summaryrefslogtreecommitdiff
path: root/sys/arch/kbus/README
diff options
context:
space:
mode:
authorgingold <gingold@cvs.openbsd.org>1997-10-14 07:25:35 +0000
committergingold <gingold@cvs.openbsd.org>1997-10-14 07:25:35 +0000
commitf6491d400ca651a8a1493d72c6a74c622aa231b1 (patch)
tree39f29255154f230f637c12e9214f12a9e64bd9d4 /sys/arch/kbus/README
parent7a9e3739a66bd0fadfdc611c72e879fcc6f9ef01 (diff)
Gingold's port for kbus Series5 machine. Not fully finished and not very stable
Diffstat (limited to 'sys/arch/kbus/README')
-rw-r--r--sys/arch/kbus/README49
1 files changed, 49 insertions, 0 deletions
diff --git a/sys/arch/kbus/README b/sys/arch/kbus/README
new file mode 100644
index 00000000000..db88456c863
--- /dev/null
+++ b/sys/arch/kbus/README
@@ -0,0 +1,49 @@
+README file for Series5.
+************************
+
+Written by Tristan Gingold 1997 (gingold@email.enst.fr)
+
+This file contains comments about the work I did in 1997.
+These comments are written from remindings, so they are not accurate.
+
+Series5 was a computer shipped by Solbourne from October 1989 on.
+However, Solbourne hardware activities are now ended.
+
+The Series5 is the second model from Solbourne, after Series4.
+It was powered by a 33 Mhz Cypress Sparc and a Weitek Abacus 3171 fpu.
+
+The model I have (Series5 model 800 with 2 SMD disks) has 2 cpus.
+In fact, Solbourne Computer was the first to offer a multiprocessor Sparc
+machine, which was software compatible with SunOS.
+
+However, it is not hardware compatible.
+
+The SMP servers are based on CPU, memory and I/O cards that connect to a
+common backplane bus, the KBUS. It is a proprietary 64-bit bus.
+The I/O cards also provides a bridge to a VME bus, on which a Xylogics card
+is connected.
+
+Device drivers:
+* zs: (ttya/ttyb) ttya is the console. At this time, there is a software bug
+ which sometimes hangs the console (but not the system).
+* le: (ethernet). Seems to be fully ok, used for loading the kernel and
+ as the main diskless interface.
+* xd: (Sylogucs 753 SMD disk controller). Seems ok. The driver was based
+ on sparc/dev/xd.c, but modified (the genuine bad144 code was too buggy
+ for me).
+* si: (WD33C93A SCSI): At least inquire is working, but no more.
+* kbd, bwtwo, ms: there is no drivers, since I have no screen, no mouse and
+ no keyboard.
+
+Note that only one processor can work. (Snif !)
+
+At this time, I can boot the system and get the login prompt on the console,
+and via telnet.
+The kernel sometimes crash, so there is still some work to do.
+
+************
+The Series5 kernel was cross-build with a gcc configured for SunOs on a
+i386-linux host. (!!!)
+The binary should be fully compatible with the Sparc one. So, there is no
+binary distribution.
+Kernel files come from sparc, pmax (for pmap.c), mvme68k (sbic*.c), pc532...