Re: Multitasking during disc access (was: Re: MiNT scheduling)

[Jo Even Skarstein <joska@nuts.edu>]

On Fri, 14 Nov 1997, Sven Karlsson wrote:

> > you hook into the kernel via the XFS/XDD-interface doesn't return
> > until they've done their stuff (e.g. read/written some stuff), and
> > AFAIK they do it in supervisor-mode. And all (x)bios-functions runs in
> > supervisor-mode for sure, so even changing MiNT won't work without
> > rewriting the (x)bios... 
> 
> It is quite possible to write nonblocking scsi/ide/floppy etc
> routines that for instance minixfs can use. There is nothing in
> the current driver/filesystem handling that prevent that.

Then I've misunderstood the docs. My apologies.

> However to be efficient a couple of things need to be implemented:
> 
> 1) New lowlevel routs (ie hard disc drivers etc) that 
>    replace bios/xbios routs completely

Yep, exactly. This is the main problem, as it has to cover a lot of
different hardware (ACSI, IDE, Falcon SCSI, TT SCSI, clones...).

> 2) New TOS/FAT fs since a lot of ppl still use these
>    systems.

Prefered, but not needed if the old FS will work with the new [x]bios
routs.

> be carefully tested. I think it would take about 6 months
> to get them all right. In which time there hopefully will
> be about three Fenix versions.

So when will we see a Fenix-version that's capable of running the
standard MiNTOS-stuff and N.AES?


/*
** Jo Even Skarstein    http://www.stud.ntnu.no/~josk/
**
**    beer - maria mckee - atari falcon - babylon 5
*/