• Please review our updated Terms and Rules here

Scientific Micro Systems SMS 1000 model 40

@gslick: There are two interesting questions here for me, because I was able to write a disklabel to my XT-1140 drive with 2.11BSD on the SMS1000.

1.) Do you use a different patchlevel of 2.11BSD than me?
Mine is:
Current Patch Level: 431
Date: April 21, 2000

2.) Does it make a difference when you boot the 2.11BSD installation via TMSCP tape drive or via vtserver?

In my 29bsd-vtserver.tgz there is also a version of the vtserver enabled 2.11BSD standalone programs. Can you test the following please?

1.) Try to create a 2.9BSD partition table with my MSCP version of 2.9BSD and vtserver (for the test you can use the XT-1140 layout). Just restore the root filesystem. Then use the "rauboot3" bootblock. This should give you a bootable 2.9BSD system. You need to boot "ra(0,0)unix73" on your PDP11/73. If this works, you can go on testing with 2.11BSD. If not, maybe it might make sense to consider a firmare upgrade of your SMS foundation module.

2.) If youn sucessfuly created a bootable 2.9BSD system, pull out your TMSCP controller and try to write the 2.11BSD disklabel with vtserver and the version in my vtserver/src/pdpvtstand. If this works, create a new installation tape with my vtserver enabled 2.11BSD standalone programs and retry it with the tape. My test with 2.11BSD went at least well for "disklabel", "mkfs" and "icheck". Only "restor" failed.

// Peter
 
I was using the standard 2.11BSD distribution tape images available here for this experiment, without any additional patches applied.

https://www.tuhs.org/Archive/Distributions/UCB/2.11BSD/
Current Patch Level: 431
Date: April 21, 2000

When installing on a SCSI hard drive attached to a CMD CQD SCSI controller on other systems, the installation will complete successfully, except that the system will not be able to boot directly from the SCSI hard drive. There is a patch #441 to sys/mdec/rauboot.s that is necessary to install to resolve that issue.

That patch wouldn't factor directly into this issue I am currently seeing as the the standalone version of disklabel uses src/sys/pdpstand/ra.c and that is were things are failing. Maybe I'll try rebuilding disklabel with extra debug output to see what is happening. Debugging via printf() might be the easiest thing for me to do for now.

As the foundation module firmware on my SMS-1000 Model 40 is version V0.6, as some point I might try updating the EPROMs to a newer version. One concern I have about that is that in addition to the EPROMs, there are quite a few PALs and PROMs, some of which have different version labels on my SMS-1000 Model 40 compared to the photo here. I wonder if the newer EPROM firmware is also dependent on any PAL and/or PROM changes.

http://www.bitsavers.org/pdf/sms/firmware/SMS_1000/
http://www.bitsavers.org/pdf/sms/firmware/SMS_1000/SMS1000.jpg

I'll have to type up a list of all the PAL and PROM versions labels on my SMS-1000 Model 40 and note the differences. It would be interesting to see if other Foundation Module boards with different firmware versions have any of these programmable parts with different version labels.

Code:
        FAB 0004554-0001 REV C      FAB 0004554-0001 REV E1
U45     AM27S35DC       2124007   * 2124010
U46     AM27S35DC       2124009     2124009
U47     AM27S35DC       2124008     2124008
U48     PAL20L10CNS     2121001     2121001
U49     AMPAL16R4LDC    2122001     2122001
U59     AMPAL16L8LPC    2120001     2120001
U77     AMPAL16R4LDC    2116001     2116001
U78     DMPAL12L6ANC    2117001     2117001
U79     AMPAL16R6LDC    2118001     2118001
U80     AMPAL16L8LPC    2119001     2119001
U91     DMPAL20X8NC     2127001   * 2127001B
U94     DMPAL20X8NC     2127001   * 2127001B
U95     DMPAL20X8NC     2127001   * 2127001B
U111    PAL18L4CNS      2096001     2096001
U112    PAL20L2CNS      2097001   * 2097002
U113    HM-7649A        2098006   * 20982001F
U114    HM-7649A        2098005   * 20982000F
U115    AMPAL16R6LDC    2099001   * 2099002B
U116    DMPAL12L6AND    2110001     2110001
U125    AMPAL16L8LPC    2115001     2115001
U134    AMPAL16R4LDC    2109001     2109001
U135    AMPAL16R6LDC    2108001     2108001
U136    PAL16R4ACN      2111001     2111001

* Devices with different version labels
 
Last edited:
At the moment on the SMS1000 the 2.11BSD installation fails because of the 2.11BSD standalone programs. Maybe the 2.11BSD kernel version of the MSCP driver will work on the SMS1000. To find this out, we first need to get at least a root filesystem restored.

This is the reason why I asked you to test my 2.11BSD standalone programs on your SMS1000 because with these ones, I can do everything except "restor" but this might be, because of a faulty disklabel on my Maxtor XT-1140 disk.

// Peter
 
At the moment on the SMS1000 the 2.11BSD installation fails because of the 2.11BSD standalone programs. Maybe the 2.11BSD kernel version of the MSCP driver will work on the SMS1000. To find this out, we first need to get at least a root filesystem restored.

This is the reason why I asked you to test my 2.11BSD standalone programs on your SMS1000 because with these ones, I can do everything except "restor" but this might be, because of a faulty disklabel on my Maxtor XT-1140 disk.

// Peter

I spent some more time on this again yesterday. Using SIMH I did a fresh install of 2.11BSD patch level 431, then applied patches up through 444. Then dumped the SIMH disk image to a physical device (a 256MB CF card) and then used that as the boot device on my SMS-1000 Model 40 attached to a CMD CQD-220/TM, and an Adtron SDDS SCSI PC Card reader drive.

The CMD CQD-220/TM was configured as the primary MSCP controller at 17772150, the SMS-1000 Model 40 on-board MSCP controller was configured as the secondary MSCP controller at 17772154. I added an entry in /etc/dtab for the second MSCP controller.

When booted into 2.11BSD from the CMD CQD-220/TM using the normal kernel version of the MSCP driver I was able to use disklabel to create a disk label on the XT-2190 attached to the SMS-1000 Model 40 on-board MSCP controller, then use newfs to create new filesystems on two partitions on the XT-2190, then use dump to dump the root partition filesystem from the boot drive to a file, then use restor to restore the filesystem dump from the file to the root partition filesystem on the XT-2190, then use fsck to check that the restored filesystem was OK. So everything seems to work fine with the SMS-1000 Model 40 on-board MSCP controller when accessed through the normal 2.11BSD kernel version of the MSCP driver.

Code:
# disklabel -r /dev/rra8a
# /dev/rra8a:
type: MSCP
disk: RD54
label: SMS1000
flags:
bytes/sector: 512
sectors/track: 18
tracks/cylinder: 15
sectors/cylinder: 270
cylinders: 1205
rpm: 3600
interleave: 1
trackskew: 0
cylinderskew: 0
headswitch: 0           # milliseconds
track-to-track seek: 0  # milliseconds
drivedata: 0

7 partitions:
#        size   offset    fstype   [fsize bsize]
  a:    17280        0   2.11BSD     1024  1024         # (Cyl.    0 - 63)
  b:    17280    17280      swap                        # (Cyl.   64 - 127)
  c:   325350        0    unused     1024  1024         # (Cyl.    0 - 1204)
  g:   290790    34560   2.11BSD     1024  1024         # (Cyl.  128 - 1204)

# newfs /dev/rra8a
newfs: /sbin/mkfs -m 2 -n 135 -i 4096 -s 8640 /dev/rra8a
isize = 2160
m/n = 2 135

# newfs /dev/rra8g
newfs: /sbin/mkfs -m 2 -n 135 -i 4096 -s 145395 /dev/rra8g
isize = 36336
m/n = 2 135

# dump f /usr/tmp/root.dump /dev/rra0a
  DUMP: Date of this level 0 dump: Fri Jun  9 06:07:26 1995
  DUMP: Date of last level 0 dump: the epoch
  DUMP: Dumping /dev/rra0a (/) to /usr/tmp/root.dump
  DUMP: mapping (Pass I) [regular files]
  DUMP: mapping (Pass II) [directories]
  DUMP: estimated 5138 tape blocks on 0.13 tape(s).
  DUMP: dumping (Pass III) [directories]
  DUMP: dumping (Pass IV) [regular files]
  DUMP: DUMP: 5138 tape blocks on 1 tape(s)
  DUMP: DUMP IS DONE
  DUMP: Closing /usr/tmp/root.dump

# ls -l /usr/tmp/root.dump
-rw-r-----  1 root      5253120 Jun  9 06:09 /usr/tmp/root.dump

# restor rf /usr/tmp/root.dump /dev/rra8a
Last chance before scribbling on /dev/rra8a.
End of tape

# fsck /dev/rra0a
** /dev/rra0a
File System: /

** Last Mounted on /
** Phase 1 - Check Blocks and Sizes
** Phase 2 - Check Pathnames
** Phase 3 - Check Connectivity
** Phase 4 - Check Reference Counts
** Phase 5 - Check Free List
558 files, 4565 used, 5198 free

# fsck /dev/rra8a
** /dev/rra8a
File System:

** Last Mounted on
** Phase 1 - Check Blocks and Sizes
** Phase 2 - Check Pathnames
** Phase 3 - Check Connectivity
** Phase 4 - Check Reference Counts
** Phase 5 - Check Free List
558 files, 4565 used, 3938 free

After doing all of that without any issues, I then removed the CMD CQD-220/TM and reconfigured the SMS-1000 Model 40 on-board MSCP controller as the primary MSCP controller at 17772150. Then I tried to use the SMS-1000 Model 40 operation menu to boot from DU0 and it just hung.

So it appears that while the normal 2.11BSD kernel version of the MSCP driver works with the SMS-1000 Model 40 on-board MSCP controller, there is something about it that is incompatible with the 2.11BSD standalone MSCP driver used when booting from tape and trying to run the standalone tape version of disklabel, and maybe also something incompatible with the MSCP boot block.

I'll have to try building standalone tape versions of disklabel which print extra debug information to see if I can get some additional clues about what might be going wrong with the MSCP driver initialization.
 
I had to reduce the number MSCP packets in the 2.10BSD driver which I ported to 2.9BSD for the SMS controller:

Snippet from my adapted ra.c for 2.9BSD:

Code:
/* PK, set packets from 3 to 2 */
#define NRSPL2  2               /* log2 number of response packets */
#define NCMDL2  2               /* log2 number of command packets */

The same lines exist also in the 2.11BSD version of ra.c:

Code:
#define NRSPL2  3               /* log2 number of response packets */
#define NCMDL2  3               /* log2 number of command packets */

Reduce it to 2 and rebuild the kernel! With a value of 3, on my SMS1000 the 2.9BSD kernel also hung during boot.

// Peter
 
Before you made that NRSPL2 / NCMDL2 change to src/sys/pdpuba/ra.c where was it hanging during the boot process?

What it at least getting this far in the boot process before hanging?

Code:
73Boot from ra(0,0,0) at 0172150
:
: ra(0,0,0)unix
Boot: bootdev=02400 bootcsr=0172150

If I understand correctly how things work, there are the following steps in the boot process:
  1. The firmware bootstrap reads the boot block from disk and then starts executing it. For example this could be the SMS-1000 Operation menu, or the CMD CQD-220/TM firmware boot.
  2. The primary boot block in this case would be src/sys/mdec/rauboot.s which has its own code to reinitialize the MSCP controller, then look for the "boot" file in the root filesystem, and then read it from disk and start executing it.
  3. The secondary boot file starts execution at src/sys/pdpstand/M.s and then src/sys/pdpstand/boot.c which will load the specified file from the specified device and start executing it. This secondary bootstrap uses src/sys/pdpstand/ra.c when accessing MSCP devices. This standalone MSCP driver has its own code to reinitialize the MSCP controller.
  4. If the kernel is loaded by secondary bootstrap, it will uses its kernel version of the MSCP driver src/sys/pdpuba/ra.c when accessing MSCP devcies, which also has its own code to reinitialize the MSCP controller.
The boot message above is printed by src/sys/pdpstand/boot.c between steps 3 and 4 above. If the change you made was to address a hang in src/sys/pdpuba/ra.c I would expect that you would see the boot message above before the hang.

In my case it was hanging immediately after starting the SMS-1000 boot without printing any messages. So it would appear to be hanging somewhere in src/sys/mdec/rauboot.s before the primary boot block loads the secondary boot file.
 
Success! I finally got 2.11BSD booting on my SMS-1000 Model 40 from an XT-2190 drive attached to the on-board MSCP controller:

Code:
 * HIT ANY KEY WITHIN TIME LIMIT *
    SMS 1000


**    SMS 1000    **

1 STATUS           <--
2 OPERATION
3 EVALUATION
4 CONFIGURATION

ENTER NUMBER - 2


**OPERATION**

1 BACKUP WINCHESTR <--
2 RESTOR WINCHESTR
3 COPY
4 IDENTIFY
5 FORMAT FLOPPY
6 WINCHESTR MGT
7 BOOT HOST CPU

ENTER NUMBER - 7
Are you sure?Y
Booting host.


    SMS 1000           BOOTSTRAP
2044KW MEMORY          11/73 CPU

BOOTABLE DEVICES:
DEVICE         DEVICE          UNIT
NAME           TYPE            NUMBERS

DU             DSA             0-3
MS             TS              0

ENTER DEVICE NAME AND UNIT NUMBER:  DU0

73Boot from ra(0,0,0) at 0172150
:
: ra(0,0,0)unix
Boot: bootdev=02400 bootcsr=0172150

2.11 BSD UNIX #5: Fri Jun 9 03:07:24 PDT 1995
    root@curly.2bsd.com:/usr/src/sys/WATNEY

ra0: Ver 0 mod 0
ra0: RD51  size=325477

phys mem  = 4186112
avail mem = 3871872
user mem  = 307200

June  9 03:50:49 init: configure system

ra 0 csr 172150 vector 154 vectorset attached
ra ? csr 172154 vector 310 skipped:  No CSR.
tms ? csr 174500 vector 260 skipped:  No CSR.
erase, kill ^U, intr ^C
#
# disklabel /dev/rra0a
# /dev/rra0a:
type: MSCP
disk: RD54
label: SMS1000
flags:
bytes/sector: 512
sectors/track: 18
tracks/cylinder: 15
sectors/cylinder: 270
cylinders: 1205
rpm: 3600
interleave: 1
trackskew: 0
cylinderskew: 0
headswitch: 0           # milliseconds
track-to-track seek: 0  # milliseconds
drivedata: 0

7 partitions:
#        size   offset    fstype   [fsize bsize]
  a:    17280        0   2.11BSD     1024  1024         # (Cyl.    0 - 63)
  b:    17280    17280      swap                        # (Cyl.   64 - 127)
  c:   325350        0    unused     1024  1024         # (Cyl.    0 - 1204)
  g:   290790    34560   2.11BSD     1024  1024         # (Cyl.  128 - 1204)

It appears that with the "Foundation Module" firmware I have on my SMS-1000 Model 40, the on-board MSCP controller will not proceed to Step 4 of the initialization sequence if during Step 1 a non-zero Interrupt Vector value is written to the SA Register without also setting the IE bit.

One of the differences between src/sys/mdec/rauboot.s src/sys/pdpstand/ra.c and src/sys/pdpuba/ra.c is that the two bootstrap versions do not enable and use interrupts, while the normal kernel version does.

In the standard patch level 431 distribution the src/sys/pdpstand/ra.c version sets a non-zero Interrupt Vector without also setting the IE bit. Starting with patch level 432 the src/sys/mdec/rauboot.s version also sets a non-zero Interrupt Vector without also setting the IE bit.

I found that if I broke into ODT while the system was hung during the execution of the primary bootstrap it was hung in this loop in src/sys/mdec/rauboot.s waiting for the MSCP controller to set the Step 4 bit in the SA Register.

Code:
/
/ RA initialize controller
/
        mov     $RASTEP1,r0
        mov     raip,r1
        clr     (r1)+                   / go through controller init seq.
        mov     $icons,r2
1:
        bit     r0,(r1)
        beq     1b

The solution for my system was simple, modify src/sys/mdec/rauboot.s and src/sys/pdpstand/ra.c so that they write a zero Interrupt Vector value to the MSCP SA Register during initialization. Then rebuild rauboot and dd that to the boot block of the drive, and rebuild boot and copy that to the root directory of the boot drive.

Code:
# pwd
/usr/src/sys/mdec

# diff orig/rauboot.s rauboot.s
303c303
< icons:        RAERR + 033
---
> icons:        RAERR

# pwd
/usr/src/sys/pdpstand
#
# diff orig/ra.c ra.c
83c83
<               raaddr->rasa = RA_ERR | (0154/4);
---
>               raaddr->rasa = RA_ERR;
 
Here are some more detailed notes about the 2.11BSD installation issue I was seeing on my SMS-1000 Model 40. Might not be many other of these systems out there that would encounter this issue, so this information is more for my own future reference than anything else, as I'll most likely forget most of these details before too long.

The issue was that when booting from the standard 2.11BSD installation tape, all of the standalone executables would fail when they tried to initialize the SMS-1000 Model 40 on-board MSCP controller. The standalone executables on the standard 2.11BSD installation tape are: boot, disklabel, mkfs, restor, and icheck.

The failure was occurring for each of these standalong executables in the standalone version of the MSCP driver /usr/src/sys/pdpstand/ra.c in raopen() when the MSCP controller was failing to advance to initialization Step 4. The failure / retry message loop would repeat forever.

Code:
73Boot from tms(0,0,0) at 0174500
: ra(0,0)unix
ra(172150) fail step 4. retrying
ra(172150) fail step 4. retrying


73Boot from tms(0,0,0) at 0174500
: tms(0,1)
Boot: bootdev=06001 bootcsr=0174500
disklabel
Disk? ra(0,0)
ra(172150) fail step 4. retrying
ra(172150) fail step 4. retrying
ra(172150) fail step 4. retrying


73Boot from tms(0,0,0) at 0174500
: tms(0,2)
Boot: bootdev=06002 bootcsr=0174500
Mkfs
file system: ra(0,0)
ra(172150) fail step 4. retrying
ra(172150) fail step 4. retrying
ra(172150) fail step 4. retrying


73Boot from tms(0,0,0) at 0174500
: tms(0,3)
Boot: bootdev=06003 bootcsr=0174500
Restor
Tape? tms(0,5)
Disk? ra(0,0)
ra(172150) fail step 4. retrying
ra(172150) fail step 4. retrying


73Boot from tms(0,0,0) at 0174500
: tms(0,4)
Boot: bootdev=06004 bootcsr=0174500
Icheck
File: ra(0,0)
ra(172150) fail step 4. retrying
ra(172150) fail step 4. retrying
ra(172150) fail step 4. retrying

Experimentially I found that that if I modified the standalone version of the MSCP driver /usr/src/sys/pdpstand/ra.c Step 1 initialization code in raopen() to write an Interrupt Vector value of zero when it is not setting the IE Interrupt Enable bit then the MSCP controller hang before advancing to initialization Step 4 no longer occurred.

Rebuilding the standalone executables boot, disklabel, mkfs, restor, and icheck after the trivial one line modification to /usr/src/sys/pdpstand/ra.c resulted in single byte changes to each of those executables, with an octal word value 100033 (hex 0x801B) changing to 100000 (hex 0x8000).

On the PC that I used to create the physical 8mm installation tape I was using to boot my SMS-1000 Model 40, I used a binary file editor to make the same single byte changes to the standard 2.11BSD distribution standalone executable files boot, disklabel, mkfs, restor, and icheck. I also used a binary file editor to make a single byte change in the root.dump file that corresponds to the boot file that is restored to the root file system during installation. After recreating a physical 8mm installation tape with these binary patched versions of the boot, disklabel, mkfs, restor, icheck, and root.dump files I was able to boot my SMS-1000 Model 40 from that installation tape and complete the installation process as described in the "Installing and Operating 2.11BSD on the PDP-11" manual without issue.

Code:
# cd /usr/src/sys/pdpstand

# od boot | grep 100033
0042200  100033 000002 012716 000002 012746 010000 010346 004737
# od disklabel | grep 100033
0043520  010346 004737 045406 022626 005700 001365 012763 100033
# od mkfs | grep 100033
0037600  004737 041464 022626 005700 001365 012763 100033 000002
# od restor | grep 100033
0042640  010346 004737 044526 022626 005700 001365 012763 100033
# od icheck | grep 100033
0040100  010346 004737 041766 022626 005700 001365 012763 100033

# mkdir orig
# mv boot orig
# mv disklabel orig
# mv mkfs orig
# mv restor orig
# mv icheck orig

# ls -l orig
total 173
-rwxr-xr-x  1 root        35236 Apr  9  2000 boot
-rwxr-xr-x  1 root        37881 Apr  9  2000 disklabel
-rwxr-xr-x  1 root        33032 Apr  9  2000 icheck
-rwxr-xr-x  1 root        33158 Apr  9  2000 mkfs
-rwxr-xr-x  1 root        35742 Apr  9  2000 restor

# diff orig/ra.c ra.c
83c83
<               raaddr->rasa = RA_ERR | (0154/4);
---
>               raaddr->rasa = RA_ERR;

# make boot disklabel icheck mkfs restor
cc -O -DSTANDALONE -I/usr/include -I. -c ra.c
ar rv libsa.a ra.o
r - ra.o
ranlib libsa.a
ld -X -o boot M.o conf.o boot.o ubmapset.o libsa.a -lc
ld -X -o disklabel srt0.o conf.o disklabel.o displaylab.o libsa.a -lc
ld -X -i -o icheck srt0-i.o conf.o icheck.o libsa.a -lc
ld -X -o mkfs srt0.o conf.o mkfs.o libsa.a -lc
ld -X -i -o restor srt0-i.o conf.o restor.o libsa.a -lc

# ls -l boot disklabel icheck mkfs restor
-rwxr-x--x  1 root        35236 Jan 12 13:51 boot
-rwxr-x--x  1 root        37881 Jan 12 13:51 disklabel
-rwxr-x--x  1 root        33032 Jan 12 13:51 icheck
-rwxr-x--x  1 root        33158 Jan 12 13:51 mkfs
-rwxr-x--x  1 root        35742 Jan 12 13:51 restor

# cmp -l orig/boot boot
 17537  33   0
# cmp -l orig/disklabel disklabel
 18271  33   0
# cmp -l orig/icheck icheck
 16463  33   0
# cmp -l orig/mkfs mkfs
 16269  33   0
# cmp -l orig/restor restor
 17839  33   0

# od boot | grep 0042200
0042200  100000 000002 012716 000002 012746 010000 010346 004737
# od disklabel | grep 0043520
0043520  010346 004737 045406 022626 005700 001365 012763 100000
# od mkfs | grep 0037600
0037600  004737 041464 022626 005700 001365 012763 100000 000002
# od restor | grep 0042640
0042640  010346 004737 044526 022626 005700 001365 012763 100000
# od icheck | grep 0040100
0040100  010346 004737 041766 022626 005700 001365 012763 100000



C:\2.11BSD>fc /b boot boot.x
Comparing files boot and BOOT.X
00004480: 1B 00

C:\2.11BSD>fc /b disklabel disklabel.x
Comparing files disklabel and DISKLABEL.X
0000475E: 1B 00

C:\2.11BSD>fc /b mkfs mkfs.x
Comparing files mkfs and MKFS.X
00003F8C: 1B 00

C:\2.11BSD>fc /b restor restor.x
Comparing files restor and RESTOR.X
000045AE: 1B 00

C:\2.11BSD>fc /b icheck icheck.x
Comparing files icheck and ICHECK.X
0000404E: 1B 00

C:\2.11BSD>fc /b root.dump root.dump.x
Comparing files root.dump and ROOT.DUMP.X
00312080: 1B 00

After the successful installation from the patched installation tape, there was one additional issue. The system would hang inside the initial rauboot boot block when attempting to boot directly from the SMS-1000 hard drive. It appears that the SMS-1000 Model 40 MSCP controller also needs the rauboot boot block patch in patch 441. However, starting with patch 432, /usr/src/sys/mdec/rauboot.s also sets a non-zero Interrupt Vector value during initialization Step 1 without also setting the IE Interrupt Enable bit. After applying patches up through patch 441, /usr/src/sys/mdec/rauboot.s needs to then be modified to set a zero Interrupt Vector value, otherwise it will have the same hang issue that was occurring in /usr/src/sys/pdpstand/ra.c After applying patches up through patch 441, and then rebuilding the /usr/src/sys/mdec/rauboot.s and copying rauboot to the boot sector of the SMS-1000 hard drive, the system could then successfully boot directly from the hard drive.

Code:
# cd /usr/src/sys/mdec

# diff orig/rauboot.s rauboot.s
303c303
< icons:        RAERR + 033
---
> icons:        RAERR

# dd if=/mdec/rauboot of=/dev/rra0a count=1
 
I am wondering about one thing. Does your 2.11BSD runs without any problems without reducing these values to "2" in the "ra.c" kenel driver:

Code:
#define NRSPL2  3               /* log2 number of response packets */
#define NCMDL2  3               /* log2 number of command packets */

Because my 2.9BSD stopped working, during a:

Code:
# find / -print

it simply hung!

// Peter
 
I am wondering about one thing. Does your 2.11BSD runs without any problems without reducing these values to "2" in the "ra.c" kenel driver:

Code:
#define NRSPL2  3               /* log2 number of response packets */
#define NCMDL2  3               /* log2 number of command packets */

Because my 2.9BSD stopped working, during a:

Code:
# find / -print

it simply hung!

// Peter

I tried this on my 2.11BSD installation on my SMS-1000 Model 40 with the XT-2190 attached to the on-board MSCP controller. As described in my previous reply, the installation was done with from a physical tape drive with the only changes being binary patches to the standalone executables: boot, disklabel, mkfs, restor, and icheck, equivalent to the trivial one line modification to /usr/src/sys/pdpstand/ra.c, plus an updated rauboot sector with patch 441 applied.

No change was made to the /usr/src/sys/pdpuba/ra.c definitions of NRSPL2 or NCMDL2.

With the root filesystem root.dump restored to ra0g and mounted on / and the file6.tar archive restored to ra0g and mounted on /usr there were currently around 4600 total files on the two mounted filesystems.

The "find / -print" command ran to completion, printing an expected number of filenames.
 
Thanks for testing and good to know that the values of NRSPL2 or NCMDL2 seem not to be related to te SMS hardware. When I have time I will maybe search for the reason for the problem of my system with greater NRSPL2 an NCMDL2 values. But for now I am really happy with my stable 2.9BSD kernel on the 11/23.
// Peter
 
Hi all,

I'm new to the forum, but am a hoarder of old computer stuff.

FsIjKGnXsAEUSAp[1].jpg

I've recently acquired this box, an SMS1000 model 40. It was full of pretty mucky oily dust, i've stripped and airlined the whole thing. The PSU had a fault, rectifier diodes in one 1/2 of the bridge were dead-short. Whilst the unit powers on, there is nothing displayed on the VFD apart from seemingly random characters when you power cycle the box. I've had a look at the management card and cannot see anything strange, doesnt look like anything has gone pop and all tants are intact etc. I'd like to get it running so am looking for some tips as to where to start.

There are quite a few IO cards in the back, one with the dual processor DEC chip on board, some serial IO, RAM cards etc, its also got a 5.25MFM from Hitachi and an 8" FDD.

I can of course provide more pictures if anyone is interested.

regards,

Paul.
 
Hello,

If you havent found it yet the link below is a very helpful starting point for this system. I have one and have had a lot of fun with it. I would strongly recommend deoxit for the backplane edge connectors as well as the edge connectors for the PSU board and SMS system board. Be sure to check all your psu outputs carefully, under load. Try it with all the dec boards out of the backplane. you should be able to get into the SMS menus on the front panel or via serial terminal. If not you have to get that 80186 system board going first. Also note that DEC board order/placement is very important to avoid damage or other issues. Be sure to read up on that in all the DEC documentation available out there. Your system may have CD/QQ slots for the first two slots. If so it is likely marked with a sticker on the backplane. The linked manual explains this and more.
 
Last edited:
Back
Top