The Sinclair QL Forum

Very curious about those 'interrupt vectors' at $FFFF3 to $FFFFF... and why are there 13 bytes reserved and not 16?

Miracle used a clever trick to link in the Trump Card ROM. When the 68008 scans for the plug-in ROM at $C000 (after the memory sizing) they paged in their ROM at $C0000 (obscuring the RAM there), so this will be the next ROM to be linked in. When it gets called its initialisation code jumps to the real location at $10000 and pages itself out at $C0000, so there is again RAM there.

BTW, Minerva keeps scanning for ROMs above RAMTOP until it finds a duplicate ROM. So it will be possible to load ROMs on a Q68 in the area above 16MB (which is the maximum supported by Minerva) and get them linked in, without occupying system RAM.

janbredenbeek wrote: Fri Aug 02, 2024 10:43 pm Very curious about those 'interrupt vectors' at $FFFF3 to $FFFFF... and why are there 13 bytes reserved and not 16?

More than likely it is an error in the "QL Advaned User Guide", I will look into this, also I Adrian Dickens has never responded to to me about releasing the copyright.

I can not fi d anything about the 13 bytrs at the top of thd QL memory map.

In the QDOS Companion:

QDOS Companion, page 35 wrote:
Level 7 interrupts can normally only be triggered by additional hardware.
However, an early debugging tool has been left in the hardware, so that pressing CTRL ALT 2, 5 or 7 triggers the equivalent interrupt. A level 7 interrupt will usually cause the machine to lock up as it destroys the communication process between the 68008 and the 8049 IPC.

This does not seem relavent, anyone whst the Additional Hardware is?

Derek_Stewart wrote: Sat Aug 03, 2024 4:00 pm

QDOS Companion, page 35 wrote:
Level 7 interrupts can normally only be triggered by additional hardware.
However, an early debugging tool has been left in the hardware, so that pressing CTRL ALT 2, 5 or 7 triggers the equivalent interrupt. A level 7 interrupt will usually cause the machine to lock up as it destroys the communication process between the 68008 and the 8049 IPC.

This does not seem relavent, anyone whst the Additional Hardware is?

This would require both the IPL0L and IPL1L pins on the expansion connector to be asserted, resulting in a Level 7 interrupt.
However, the normal way for external hardware to assert an interrupt will be through the EXTINTL pin, which triggers the system interrupt and causes the CPU to execute the External Interrupt linked list.

You might use the L7 interrupt vector to start a debugger to recover from a lock-up using CTRL-ALT-7, but this is not guaranteed to work (it usually causes another lock-up).

The problem with the CTRL ALT 7 combo is it also triggers a reset in the IPC code too - and that's what causes the loss of synch as the QL and the IPC get out of step.

Hermes has some code changes to help with this a little, but due to the 'fragile' nature of the link, it's easily screwed up.

Wicksy wrote: Thu Aug 01, 2024 8:09 am Am interested how they would have achieved getting an individual Psion file of around 60k, when only 16k ROM is on offer.
Reading QL Wiki on Toolkit 2, it seems it uses a small amount of RAM as a ROM cartridge. That leads me to believe that would be the solution Psion would go with, albeit using a lot more RAM. Am happy to be corrected on this assumption.

If it were about just launching one of the Exchange programs from ROM, you could maybe store a smashed version in 16K ROM, which would allow it to be as big as 32K. I've used smash'ed programs before (i.e. self-extracting executables) on the QL and it works pretty well and the overhead isn't too big. That's still much smaller than what the Exchange programs are, plus you'd have to have a different ROM plug-in for each of the four programs. I suppose they could maybe offload the less used features onto microdrives and create a leaner version to get closer to the 32Kish limit.

Could they have somehow bank switched using a larger ROM/EPROM along with a smaller running RAM program to get this work? Unfortunately I don't know enough about the ROM/RAM layout of the QL to better understand how all that works.

It would have been possible to use something like the method used by ROMDisq - which offered upto 8Mb of storage accessible through the ROM port on the QL.

The other possibility is to have the Psion programs in a plug in cartidge in the slot on the side - as an Expansion device - ROM boards have been offered before (Eprom / GST) and that could also have offered RAM expansion too.

Pr0f wrote: Sat Aug 03, 2024 11:33 pm It would have been possible to use something like the method used by ROMDisq - which offered upto 8Mb of storage accessible through the ROM port on the QL.

Ah, that's right, forgot about ROMDisq, which came out in the late 90s.

Pr0f wrote: Sat Aug 03, 2024 11:33 pm The other possibility is to have the Psion programs in a plug in cartidge in the slot on the side - as an Expansion device - ROM boards have been offered before (Eprom / GST) and that could also have offered RAM expansion too.

That would have been my less preferred way, but likely could have preserved more RAM if no new one was added.

Thankyou to everyone who replied, it seems then the expansion bus is the answer.

Wicksy wrote: Mon Aug 05, 2024 1:58 pm
On the clones, the Q68 with 96k is only going to fit one package. I don't know what the pc emulators can do, other than Qemulator which is too small for the task.

How would you produce a ROM image from one of the Psion 4, as it's not the same as getting your system rom is it?

I do not understand this part, can you explain what you going to with thd Q68?