Return to BSD News archive
Newsgroups: comp.os.386bsd.questions Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!olivea!charnel!psgrain!ee.und.ac.za!ucthpx!hippo!kudu!g89r4222 From: g89r4222@kudu.ru.ac.za (Geoff Rehmet) Subject: Re: 386bsd with >16M Message-ID: <g89r4222.736070103@kudu> Sender: news@hippo.ru.ac.za (Usenet News Admin) Organization: Rhodes University, Grahamstown, South Africa References: <g89r4222.735937533@kudu> <C677BI.20Dz@austin.ibm.com> <C67F39.Eov@flatlin.ka.sub.org> Date: Thu, 29 Apr 1993 07:55:03 GMT Lines: 27 In <C67F39.Eov@flatlin.ka.sub.org> bad@flatlin.ka.sub.org (Christoph Badura) writes: >In <C677BI.20Dz@austin.ibm.com> guyd@austin.ibm.com (Guy Dawson) writes: >>The ISA bus has 24 address lines so ANY card attempting DMA can only >>do so with the first 16MB of memory. >That's only correct for busmaster DMA devices. Devices that use the >mainboard's DMA controller have the same problems for similar reasons >though. I have had a look at the DMA handling code in /sys/i386/isa/isa.c. The code for handling "DMA bounce buffers" seems never to get used normally, and looks horribly buggy. (A few strategically placed panic() statements have shown that this code isn't being used at present.) Once I can get another machine set up with more than 16M of memory - I hope that will be soon, I will start fiddling with the code in isa.c to try and get the 'bounce buffers' working. Geoff. -- ============================csgr@alpha.ru.ac.za=============================== Geoff Rehmet, Parallel Processing Group, |#define DISCLAIMER These are my Computer Science Department, | ramblings, not the Rhodes University, RSA. | University's