Atari Mega Archive 1/Atari Mega Archive - Volume 1.iso/lists/mint/l_0399/339

Internet Message Format | 1994-08-27 | 2KB

From: Marinos Yannikos <[e-mail]>
Subject: Re: Compiling 1.07 with GCC
Date: Mon, 28 Jun 93 13:02:52 MET DST
In-Reply-To: <[e-mail]>; from "Daniel Eriksson" at Jun 25, 93 1:11 pm

> Also, what on earth has happened to MiNT between v1.04 and v1.06?
> Neither Ydisk nor the other (French) .XFS-RAMdisk works any more. :-(
> They (or MiNT I suppose) both return ".: unknown device" whenever
> I try to access them. Any hints?

The problem is, that MiNT assumes that all drives with device numbers < 32
also have a BIOS part (perhaps I'm wrong, but that's what it looks like)
and thus calls Mediach() to detect media-changes forced by programs
that bend vectors for that purpose. AHDI doesn't like such drives,
obviously. According to Eric (filesys.c, line 375) the solution is
not to change the default "PSEUDODRVS" settings, but that only
helps with MiNT's built-in pseudo-drives (u:\PROC, ...). There are
2 solutions for a ramdisk I could think of:

1. Install an own routine (in the BIOS-vectors for disk access) to
   return 0 for a media change

2. recompile MiNT with the ramdisk built-in (similar to the way
   the current psuedo-drives are handled, with drive-number >= 32).

Obviously, 2 would be more comfortable, but not very practical, so
I guess one has to go for 1. I hope Eric will do something about it,
e.g. have the file-systems return information whether they have
a BIOS part or not. Of course, I could be wrong, and there *is* a way
to tell MiNT not to assume this, but I couldn't find anything.

-MY
 _________________________________________________________________________
 Marinos Yannikos       |@-mail: Blumberggasse 14/1,  |Tel.[phone]
 [e-mail]|        A-1160 Wien, Austria |(9-11pm MET)
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~