Re: [B2-devel] Hello, and a few questions

[Christian B. <cb...@st...>]

Hi!

On Tue, Nov 28, 2000 at 11:41:15AM +1100, ni...@in... wrote:
> * Since this isn't an X windows port, I really cannot use
>   the Unix directory for video and main. So, is a directory
>   'MacOSX' at the same level as Amiga, BeOS and Unix OK?

This is better, especially if you intend to switch from autoconf/Makefiles
to an IDE.

> [NMI/Reset]
>   The sensible way to do this would seem to be to define two new
>   interrupt flags in main.h, e.g.:
> 	INTFLAG_NMI = 32	// Non maskable interrupt
> 	INTFLAG_RESET = 64	// Hard reset
>   and handle those in the emulator.

Not really. The INTFLAGS are for decoding level 1 interrupts. There's a
TriggerNMI() in cpu_emulation.h which is currently unimplemented for the
UAE CPU emulation (it works under AmigaOS, though). I think the BeOS port
once had a "reset" capability, but it seems to have got lost somehow.
Adding a Reset680x0() function might be in order.

>   Now, I can see that TriggerInterrupt() sets a bit to tell
>   something else to handle an interrupt, but I cannot see what
>   actually processes the InterruptFlags variable.

EmulOp()/M68K_EMUL_OP_IRQ tests them.

> (e.g. why does Unix need an XPRAM watchdog?

The XPRAM contents are normally only saved when B2 quits. To avoid changes
to the XPRAM from getting lost when the emulator crashes, a background
thread looks for XPRAM changes every minute and saves the XPRAM if necessary.

>       why does Unix use this:
> #ifdef NANOSLEEP
> 		struct timespec req = {1, 0};
> 		nanosleep(&req, NULL);
> #else
> 		usleep(1000000);
> #endif
>       to sleep for a minute when sleep(60); would suffice?)

I think I just copied the sleep code from some other function. I also like
nanosleep() better than sleep() because it doesn't mess with signals.

Bye,
Christian

-- 
  / Coding on PowerPC and proud of it
\/ http://www.uni-mainz.de/~bauec002/