Re: pre-2.1, AHA1542 fixes, WD7000

Frank Breure (breure@serc.nl)
Fri, 27 Sep 1996 09:09:15 GMT


>>>>> "Linus" == Linus Torvalds <torvalds@cs.helsinki.fi> writes:

Linus> On Thu, 26 Sep 1996, Keith Owens wrote:
>>
>> aha1542.c assumes kernel == real in quite a few places, fixes follow. I
>> don't claim to have got all of them but it works for my 1542C with 2 disks
>> (Micropolis 3243, IBM DORS-32160) and a cdrom (Panasonic CR-533). I also
>> changed the 16M tests slightly to look at the end of the data, not just
>> the start.

Ok, I applied the patch and it works for me. I don't get any errors
when using WorkMan though (with a NEC 4 speed CD-ROM player).

I found the explanation about the different kinds of memory very
interesting. But there is one question left, if you have an ISA card
(like the AHA1542 I have) there is a limitation on the DMA memory
addresses, they must be below 16M. What are these addresses in the
terminology of this discussion: bus addresses ?

Linus> unsigned long signature = *(unsigned int *) 0xC0000;
Linus> vs
Linus> unsigned long signature = readl(0xC0000);

Linus> I think the second version actually is more readable, no?

Agreed.

Linus> Linus

Greetings,

Frank.