On Fri, May 18, 2001 at 10:00:32PM +0200, Ignatios Souvatzis wrote:

> notmuchprogress report: compiled amiga/dev/siop.c with -DDEBUG and run.
> 
> Apparently, the drive in question either likes to disconnect w/o save data
> pointer first, and the code saves the data pointer (updates DMA chain) anyway.
> 
> I'm compiling a kernel with this quick and dirty changed right now.

mostly works. however, mode sense doesn't work. 

Old results:

scsibus0: waiting 2 seconds for devices to settle...
drsc0: target 0 now synchronous, period=100ns, offset=8
sd0 at scsibus0 target 0 lun 0: <QUANTUM, TRB850S, 0404> SCSI2 0/direct fixed
sd0: 810 MB, 3653 cyl, 4 head, 113 sec, 512 bytes/sect x 1660299 sectors
drsc0: target 2 now synchronous, period=100ns, offset=8
sd1 at scsibus0 target 2 lun 0: <IBM, DDRS-34560, S97B> SCSI2 0/direct fixed
sd1: 4357 MB, 8387 cyl, 5 head, 212 sec, 512 bytes/sect x 8925000 sectors
drsc0: target 3 now synchronous, period=100ns, offset=8

New results:

scsibus0: waiting 2 seconds for devices to settle...
drsc0: target 0 now synchronous, period=100ns, offset=8
sd0 at scsibus0 target 0 lun 0: <QUANTUM, TRB850S, 0404> SCSI2 0/direct fixed
sd0: could not mode sense (4/5); using fictitious geometry
sd0: 810 MB, 810 cyl, 64 head, 32 sec, 512 bytes/sect x 1660299 sectors
drsc0: target 2 now synchronous, period=100ns, offset=8
sd1 at scsibus0 target 2 lun 0: <IBM, DDRS-34560, S97B> SCSI2 0/direct fixed
sd1: could not mode sense (4/5); using fictitious geometry
sd1: 4357 MB, 4357 cyl, 64 head, 32 sec, 512 bytes/sect x 8925000 sectors


Hm, when I browsed the may 14 diffs, I noticed that you're using nested
structs and unions for on-wire data in the new code. Maybe this stuff needs
to be __attribute__((packed)) for it to work with m68k compilers?

Regards,
	-is