Previously, I wrote:
>It seems to me that the code we use for floppy pseudo-DMA at the moment is
>unnecessarily complicated. [...]
Russell, you mentioned that you thought whole-track reads might fall foul of
this. By my calculations the maximum buffer size with my code is 0x18004/4
or 24577 bytes. If you extrapolate this to an 80-track double sided disk you
get a capacity of 3840 kilobytes which is (as far as I know) way beyond
what any drive fitted to Acorn equipment is capable of. In fact the Acorn
PRMs give the maximum "actual workable track length" as 12312 bytes.
It does look like there's some support in the floppy driver for whole-cylinder
transfers, which would indeed come to grief. I don't know whether this is
actually ever used on the A5000 or the RiscPC. I suspect not on the former
which is actually the machine I'm more interested in.
(For those wondering why I'm interested in this stuff: my original goal was
actually to make the floppy FIQ code short enough that it would be safe to
disable FIQs for a couple of instructions from time to time without running
the risk of disk corruption. Apparently at the moment this isn't possible
which could cause problems with the IRQ masks under some circumstances.)
p.
unsubscribe: body of `unsubscribe linux-arm' to [EMAIL PROTECTED]
