Re: [MiNT] FreeMiNT1.16.x/XaAES on 68000-machine

[Johan Klockars <johan@klockars.net>]

> A small update on (b): if I start the latest fVDI (68000-version) 
> before MiNT, XaAES loads all the way and I get to the Tera-desktop. 
> With keyboard shortcuts I can open/close folders etc, but as soon as I 
> move the mouse I get "pid0 (): assert


It's interesting that it works better with fVDI than with NVDIv4 (and 
was that the original VDI you tested with also?), though I have no idea 
why that would be so. We could possibly find out, however, if you turn 
on the fVDI debugging to see what's happening.

> (`curproc->stack_magic == STACK_MAGIC') failed at line 505 of 
> ../sig_mach.c." again and the system freezes ...


Well, this is probably my fault, somehow. ;-)
Was this with or without "oldmouse"? (Better to discuss this in private 
email, I think.)

In any case, at this point any problems that show up when using the new 
fVDI bitplane driver are very likely to be caused by that rather than 
some other software. For example, I'm pretty sure the strange mouse 
behaviour you mentioned in a private mail is caused by the mouse cursor 
drawing taking too long (it's done using a series of calls to the 
internal equivalents of the vro_cpyfm and vrt_cpyfm blit routines), 
causing other interrupts to be delayed or even missed. The stack 
overflow (at least that's what it looks like) you mention above could 
possibly be caused by the same thing, perhaps in combination with too 
much interrupt stack is being used by the blits.

I'll write a decently fast mouse cursor drawing routine for the bitplane 
driver shortly, which should take care of the "unguided projectile" 
mouse cursor and hopefully the stack overflow. The current routine was 
never intended to stay in there, but it was a quick way to get the 
driver working (I don't think I've spent two days on that driver in total).

--
| Why are these | johan@klockars.net
|  .signatures  |
| so hard to do | http://www.klockars.net
|     well?     | (fVDI, MGIFv5, QLem)