Assembly differences between PNut and Flexgui

evanh

This fails to loop properly. There is a +$400 byte offset ($100 longwords) to the branch distance, or it might be +$3fc bytes ($ff longwords). If you comment the JMP #loop1 it'll work correctly using the JMP #$400.

dat	org
	jmp	#loop1


	orgh	$400
loop1
	drvnot #56
	waitx ##25_000_000/20

	long	0[5000]

	drvnot #57
	waitx ##25_000_000/20

	jmp #loop1
	jmp #$400

whicker

Does it also work if you comment out the

long	0[5000]

?

ozpropdev

evanh wrote: »

This fails to loop properly. There is a +$400 byte offset ($100 longwords) to the branch distance, or it might be +$3fc bytes ($ff longwords). If you comment the JMP #loop1 it'll work correctly using the JMP #$400.

dat	org
	jmp	#loop1


	orgh	$400
loop1
	drvnot #56
	waitx ##25_000_000/20

	long	0[5000]

	drvnot #57
	waitx ##25_000_000/20

	jmp #loop1
	jmp #$400

Like previous test results this code also works fine if I go back to V34L.

evanh

whicker wrote: »

Does it also work if you comment out the

long	0[5000]

?

It crashes because the +$400 offset makes the branch end up beyond the end of the program ... That said, if you have the remnant of a functioning version in hubRAM that isn't getting overwritten by the new download then, when run, it'll hit that old code and loop back just fine.

evanh

Here's my more long-winded version that always erases hubRAM before the test.

dat	org

'backup first part
	setq2	#9
	rdlong	0, ##@loop1

'backup second part
	setq2	#9
	rdlong	10, ##@hubcode2

'erase entire 512 kB hubRAM
	wrfast	#0, #0
	rep	@.rend, ##512*1024/4
	wflong	#0
.rend
	rdfast	#0, #0

'restore first part
	setq2	#9
	wrlong	0, ##@loop1

'restore second part
	setq2	#9
	wrlong	10, ##@hubcode2

'and away we go ...
	wrpin	#0, #56|1<<6
	jmp	#loop1



	orgh	$400
loop1
	drvnot	#56
	waitx	##25_000_000/20

	long	0[1000]
hubcode2
	drvnot	#57
	waitx	##25_000_000/20
	jmp	#loop1
	jmp	#$400