Path: utzoo!attcan!uunet!munnari!otc!metro!basser!natmlab!ditsyda!evans
From: evans@ditsyda.oz (Bruce Evans)
Newsgroups: comp.os.minix
Subject: Re: Hard drives on ST
Message-ID: <1803@ditsyda.oz>
Date: 14 Feb 89 11:01:11 GMT
References: <1630@psu-cs.UUCP>
Organization: CSIRO DIT Sydney, Australia
Lines: 24

in article <1630@psu-cs.UUCP>, kirkenda@psu-cs.UUCP (Steve Kirkendall) says:
> Basically, the problem is that the older Supra board send extra interrupts
> after the DMA transfer is complete.  This either causes the kernel to print
> "dma interrupt discarded" (a minor nuisance) or to panic when the hard disk
> driver task received a message from -1, which is hardware.

In the PC at_wini.c, there is a potential bug which may be related to this.
It doesn't matter yet because of the way interrupts are handled, but
probably will matter for PS/2's.

The bug is around line 215 of w_transfer(). After setting up DMA, it does

	receive message
	read buffer out of DMA port to memory
	get results

Two interrupts should be geneated since there are two sectors per block. The
code is in a loop to take care of this. But the interrupts are generated when
the buffer is ready, before the buffer is read and not after. So the second
interrupt may arrive before the results of the first have been read, leading
to the second receive hanging. The fix is to get the results immediately after
receiving the message.

Bruce Evans		evans@ditsyda.oz.au