Interrupt jmp to $0

Fook42 · 17 January 2021, 00:04

Hello,
we encountered a strange behaviour of our Amiga1000 which is also seen in the WinUAE-Emu. But the cause for this is unclear to us, so maybe someone has a clue, what WinUAE is doing exactly to have this result.

situation: the code runs in kickstart-memory (fc0000->ffffff) after being called by cold-capture vector and then suddenly it is interrupted and a jump is executed to the address which is stored at $0. (see the red-marked data in the image of the winuae-debugger).

the intreqr-register holds a value of $3000 which should trigger an interrupt at $78 (marked in blue in the image).

we expect an interrupt for disk-handling to arrive, but this should not use the value at adress 0 for the destination.

so what causes this jump to $0 /to the address there?

jotd · 17 January 2021, 00:22

Easy way to debug that.

First set a breakpoint at 0, then set history to for ex 100:

Code:

f 0
H 100

run your code/reboot/whatever, reach the breakpoint. Using H again you can see which instructions led to 0.

ross · 17 January 2021, 00:25

My crystal ball tells me you have something like this at end of 24-bit memory:

00FFFFF0 0000 0000 0000 0000 0000 0000 0000 0000

Amiga do not use Autovector for IRQs..

Fook42 · 17 January 2021, 00:27

Quote:

Originally Posted by jotd

Easy way to debug that.

not that easy

the breakpoint to adress $0 will not help as the PC never reaches $0 in this scenario.. only the value from $0 is taken for the jump to the next instruction..
and i took the value as a breakpoint (as you can see in the attached screenshot)... but the "History" does not show any jump or specific command for this.. only the PC suddenly changes to the new value (see the red line in the screenshot).

ross · 17 January 2021, 00:42

Ok, it took me a while but I found my post on why it happens

:
http://eab.abime.net/showpost.php?p=1157566&postcount=8

jotd · 17 January 2021, 00:45

oh you already know "H"... read too fast. There's also the "il" command that can breakpoint on interrupts. You can specify mask. Just ... a lot of values to try because the command doesn't tell you which interrupt did it.

Fook42 · 17 January 2021, 00:53

Quote:

Originally Posted by ross

Amiga do not use Autovector for IRQs..

thank you!!!.. that and your post seems really promissing!

indeed there is some "garbage" at $FFFFF0 .. which may lead to that problem.

17 January 2021, 00:04	#1
Fook42 Registered User Join Date: Aug 2016 Location: germany Posts: 58	Interrupt jmp to $0 Hello, we encountered a strange behaviour of our Amiga1000 which is also seen in the WinUAE-Emu. But the cause for this is unclear to us, so maybe someone has a clue, what WinUAE is doing exactly to have this result. situation: the code runs in kickstart-memory (fc0000->ffffff) after being called by cold-capture vector and then suddenly it is interrupted and a jump is executed to the address which is stored at $0. (see the red-marked data in the image of the winuae-debugger). the intreqr-register holds a value of $3000 which should trigger an interrupt at $78 (marked in blue in the image). we expect an interrupt for disk-handling to arrive, but this should not use the value at adress 0 for the destination. so what causes this jump to $0 /to the address there? Attached Thumbnails

17 January 2021, 00:22	#2
jotd This cat is no more Join Date: Dec 2004 Location: FRANCE Age: 49 Posts: 5,098	Easy way to debug that. First set a breakpoint at 0, then set history to for ex 100: Code: f 0 H 100 run your code/reboot/whatever, reach the breakpoint. Using H again you can see which instructions led to 0.

17 January 2021, 00:25	#3
ross Defendit numerus Join Date: Mar 2017 Location: Crossing the Rubicon Age: 50 Posts: 3,127	My crystal ball tells me you have something like this at end of 24-bit memory: 00FFFFF0 0000 0000 0000 0000 0000 0000 0000 0000 Amiga do not use Autovector for IRQs..

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
CIA periodic interrupt	jotd	Coders. Asm / Hardware	12	02 January 2021 22:22
Blitter interrupt during VERTB interrupt	phx	Coders. Asm / Hardware	32	03 December 2017 12:34
Q: Copper Interrupt	Herpes	Coders. Asm / Hardware	3	25 April 2016 14:31
Interrupt 7	BigT	support.WinUAE	2	07 September 2013 23:25
Level 7 interrupt	Kintaro	support.WinUAE	1	21 January 2004 18:31

17 January 2021, 00:42	#5
ross Defendit numerus Join Date: Mar 2017 Location: Crossing the Rubicon Age: 50 Posts: 3,127	Ok, it took me a while but I found my post on why it happens : http://eab.abime.net/showpost.php?p=1157566&postcount=8

17 January 2021, 00:45	#6
jotd This cat is no more Join Date: Dec 2004 Location: FRANCE Age: 49 Posts: 5,098	oh you already know "H"... read too fast. There's also the "il" command that can breakpoint on interrupts. You can specify mask. Just ... a lot of values to try because the command doesn't tell you which interrupt did it.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)