Re: Documentation on minixfs and ext2fs

[Claus Brod <clausb@hpbbn.bbn.hp.com>]

Steve Woodford wrote:
> > If you are inside Rwabs(), your code is running in supervisor mode,
> > and nothing will ever preempt you in all versions of MiNT that I know.
> > Or are we talking about different problems?
> 
> We're talking about an extension to MiNT to pre-empt Rwabs() calls. I only
> wish I could spare time to work on it :(

Hmmm, I must have missed something. Why in the world would you want MiNT
to *pre-empt* an Rwabs() call? IMHO, the proper way to do background
transfers is that the hard disk driver sets up the transfer and
then voluntarily sleeps until the transfer is over, i.e. it yields
CPU control explicitly, but is not preempted by the kernel. At least
this is the model which I used a year or two ago when experimenting
with background transfers under MiNT (I stopped experimenting when
I saw that I needed more kernel support, i.e. GEMDOS reentrancy).

--clausb@bbn.hp.com-------------------------------------------------------
  Claus Brod, MDD, HP Boeblingen     Have you hugged your manager today?
--#include <std_disclaimer>----------http://hpbmdd28.bbn.hp.com/~clausb---
   Explicit permission required for distribution in Microsoft Network.