Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cornell!uw-beaver!ssc-vax!bcsaic!paula
From: paula@bcsaic.UUCP (Paul Allen)
Newsgroups: comp.os.minix
Subject: 1.5.8 initial reaction
Message-ID: <23701@bcsaic.UUCP>
Date: 27 Apr 90 05:46:38 GMT
Organization: Boeing Computer Services ATC, Seattle
Lines: 41

I just finished building a 1.5.8 boot disk and have two quick
comments.

The subscript expressions on lines 259 and 260 in tools/menu.c
(where it's setting up the default values for the ram image and
the root device) are bogus.  Someone else also mentioned this.
Since it's late and I need to get to bed, I didn't try to figure
out a fix.  Menu.c compiles with warnings about expression overflows,
and the resulting kernel works if you manually specify the root
and ram image values at boot time.

When I booted my initial 1.5.8 kernel, it complained about
"controller not ready" and hung when I hit the "=" key.  Thorsten
Lockert (did I spell your name right?) posted a fix for this
relative to 1.5.6.  His fix worked for me.  I used to ask the boot
menu to fsck a small partition, just so it would initialize the 
controller so it would talk to the hard disk.  Now, with no fsck
options in the menu, this problem is fatal.  Here's the patch
I applied to at_wini.c:

397c397
< 	 (in_byte(WIN_STATUS) & (WIN_BUSY | WIN_OUTREADY)) != WIN_OUTREADY)
---
> 	 (in_byte(WIN_STATUS) & WIN_BUSY) == WIN_BUSY)

Someone who hasn't had this problem should probably apply this patch
and see if their hard disk still works.  It seems to be critical for
at least some controllers.  (Note: I've simply applied Thorsten's
patch blindly.  I'm too sleepy to try to verify that this is the
best way to fix my problem.  :-) )

I'll compile the commands over the weekend and post more if I find
anything interesting.

G'nite.
Paul Allen

-- 
------------------------------------------------------------------------
Paul L. Allen                       | pallen@atc.boeing.com
Boeing Advanced Technology Center   | ...!uw-beaver!bcsaic!pallen