Annotation of wikisrc/ports/luna68k/luna68k_info.mdwn, revision 1.2
1.1 ryoon 1: [[!meta title="NetBSD/luna68k: Information"]]
2:
3: *This page is under construction and more information will be added
4: to make it helpful for hardware owners suffering from missing documents.*
5:
6: # Hardware and Operation<a name="hardware"></a>
7: [[!table data="""
8: Model |Processor |RAM |RAM configuration
9: LUNA |20MHz 68030 + 20MHz 68882 FPU |16MB |4/8MB + two 4MB modules
10: LUNA-88K |25MHz 88100 + pairs of 88204 CMMU |64MB |16 30-pin SIPP slots
11: LUNA-II |25MHz 68040 |64MB |16 30-pin SIMM slots
12: """]]
13:
14: LUNA-II has a room for piggy back 2nd 68040 processor,
15: and apparently designed as 2 processor SMP box.
16: LUNA-88K Plus product announcement was done in mid 1993.
17: It's not certain volume production was made.
18: The company released later rebagged DG AViiONs under LUNA brandname
19: which run m88k DG-UX.
20:
21: ## Photos; front and back
22:
23: [ ... professional looks photos here ... ]
24:
25: ## Front panel DIP switch #1
26:
27: [[!template id=programlisting text="""
28: [] [] [] [] []
29: [] [] []
30: 1 2 3 4 5 6 7 8
31: """]]
32:
33: [[!table data="""
34: switch # |description
35: 1 |down - boot system in ROM monitor mode<br />up - load and start UNIX (always goes to multiuser mode)
36: 2 |down - serial console on ttya<br/ >up - console on bitmap display
37: 3 |down - force to have monochrome display<br />up - color display
38: 4 |down - verification on every harddisk write operation<br />up - no write verification
39: 5 |down - operating system is UniOS-B (4.3BSD/a.out OMAGIC)<br />up - operating system is UniOS-U (SystemV/COFF)
40: 6 |down - force monochrome display<br />up - color display</br />*uncertain about functional difference from sw3.*
41: 7 |down - boot from network<br />up - boot from local devices
42: 8 |down - start diagnostics<br />up - normal boot
43: """]]
44:
45: DIP switch #2 is not used for any purpose.
46:
47: ## ROM monitor operation
48:
49: ROM monitor commands are not like those found in popular UNIX boxes.
50:
51: <dl>
52: <dt>k</dt>
53: <dd>
54: Display and change the boot device and filename. Boot device can be harddisk
55: (dk), netboot (et), cassette tape (sd), or floppy disk (fl).
56: </dd>
57: <dt>g</dt>
58: <dd>
59: Load the boot program into memory. It can take a different filename
60: as optional argument. Note ROM monitor recognizes only a.out format binaries.
61: </dd>
62: <dt>x</dt>
63: <dd>
64: Execute the loaded program. It accepts optional arguments. NetBSD/luna68k
65: takes any letter of s (boot in single user mode), d (start DDB session)
66: or a (ask root device name).
67: </dd>
68: </dl>
69:
70: [ ... more info, differences between models ... ]
71:
72: Here is an example of netbooting:
73:
74: [[!template id=programlisting text="""
75: >k
76: ctlr: dk et
77: host: omron [enter]
78: sver: servername [enter]
79: fnam: server:/vmunix lala:netbsd.aout
80: >g
81: text(1585988)+data(0)+bss(72780)
82: >x
83: [ Kernel symbol table missing! ]
84: Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
85: 2006, 2007, 2008, 2009, 2010, 2011
86: The NetBSD Foundation, Inc. All rights reserved.
87: Copyright (c) 1982, 1986, 1989, 1991, 1993
88: The Regents of the University of California. All rights reserved.
89: ...
90: """]]
91:
92: ## How to determine the station's Ethernet address
93:
94: For the ROM monitor prompt '>', type the following hexadecimal address;
95:
96: [[!template id=programlisting text="""
97: >4101ffe0[enter]
98: 4101FFE0: 30 30 30 30 30 41 30 32 *00000A02*
99: >[enter]
100: 4101FFE8: 34 33 30 46 00 00 00 00 *430F....*
101: """]]
102:
103: in this example, the station address is 00:00:0A:02:43:0F.
104:
105: Some older models do not store the station address in CPU ROM,
106: but in NVRAM storage (found in V4.02 ROM monitor dated Oct 12 1988).
107: See also next section.
108:
109: ## Dead timekeeper NVRAM syndrome
110:
111: The LUNA are equipped with the infamous (non)volatile timekeeping
112: RAM Mostek 48T02. Like for Sun3s and SPARCstations,
113: the backup battery wears out and eventually loses its contents.
114: Whenever the LUNA ROM monitor detects the situation,
115: it initializes the NVRAM with factory default values.
116: A Dead NVRAM is cumbersome for daily operations because there is no way
117: to automatically boot NetBSD/luna68k at powerup;
118: you have to boot the machine in ROM monitor mode and enter
119: the boot commands by hand, specifying explicit boot configuration every time.
120:
121: [[!template id=programlisting text="""
122: Diagnostic Start....nvram initialize.
123: No problems in hardware
124: OMRON WS Monitor Ver4.22 (Thu Jul 27 11:45:42 1989) - 0x01000000 bytes of memory
125: "BSD Monitor-mode"
126: >
127: """]]
128:
129: If you see the ROM monitor message shown above on powerup,
130: the NVRAM battery is gone and the contents have been initialized
131: with factory default values. To learn about the dead NVRAM syndrome,
132: please refer to [Frequently Asked Questions about Sun NVRAM/hostid](http://www.squirrel.com/squirrel/sun-nvram-hostid.faq.html).
133:
134: Older models have Ethernet station address in the ENADDR environment
135: variable. If the NVRAM wears out, the ROM monitor resets the value
136: to 0:0:0:0:0:0, and the ROM monitor will refuse to netboot
137: before until a valid address is set.
138:
139: ## Dead NVRAM replacement
140: The author succeeded in installing new DS1642-70 NVRAM.
141: The PROM detected it and initialized ("kick-start") the timekeeping circuit
142: automagically. I purchased it from a Dallas Semiconductor credit card sale
143: representative. The price was $16.89 on Jan 4 2000.
144:
145: ## Disk drive replacement
146:
147: The LUNA is equipped with one 3.5" 'half height' hard disk drive made
148: by either Conner, Fujitsu or Hitachi. They are slow and small disks of
149: at best 172MB or 270MB capacity.
150: It's a good idea to replace these low capacity drives with high capacity
151: faster drives.
152:
153: ## Removing the front bezel
154:
155: Remove the top cover first. It has one screw on the back.
156: Remove the metal sheet on top, which also has one screw.
157: Then, remove the side covers; look for one screw on the back of each one.
158: The front bezel might be secured with screws on both sides.
159: It has three leads on the top. Unlock them gently from metal notches,
160: release the top first, then remove the bottom end.
161:
162: ## Removing the tape/disk unit
163:
164: The tape and disk unit forms a block secured by a metal lead on the top.
165: Loosen the lead and slide the block to front end gently.
166: When you got a little space, release the power and SCSI cables on the back.
167: Pay attention to avoid finger injuries due to the low quality metal work.
168:
169: ## Replacing the disk drive
170:
171: The ROM monitor requires the disk drive to have SCSI ID #6.
172: The disk drive may be at the end of SCSI cable or not.
173: The LUNA requires at least one SCSI termination. Without a SCSI termination,
174: the system won't be able to run.
175:
176: ## ttya connector replacement
177:
178: The LUNA uses obscure serial connectors inherited from the old deskside
179: VME machine design. It's close to impossible to obtain genuine cables,
180: and the connector parts are very hard to find in market.
181: See [Tadashi Okamura's post](http://mail-index.netbsd.org/port-luna68k/2002/03/21/0000.html) for details.
182:
183: ## SCSI connector replacement
184:
185: The LUNA is equipped with an external SCSI connector.
186: It's a micro DB50 connector, looks like the so called 'SCSI-2 connector'
187: commonly found in any modern UNIX boxes, but it has a different gender!
188: You can replace it by decomposing parts from a PC SCSI assembly;
189: micro DB50 connector on ISA/PCI metal plate with 50-pin flat cable connector
190: on back side. The item is available at conventional PC shops.
191:
192: # Installation
193:
194: Instructions are available
195: in the [[How to install NetBSD/luna68k|luna68k_install]] document.
196:
197: # Future work<a name="futurework"></a>
198:
1.2 ! snj 199: X server; should not be hard.
1.1 ryoon 200: The X Consortium X11 release contains code for UNIOS-B/Mach2.5.
201: NetBSD/luna68k has the WSCONS interface, and the porting effort
202: will be concentrated around this point.
203: However, it's uncertain that color support will be implemented,
204: because the LUNA framebuffer is designed in 'planer format',
205: not 'packed pixel format.'
206:
207: LUNA-88K support would be a fun project;
208: CMU Mach3 MK84 release contains code for the LUNA-88K hardware,
209: and its peripheral devices are nearly identical to those of the LUNA.
210:
211: # History and Background of LUNA<a name="behindthescene"></a>
212:
213: The LUNA has an interesting behind the scene story.
214:
215: The LUNA hardware had two different operating systems;
216: a 4.3BSD derivative and a SVR3 variant.
217: The first one, named UNIOS-B, was was a port of Integrated Solution Inc.
218: UNIX product. ISI manufactured m68k based VME UNIX boxes.
219: Their OS had an interesting feature of TRFS (Translucent Remote File System)
220: as well as the popular SMI's NFS. The paper of TRFS was published
221: at USENIX Technical Conference (late '80, details unknown in this moment).
222:
223: TRFS runs atop its own RPC layer with a distinct ethertype;
224: the protocol can not operate across routing segments.
225: TRFS supports diskless client nodes. TRFS is a remarkably small network
226: filesystem. It is not built on VFS nor fssw[].
227: Each remote client process has a phantom kernel process in the TRFS server,
228: and operational semantics of UNIX I/O model is preserved across machines
229: unlike to NFS.
230:
231: ISI once made m68020 based deskside UNIX workstations,
232: which were available for the Japanese market.
233: The machine had its own bitmap windowing system, and marketed
234: against then successful SMI's sun2.
235:
236: The LUNA was a straight port of the ISI combined with OMRON's hardware design.
237: LUNA could boot via network and operate as a diskless TRFS client.
238: For unknown reasons, OMRON published little about how the LUNA
239: could be used as TRFS network nodes.
240:
241: Upfront to RISC computer boom, the company made a multiprocessor
242: variant of the LUNA; a 4 processor m88000 SMP box geared by CMU's Mach2.5.
243: The company was affiliated with Motorola, and the choice was
244: natural to them (the company acquired a Japanese unit of Data General
245: when it was abandoned by the parent company).
246: A m68040 variant was made later, and marketed as LUNA-II.
247:
248: The LUNA also has an interesting side story; it was a development
249: platform of the Japanese Xterminal for a while.
250: At late '80, a software company named ASTEC started developing an Xterminal
251: prototype. Engineers in the company knew that a LUNA could be netbooted
252: by another one. They started designing and building a propriety OS featuring
253: an UNIX-like API and homebrew TCP/IP protocol stack.
254: The LUNA was a quasi target hardware, and Xterminal images were downloaded
255: to it by TRFS network boot.
256:
257: Three Japanese companies bought the prototype design;
258: OMRON, JCC and Takaoka Electric. The last two made their own 68030 based
259: Xterminal hardware and deployed their products to the domestic market.
260: The Xterminal business was a moderate success. ASTEC provided prototype
261: Xserver software based on each release of X11 sample implementation
262: to the companies.
CVSweb for NetBSD wikisrc <wikimaster@NetBSD.org> software: FreeBSD-CVSweb
![[BACK]](/icons/back.gif){kind=icon}
Return to luna68k_info.mdwn CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [NetBSD Developer Wiki] / wikisrc / ports / luna68k