12 Aug
2007
12 Aug
'07
4 a.m.
My attempts to boot the MicroVax continue:
After much digging around at the house, I finally
found another TK50 cartridge. This one is in good
condition. I cleaned my SCSI TK50 drive, and mounted
the tape. It read fine, I was able to dd the contents
(one ~70mb VMS data file, with header and trailer) off
the tape and onto my PC. So far so good. I erased the
tape, and tried to write the NetBSD boot.fs to it.
dd if=boot.fs of=/dev/nst0 bs=512
And the drive whirred to life, shoeshining tape back
and forth,but actually making progress through the
media (it was accumulating on the take-up reel,
anyway). So I let it run. Eventually it failed with an
I/O error at a point where most of the tape was out of
the cartridge. It took over an hour to get there, and
only made it through one length of tape (no track
change/reverse) Rewind, erase tape again.
So, I tried just plain 'dd if=boot.fs of=/dev/nst0'
without specifying the block size (although from what
I've read, NetBSD wants a block size of 512 to boot)
That shoeshined like crazy too, so I stopped it.
Try again, with a block size of 16k (the same block
size I used to read in the tape). Still no go, it
still shoeshines, but not quite as much (barely).
What do I need to do here? How long should it take to
write two and a half megabytes of data to tape?
Thanks!
-Ian
12 Aug
12 Aug
4:40 a.m.
Mr Ian Primus wrote:
My attempts to boot the MicroVax continue:
...
And the drive whirred to life, shoeshining tape back
and forth,but actually making progress through the
media (it was accumulating on the take-up reel,
anyway). So I let it run. Eventually it failed with an
I/O error at a point where most of the tape was out of
the cartridge.
My notes for 2.10bsd say that tape 1 should look like this:
File RecSize #Recs Content
---- ------- ----- -------
0 512 41 tapeboot,tapeboot,boot
1 1024 28 mkfs
2 1024 27 restor
3 1024 26 icheck
4 10240 451 dump of root fs
5 10240 2277 tar of /usr without usr/src
6 10240 466 tar of /usr/src/{sys,include}
Which means one way to do this is to write a shell script
using the "non rewind device" (/dev/nst0) to write the proper
files, each with dd and the proper block size.
You can't just dd the entire thing in one go. And, if what
you have is a "tape image", you need to break it into files
first.
(or, use a utiltity which knows about the tape image and
will write the image back to the tape with the right block
size)
I'd start by just writing the boot file and see if that
works.
Here's the maketape.sh script from 2.11
#!/bin/sh
# File Blocksize Content
# 0 512 mtboot
# 512 mtboot
# 512 boot
# 1 1024 disklabel
# 2 1024 mkfs
# 3 1024 restor
# 4 1024 icheck
# 5 10240 root.dump
# 6 10240 file6.tar
# 7 10240 file7.tar
# 8 10240 file8.tar
TAPE=nrst0
cat mtboot mtboot boot | dd of=/dev/nrst0 obs=512
dd if=disklabel of=/dev/nrst0 obs=1024
dd if=mkfs of=/dev/nrst0 obs=1024
dd if=restor of=/dev/nrst0 obs=1024
dd if=icheck of=/dev/nrst0 obs=1024
dd if=root.dump of=/dev/nrst0 obs=10240
dd if=file6.tar of=/dev/nrst0 obs=10240
dd if=file7.tar of=/dev/nrst0 obs=10240
dd if=file8.tar of=/dev/nrst0 obs=10240
-brad
13 Aug
13 Aug
10:13 a.m.
On 8/12/07, Brad Parker <brad at heeltoe.com> wrote:
Here's the maketape.sh script from 2.11
#!/bin/sh
# File Blocksize Content
# 0 512 mtboot
# 512 mtboot
# 512 boot
# 1 1024 disklabel
# 2 1024 mkfs
# 3 1024 restor
# 4 1024 icheck
# 5 10240 root.dump
# 6 10240 file6.tar
# 7 10240 file7.tar
# 8 10240 file8.tar
TAPE=nrst0
cat mtboot mtboot boot | dd of=/dev/nrst0 obs=512
dd if=disklabel of=/dev/nrst0 obs=1024
dd if=mkfs of=/dev/nrst0 obs=1024
dd if=restor of=/dev/nrst0 obs=1024
dd if=icheck of=/dev/nrst0 obs=1024
dd if=root.dump of=/dev/nrst0 obs=10240
dd if=file6.tar of=/dev/nrst0 obs=10240
dd if=file7.tar of=/dev/nrst0 obs=10240
dd if=file8.tar of=/dev/nrst0 obs=10240
-brad
I believe that script will not actually work in practice to create a
bootable 2.11BSD tape without inserting padding to 512 byte boundaries
after the mtboot files.
The mtboot files here are only 490 bytes. If you use the maketape
program found there it takes care of padding to block boundaries.
http://www.tribug.org/pub/tuhs/PDP-11/Distributions/ucb/2.11BSD/
12 Aug
12 Aug
9:16 a.m.
Mr Ian Primus wrote:
My attempts to boot the MicroVax continue:
After much digging around at the house, I finally
found another TK50 cartridge. This one is in good
condition. I cleaned my SCSI TK50 drive, and mounted
the tape. It read fine, I was able to dd the contents
(one ~70mb VMS data file, with header and trailer) off
the tape and onto my PC. So far so good. I erased the
tape, and tried to write the NetBSD boot.fs to it.
dd if=boot.fs of=/dev/nst0 bs=512
Which drive do you have ? (there are F or G variants),
and which linux version are you using ? I thought the TK SCSI don't work
on "standard" linux, as the return an error in the SCSI Inquiry command
(it is too long)
Cheers
10:03 a.m.
Which drive do you have ? (there are F or G
variants),
According to the back of the enclosure - TK50Z-GA. So
it's a G then? The tape mechanism is an E4.
and which linux version are you using ? I thought
the TK SCSI don't work
on "standard" linux, as the return an error in the
SCSI Inquiry command
(it is too long)
Debian Linux. It certainally seems like it works,
although as it discovers the devices on bootup,
neither the SCSI controller nor the Linux kernel can
read anything off the drive as far as manufacturer or
model. It manages to figure out that it's a tape drive
though, because I can use it via /dev/nst0, and mt
works fine with it. I was also able to read a tape
just fine.
-Ian
13 Aug
13 Aug
4:44 a.m.
On 8/12/07, e.stiebler <emu at e-bbes.com> wrote:
Which drive do you have ? (there are F or G variants),
I have not worked with the TK50Z-G, only the -F, but AFAIK, the -F
ROMs aren't really SCSI compliant. There is a 5380 on a MicroVAX2000,
but something in the packets isn't quite kosher, and as such, you can
*not* put a TK50Z-F on a SCSI chain with other targets.
The TK50Z-G was supposed to, IIRC, remove those issues and work on
other targets than just a MicroVAX2000.
and which linux version are you using ? I thought the
TK SCSI don't work
on "standard" linux, as the return an error in the SCSI Inquiry command
(it is too long)
Hmm... that would be interesting to see the logs for.
-ethan
8:22 a.m.
"Ethan Dicks" wrote:
The TK50Z-G was supposed to, IIRC, remove those issues and work on
other targets than just a MicroVAX2000.
I have a TK50-GA right here which I have never powered on. I'm game
to attach it to a linux box and try to write tape using the dd script...
I'll try it later today and report back.
-brad
9:07 a.m.
I plugged my TK50Z-GA into a an ancient laptop (166mhz) running a linux
2.2 kernel and it worked as expected.
The kernel saw the device and noticed it was a tape.
The only writable media I have is "compactape-II", so caveat emptor. I
was able to copy a 1mb file onto it with dd using block sizes of 512,
10240, 20480 and 32768. It did not seem to want to stream (possibly
because of my slow machine) but the 32768 did a lot less rocking.
It sure seems like it would build a bsd tape cleanly, but with a lot of
rocking back and forth, or "shoeshining" (a phrase which makes me smile).
I have some tk70's also, hence the media. I hoped to write a few tapes
with the TK50Z-GA and the read them with a tk70 plugged into a qbus.
I guess I should find some plain old compacttap somewhere.
-brad
10:02 a.m.
It sure seems like it would build a bsd tape
cleanly, but with a lot of
rocking back and forth, or "shoeshining" (a phrase
which makes me smile).
I have been messing with it today as well, and I was
successful in finally getting the drive to write
straight through. I did:
mt -f /dev/nst0 drvbuffer 1
dd if=boot.fs of=/dev/nst0
By turning on buffering, it wrote cleanly. I can now
boot the tape on the Vax (sorta) - I was able to get
the boot loader to load, then it dies. But I think I
have solved the "tape writing" part of this.
-Ian
1 p.m.
Brad Parker wrote:
I plugged my TK50Z-GA into a an ancient laptop (166mhz)
running a linux
2.2 kernel and it worked as expected.
The kernel saw the device and noticed it was a tape.
The only writable media I have is "compactape-II", so caveat emptor. I
was able to copy a 1mb file onto it with dd using block sizes of 512,
10240, 20480 and 32768. It did not seem to want to stream (possibly
because of my slow machine) but the 32768 did a lot less rocking.
Jerome Fine replies:
I used to use a TK70 as my primary backup drive about 15 years
ago. Since I was reluctant to purchase "Compactape II" tape
media at the still exaggerated prices sold by DEC, I checked to
data sheets on a Memorex site - now long gone. Both the
physical properties AND the magnetic properties of the regular
Compactape and Compactape II were IDENTICAL. Based
on the few Compactape II media that I found by accident, the
only 2 differences that I ever found were the blue vs brown
label name and the PRICE!!!!!!!!!!
But in order to use a non-blank tape written by a TK50 on a
TK70, a bulk erase was required. I suspect that the same might
be required in the reverse direction. During the 10 years I used
the Compactape media on TK70 drives, I never had a failure -
and I ALWAYS verified EVERY tape that was written (almost
all with 256 MB of files) against the original drives after writing
to the media during a separate operation.
It sure seems like it would build a bsd tape cleanly,
but with a lot of
rocking back and forth, or "shoeshining" (a phrase which makes me smile).
I have some tk70's also, hence the media. I hoped to write a few tapes
with the TK50Z-GA and the read them with a tk70 plugged into a qbus.
I guess I should find some plain old compacttap somewhere.
I still have a few ready for the TK70. As I suggested,
they may need to be bulk erased. Contact me in a month
if you can't find any tapes. I am in Toronto.
Sincerely yours,
Jerome Fine
--
If you attempted to send a reply and the original e-mail
address has been discontinued due a high volume of junk
e-mail, then the semi-permanent e-mail address can be
obtained by replacing the four characters preceding the
'at' with the four digits of the current year.
12 Aug
12 Aug
10:04 a.m.
Mr Ian Primus wrote:
My attempts to boot the MicroVax continue:
Ian,
If you can turn up the complaint logging level of the scsi driver you
have running
the drive, try to write a 1 or 2 block file, and grab the error message
bytes from
the kernel, assuming that is logged.
You may not have a device which returns much. If there is a TK50 oem manual
online somewhere that has the messages that would help.
the shoeshining can be because the drive is not able to stream from the data
being sent from your system, or it can be continuous error retries.
A lot of drivers will try so many times, then rewind tape devices. The
rewind
may not be coded in Linux if you are using the norewind on close device.
Also if your linux box has more than the tape on the scsi controller,
you may
be overwhelmed by log entries and have problems finding the tape traffic.
Jim
10:14 a.m.
Mr Ian Primus wrote:
My attempts to boot the MicroVax continue:
After much digging around at the house, I finally
found another TK50 cartridge. This one is in good
condition. I cleaned my SCSI TK50 drive, and mounted
the tape. It read fine, I was able to dd the contents
(one ~70mb VMS data file, with header and trailer) off
the tape and onto my PC. So far so good. I erased the
tape, and tried to write the NetBSD boot.fs to it.
dd if=boot.fs of=/dev/nst0 bs=512
Another advice: get Eric Smith's tapecopy program, and check, if the
block sizes on tape are really the ones you thought you are writing.
Helped me a lot in the past.
Cheers
6341
days inactive
6342
days old
11 comments
7 participants
participants (7)
-
brad＠heeltoe.com -
emu＠e-bbes.com -
ethan.dicks＠gmail.com -
glen.slick＠gmail.com -
ian_primus＠yahoo.com -
jhfinedp3k＠compsys.to -
jwstephens＠msm.umr.edu