I just finished rebuilding the world on a 20MB Sun 3/60 with a 662MB SCSI-1
Micropolis disk under the SCSI_TEST kernel. I also did some "dd" tests of
the raw device while the compiles were going:

dd if=/dev/rsd0b of=/dev/null bs=50k count=200
200+0 records in
200+0 records out
10240000 bytes transferred in 18 secs (568888 bytes/sec)
0.0u 2.8s 0:18.34 15.7% 0+0k 0+2io 0pf+0w
dd if=/dev/rsd0d of=/dev/null bs=50k count=200
200+0 records in
200+0 records out
10240000 bytes transferred in 17 secs (602352 bytes/sec)
0.0u 2.7s 0:17.33 15.6% 0+0k 1+1io 0pf+0w
dd if=/dev/rsd0e of=/dev/null bs=50k count=200
200+0 records in
200+0 records out
10240000 bytes transferred in 19 secs (538947 bytes/sec)
0.0u 2.9s 0:18.58 15.8% 0+0k 0+2io 0pf+0w
dd if=/dev/rsd0g of=/dev/null bs=50k count=200
200+0 records in
200+0 records out
10240000 bytes transferred in 17 secs (602352 bytes/sec)
0.0u 2.7s 0:17.20 16.1% 0+0k 0+0io 0pf+0w

As reported earlier, attempts to transfer more than 63K resulted in kernel
printf's, so I did 50K transfers. I did have some trouble with this
configuration initially, but that turned out to be a loose SCSI connector
at the disk. Once it was tightened up, no more trouble.

This is hardly a full stress test, but it's a good start. Congratulations
on getting SCSI DMA going.

One other item: the make died twice trying to rebuild libc, just as before,
and I had sup'd and reinstalled "make" before starting this whole process,
so the bug is still there. I also have some kind of libkvm mismatch because
vmstat will not run, and ps reports funny results in places. Hopefully this
will all clear up after I do "make install" this morning.

Erik Fair