: [Simulavr-devel] initial patch for interrupt flag/enable bits |
Date: |
Fri, 21 Nov 2003 16:03:15 -0800 (PST) |
On Fri, 21 Nov 2003, John Regehr wrote:
> Hi Ted-
>
> I've attached my interrupt patch. It only covers the ATmegas but I can do
> the others soon, and at that point I can hack up a small change to the
> interrupt selection logic in irq_get_pending_vector.
I'd rather not commit this until all of the devices are handled
otherwise the build will be broken.
I'll try to add some of the others tonight.
>
> Two things bother me a little. First, there is redundancy across chips;
> for example, the timer structure is the same between the 103 and 128. I
> have not factored out this redundancy anywhere because there are many
> differences as well as similarities.
This is a hard one to handle. It's been a thorn in my side for a long
time and I've never come up with a clean solution. If you have any
ideas, I'm all ears.
> Second, some of the register locations are available as named
> constants elsewhere in the source; for example, in eeprom.h we find
> EECR_ADDR. I have not exploited these because not nearly all of the
> register locations are named, and because using any of them required
> includeing a bunch of extra header files and I didn't know if you'd
> want to do that. Also, register locations are not always consistent
> across members of the AVR family.
Back when I started simlulavr, I was really only looking at the
datasheets for the 1200, 2313 and 8515 devices. I made some poor design
decisions because of that. The changes I hinted at the other day might
clean that up a bit.
For now, I'd rather stay away from using the names since the current
implementation is sub-par.
>
> Very frequently simulavr's CVS server aborts when I try to do something.
> This is frustrating. Is the a workaround?
That's the first I've heard about that. Sounds like savannah is having
load issues for anon-cvs users. Sadly, I don't really have any control
over it since I don't run the server.
About the only thing I could do is send you a tarball of the latest cvs
repo and you work from that. Let me know if you want that.
Ted Roth