Subject: Re: Compiling Kernel (long!)
To: Howard Chu <hyc@locus.com>
From: Leo Weppelman <leo@wau.mis.ah.nl>
List: port-atari
Date: 02/13/1996 22:30:41
Hi Howard,

>   > > fd?a =3D> single sided floppy (80 tracks, 9 sectors/track)
>   > > fd?b =3D> double sided floppy (2*80 tracks, 9 sectors/track)
>   > > fd?c =3D> double sided, high density (2*80 tracks, 18 sectors/track)
>   > 
>   > why not fd?.360, fd?.720 fd?.1440? Yes, I know I may link them ...
> 
>   This looks even worse than the current kludge. The correct solution
>   is an auto-detecting fd-driver.
> 
> I agree. I wrote just such a driver for Minix 1.4 at least 8 years ago.
> I'm apalled to see the world still using this approach. My driver would
> also dynamically determine the number of sectors per track, thus supporting
> the oddball ST formats like 10 or 11 sector tracks on double-density,
> or 20-22 sector tracks on high density.
> 
> All that's required is to read track 0 on any floppy open(), specifying a
> large number of sectors, and checking the residue after the sector-not-found
> interrupt. The same operation will determine single or double-sided. For
> formatting purposes, you can provide an alternate minor device that skips
> the read check on open, although that's not strictly necessary. 
As you obviously know what to do about this and because the floppy format
autodetection is on my todo/whish list, you are greatly invited to send
me a patch on the floppy driver...

> It's amazing what kind of braindead kludges people will put up with. But
> then again, I guess that's why Bill Gates is a billionaire.
You won't get rich from sending in the patch, you will however relieve the
NetBSD/Atari community from one of my braindead kludges ;-))

Leo.