RMAC and "stop" psuedo-ops

[Cyprian]

According to "Jaguar Object Processor Mode" "stop" doesn't take any parameter. How pass "O_STOPINTS" then?

http://d-bug.me/various/rmacman/rmac.html#jaguar-object-processor-mode

 

 

thanks

[ggn]

Hello,

 

A quick glance at the codebase (http://shamusworld.gotdns.org/cgi-bin/gitweb.cgi?p=rmac;a=blob;f=op.c;h=836f4de173b01add2b452414574c98b9c5c13953;hb=HEAD#l428)shows that the implementation of stop does not not take a parameter.

 

We can add it easily though, no probs. As a workaround for now you can use something like the following in lieu of the stop opcode:

 

dc.l 4<<28,O_STOPINTS

 

(not tested, but it's probably fine)

Edited September 28, 2020 by ggn
Geez, bbcode is kind of RIP

[Cyprian]

would be cool @ggn thanks

 

[Cyprian]

12 minutes ago, ggn said:

We can add it easily though, no probs. As a workaround for now you can use something like the following in lieu of the stop opcode:

 

dc.l 4<<28,O_STOPINTS

 

(not tested, but it's probably fine)

 

should not be as follow?

dc.l 0,(STOPOBJ|O_STOPINTS)

 

[ggn]

Dunno, probably. I just work here, man!

 

(by that I mean: I didn't write the OP portions, @Shamus did actually. And I haven't looked that serious at how the bits are laid out, sorry!)

 

(but it'll be correct by the time we issue the fix!)

Edited September 28, 2020 by ggn

[Shamus]

O_STOPINTS has nothing to do with the STOP opcode (at least according to the JTRM); the OP ignores anything in bits 3-63 of a STOP object.  It looks like it should be some kind of interrupt control bit, but it matches nothing (the closest is INT1 [$F000E0], but it should be $4 in that case instead of $8).

 

Dunno what to tell you other than, what do you need it for?  Does it really do anything?

 

N.B.: STOP is *not* a pseudo-op, it's a geniune, bona-fide, real as real can be opcode.  : )

Edited September 28, 2020 by Shamus
Corrections : P

[Cyprian]

that was from Atari workshop "GPUINT"

; Write a STOP object for the end of the list

		clr.l   d1
		move.l  #(STOPOBJ|O_STOPINTS),d0

		move.l  d1,(a0)+
		move.l  d0,(a0)+

 

[Shamus]

Well, it wouldn't be the first time they got things wrong.  As far as I can tell, OP STOP interrupts are controlled by bit 2 of INT1 ($F000E0), not by some magick bit in a STOP object.  No doubt the guy writing the documentation (LOL) was pulling his hair out. 

Edited September 29, 2020 by Shamus
:)

[ggn]

All right, so unless someone proves otherwise, rmac isn't getting a parameter in the STOP opcode.

[SCPCD]

Bit 3 of Stop object generate an interrupt if equal to 1. (documented on the most up to date JTRM and confirmed in netlist)

 

If bit 2 of INT1 register is enabled, then the stop interrupt will be sent back to the 68K.

 

Edited September 29, 2020 by SCPCD

[Cyprian]

thanks for clarification @SCPCD

[Cyprian]

15 hours ago, SCPCD said:

Bit 3 of Stop object generate an interrupt if equal to 1. (documented on the most up to date JTRM and confirmed in netlist)

 

If bit 2 of INT1 register is enabled, then the stop interrupt will be sent back to the 68K.

 

and actually I wonder whether other STOP bits are somehow used,

and what else we can see in the Jag Netlist?

[SCPCD]

Other bits are only used for software  (OB[0-3] registers) like for GPU objects.

 

[Shamus]

Well, it's an easy enough thing to add the logic to handle this in the OP assembler.  Will add it now that I have confirmation that it does something.