From my original post :
Daniel Edward Lovinger <email@example.com> writes:
> We are experiencing a very strange problem on a set of
> Sparcstation IPCs we acquired about three weeks ago.
> Sparcstation IPC (Sun 4/40), Rom rev. 1.6
> Two esp SCSI adapters at esp0 and esp1
> Two CDC Wren 7 drives (94601-12G) on esp0, sd0 and sd1
> For our purposes, we need to run SunOS 4.0.3c on these
> machines, but during the network boot to download, as the kernel looks
> down the SCSI ....
> esp0: Old stat=0x93<RES,XZERO,CD,IO>, New stat=0x13<XZERO,CD,IO>
> esp0: Old stat=0x91<RES,XZERO,IO>, New stat=0x11<XZERO,IO>
> appear in seemingly random order, about 6 resets per drive. The only
> thing that changes is that it is stripping the RES flag off of
> (whatever). This happens on every disk access. The disk seems to work,
> but we haven't done any extensive testing of that (yet).
> The confusing part is that we have two other IPCs similarly
> configured, running 4.0.3 with no problems. The only change is in the
> part number on the IPC unit, ending in -04 rather than -05 (the newer
> units have the esp0 problem).
It turns out that if you have Sparcstation IPCs (or SLCs -
4/20?) with CPU board numbers
501-1680-03 and greater 0MB Sun-4/20 CPU Board
501-1748-03 and greater 0MB Sun-4/20 CPU Board
501-1689-08 and greater 8MB Sun-4/40 CPU Board
501-1835-02 and greater 8MB Sun-4/40 CPU Board
you will need to apply a wee patch to your kernel and
/sys/sun4c/OBJ/esp.o for future builds. For both, the incantation
adb -w [vmunix or esp.o]
esp_phasemanage+7c?W 1000000 <-- adb command
^D (control-D to finish)
will fix the problem. A magic number tweak. The Sun folk I talked to
indicated that the salient difference is a new SCSI chip that was
installed on these CPU boards (with different timing characteristics,
- but they didn't get specific). This only affects SunOS 4.0.3c, and
the patch (actually, the README file) is SunOS 4.0.3c SCSI patch kit,
P/N 795-1657-01 (patch id 100204-01).
For what it is worth, the modified kernels seem to be perking
along alright on the older IPCs with the older model SCSI chip.
Anyone interested in getting a copy of the README, please drop
a line and I'll mail it out. I don't think the interest level
justifies posting an 8k file :-).
Dan Lovinger Computing & Comunications Carnegie Mellon U.
Internet: firstname.lastname@example.org Bitnet: dl2n+%andrew.cmu.edu@carnegie
"... and in the stillness, they heard the cry of the golden banana ..."
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:13 CDT