NetNews Usenet Archive 1992 #16/NN_1992_16.iso/spool/comp/os/minix/3774

Internet Message Format | 1992-07-26 | 1.6 KB

Path: sparky!uunet!mcsun!sunic!isgate!krafla!adamd
From: [e-mail] (Adam David)
Newsgroups: comp.os.minix
Subject: Re: Help!!! Minix ignores 2nd HD (ST)
Keywords: minix ST
Message-ID: <[e-mail]>
Date: 26 Jul 92 01:54:36 GMT
References: <[e-mail]>
Sender: [e-mail]
Lines: 24

I wrote:
>but I keep getting "DMA interrupt discarded" warnings whenever the 2nd disk
>is accessed. This is ST Minix 1.5.10.4 with OLDSUPRA defined so that the
>driver interacts correctly with the early type Supra hostadaptor.

It was the timing loop governed by PNK_DELAY that had to be lengthened because
the 2nd disk is a fair bit slower than the first. PNK is presumably an acronym
for Parameter Not Known, it was pure hocus pocus finding a useful value for it,
and it was quite critical timing. Anything less than 140 or so discarded the
interrupt, values between 140 and 160 gave a "kernel panic: Winchester task got
message from -1". 160 worked alright. The first disk (which needed the OLDSUPRA
mod in order to boot without a kernel panic) really didn't care about the timing
delay and worked perfectly with all values that I tried, from 0 to 200.

I wonder, with the wide variation of CPU clock speeds available these days, is
there not a better way of handling timing routines than using software loops
written in C? Surely there are some spare timers on the MFP and sound chips
that can be used to advantage. A lot of time is wasted executing busy loops
while disk transfers are taking place, and this is a DMA based system (!!!).
During the actual transfers, everything else just hangs. At least it is
working now :-)

--
Adam David. ([e-mail])