bet@orion.mc.duke.edu (Bennett Todd) (04/27/89)
We ordered a Sun 900M disk drive system, and specified that we wanted one
of the new fast controllers. Long, long after the order was placed (but
long, long before it arrived) our Sun salesman mentioned that the new
controller isn't supported under SunOS 3.5, we'd have to upgrade to 4.0.
After some exchanging of insults, I gave up on Sun and started looking
around. It was mentioned on the net that the 7053 is the same as a 753 to
software, so I called up Xylogics and purchased (for essentially just a
handling fee) their device driver for the 753 under SunOS 3.5 (which they
had been selling before Sun added the 7053 to their product line, and thus
I assumed before Sun added support to SunOS). It worked. Sun ships the
7053 with the address and interrupt vectors preset, to values different
from those that Xylogics used, but you can get the values Sun sets them to
from xd(4) off a SunOS 4.0 machine, which reads, in part:
controller xdc0 at vme16d32 ? csr 0xee80 priority 2 vector xdintr 0x44
so I just used those values when I configured in the Xylogics driver.
Also, Sun omitted documenting the geometry details anywhere I could find,
but between asking questions through software and reading DIP switches I
came up with the following, for our Hitachi "DK815-10" drives:
1735 normal cylinders, plus 2 alternates, for 1737 total.
15 heads
67 sectors/track
600 bytes/physical sector
1:1 interleave
Finally, after I had everything else just about working, I noted that the
machine (a 3/280) prefered to try and autoboot from the new controller
rather than the old one; type "b" at the firmware prompt and it would hang
up trying to boot off of xd (which I don't have boot blocks for). I could
get it up using "bxy()" but that's no fun. Happily, going:
eeprom bootdev=xy(0,0,0)
eeprom default_boot=true
I was able to convince it to resume default booting off the old 451.
If Sun had included a bit more in-depth documentation (like maybe the
manufacturer's manuals like they used to ship with disks?), and had
included a version of their xd(4) driver for SunOS 3.5, I would have been
saved a lot of work. But it is satisfying to actually be able to used the
damned things, without being railroaded into converting over to SunOS 4.0,
*despite* Sun's efforts!
-Bennett