• Please review our updated Terms and Rules here

New PDP-11/53 Build

Continuing the saga...

Haven't reprogrammed the ATMEGA. But if I create a couple of RL02 partitions on the 2GB card and copy ZERORL.DSK to them, it does activate those partitions and shows them as DL0 and DL1. Copying a different disk image file, xxdp25.rl02, to one of the partitions causes automount to fail. This image file is a full 20480 records in length (=512 tracks * 40 sectors/track), while the ZERORL file is only 20450 records, but it appears to be overwriting the start of the next partition. Truncating the xxdp image to 20450 sectors before copying it results in automount working once again. But it won't boot. This does not appear to be related to the truncation, but a DMA failure. BOOT DL1 (the XXDP drive) reports an E.10 controller error.

The RLV12 registers contain:
17774400/120615 (CSR)
17774402/000000 (Bus address)
17774404/000000 (Disk address)
17774406/000000
17774410/000000
17774412/000000
17774414/000000
17774416/000000

Decoding the CSR, the following bits are set: ERR, E3, DS0, CRDY, F2, F1, DRDY
Controller and drive ready, operation=READ DATA, selected drive=1
Bus address and Disk address=0, as you expect from a boot
Indicated error = Non-Existent Memory

So at this point there are 3 possible failure points:
1. The RLV12 emulator (either the ATMEGA code or one of the CPLDs)
2. The KDJ11-D/S CPU module (Qbus interface)
3. The backplane

More investigation required...
 
Now we're getting nowhere!

To rule out #2 above, I removed the CPU and memory from my 11/73, which has a SCSI controller, and replaced them with the KDJ11-D/S. It booted up fine, so no problems with the CPU module. And since it was out of the 11/53 case, giving me easy access to the RLV12 emulator board, I reflashed the ATMEGA microcontroller with the hex file I built from the sources last week. Now on power-up I still get the spurious interrupt, but no output to the RLV12 emulator's console. :(

Even worse, reverting to the previous version of the firmware produces no output on the emulator console o_O
 
Hi - it sounded like a bus grant / board position problem - can you show the actual board layout? - Or put the scsi controller in your chassis and see if you can boot that?
 
I don't think it's a board position problem. The board layout couldn't be simpler: slot 1 has the KDJ11-D/S (M7554) and the RLV12 emulator is in the AB side of slot 2.

Turns out the no output problem was user error (no surprise) - I didn't realize I had to reprogram the fuses every time, so they were going back to their default values. Meanwhile, the developer sent me a new hex file and that solves all the SD card issues. Partitions get activated OK and I can do a DIR on the FAT32 (C:) drive. The 8GB card works now. However, the Qbus side of things is now not working at all - I'm thinking I need to update the CPLDs as well...
 
Ah - well that is about as minimal as you can get... I'm not sure if you have had it booting OK? - It would be good to see it boot from the known good SCSI card, for instance, just to prove the where the issues are - sorry if you have already done this!

Robin
 
Making progress... I switched to a different hex file that was included in the zip that the developer sent me, and now:
I still get a spurious interrupt at power-on. Restart gets me through the self-test and to the monitor, so it's usable but not optimal.
  • autoload works
  • show partition/show units work
  • dir shows the DOS partition
  • I am able to attach disk images in the DOS partition as drives
However, BOOT DL1 (which has XXDP25 disk image) gives E.10 controller error.
The RLV12 emulator has a bootloader that can be copied to RAM via DMA (LOAD command). Using that (after remounting the boot drive as DL0) produces a different failure...

Here's a gotcha: the drive images I have seem to have a 2-sector bootstrap. The boot ROMs on the 11/53 only read a single sector (256 bytes/400 octal). The emulator's bootstrap code reads 2 sectors (512 bytes/1000 octal). So it appears autoboot is not going to work unless I update the boot ROMs or get disk images with single-sector bootstraps.

At this point, I can see that the emulator is reading disk data into the PDP-11 RAM OK, but I haven't successfully booted anything (tried XXDP25 and RT11 disk images).

More to come...
 
Hi! - What did you do? - The bootstrap code only needs to load 1 sector (256 bytes), - the primary bootstrap.. loading 2 sectors is odd, but won't break it as the bootstrap will not care what is in those locations and it will load the secondary bootstrap from disk, if all is working as expected
 
I didn't really do anything except use the boot loader that the emulator's LOAD command provides and used an image from Jerome Fine's collection.

The boot code in this disk image jumps to location 000600 after the initial NOP/BR that identifies it as a valid boot sector, but 000600 is in the second sector, hence the need to read 2 sectors on boot. Maybe it's a Mentec thing, who knows?
 
It can't be a mentec thing - the RT-11 boot block is the same, but is it a confusion between loading one 256 byte sector - and loading one disk block? One disk block is 2 sectors and 512 bytes, so my RT-11 boot block starts:
1714846203515.png

so the jump is to 0566 - the second half of block 0
1714846263884.png

So any working bootstrap needs to load 512 bytes...

I always thought it was the first sector only that got loaded, but it can't be!
Robin
 
You must log in or register to reply here.
Back
Top