While trying whether Darrin's SCSI patch would work with turbo hardware
(no, it didn't), I noticed that the controller chip identifies as ESP100.

Since my hardware definitely has a NCR53C90A, I would have expected to see
ESP100A.

It turned out that the cfg2 register reads out as zero after writing to
it, even after sprinkling in some DELAY(10)s.

(In general the control registers seem to be mapped properly: DCTL has
reasonable values, and some select commands seem to fail/work as
expected.)

Is this the same with non-turbo hardware? Is it technically possible that
some of the control registers are simply not wired; or is this possibly a
problem with bus_space?


BTW, currently things crash after the first SCSI DMA transfer (request
sense to the first target found). One problem is that there is no obvious
place to find out how far the last dma transfer went (the usual registers
point to the next buffer). Trying to ignore this for the moment, I got
a kernel bus error during esp_read_register.

Since the NS scsi code doesn't seem to have many turbo-specific parts
(much in contrast to the ethernet code), most likely my general dma
routines for turbo are at fault.

Random idea: Frobbing the SCSI-floppy bit to make sure the shared dma
channel doesn't act up?

Comments/suggestions welcome.

Timm
-- 
Timm Wetzel     	<twetzel@gwdg.de>
Biomed. NMR GmbH        Tel +49 551 201-1091    FAX +49 551 201-1307
Am Fassberg 11, D-37077 Goettingen, Germany