Questions about the P2 instrution set for anyone but Chip

rjo__ · 2018-02-13 15:10

I wanted to focus on the offending code.

Here is the complete code... I think it is failing hardware, but I have thought that before.
The story is I spent a couple of days soldering a couple of LaserPings to a slip ring and mounting the thing to a Parallax feedback 360 servo... assuming that the code would be a no brainer... first I couldn't get the smart pins to cooperate.. then I couldn't get the edge detectors to cooperate... and now I am reduced to this:) The idea is that when you make momentary contact with 3.3v... the moment is spit out to serial, with the first clock and the total period reported AND the LED lights up during contact... this actually does work!!! But seeing that LED stay dimly lit has me befuzzled.

rjo__ · 2018-02-13 15:13

The code snippet starts around line 74

rjo__ · 2018-02-13 15:16

ok... I see it... thanks.

rjo__ · 2018-02-13 15:17

looked at it for two ....ing days... Lordy

Cluso99 · 2018-05-11 05:14

Questions...

I presume that this is invalid

    _ret_   call     #label

Cluso99 · 2018-05-11 05:28

Thinking some more...
It would just require the verilog to NOT push the new return address onto the stack. I wonder if Chip thought about this?

msrobots · 2018-05-11 05:36

how about doing a JMP instead of _RET_ CALL.....?

there is probably a ret at the end of the code of label?

Mike

cgracey · 2018-05-11 05:43

Cluso99 wrote: »
Questions...

I presume that this is invalid
    _ret_   call     #label

The _RET_ only executes if not preempted by a branch. This means that '_RET_ DJNZ X,#loop' will only RETurn when the looping is done. In your example, the CALL would always take precedence over the _RET_.

Peter Jakacki · 2018-05-11 06:05

Now here's another one. I've been setting the carry with "modc 1 wc" but I'm told it should be "modc %1111 wc" however this test with MODZ says only the lsb is used.

First, the test word in TAQOZ which is simply called MODZ at the high level.

' MODZ ( n -- set ) 
_MODZ			altd	tos,#0
			modz	0 wz
			mov	tos,#0
		if_z	mov	tos,#1
			ret

So it will return a 1 if the z is set.
Now the test....

TAQOZ# 0 16 ADO CRLF I .B SPACE I MODZ . LOOP  
$00 0
$01 1
$02 0
$03 1
$04 0
$05 1
$06 0
$07 1
$08 0
$09 1
$0A 0
$0B 1
$0C 0
$0D 1
$0E 0
$0F 1 ok

Only the lsb was recognized.

cgracey · 2018-05-11 06:40

Peter Jakacki wrote: »
Now here's another one. I've been setting the carry with "modc 1 wc" but I'm told it should be "modc %1111 wc" however this test with MODZ says only the lsb is used.

First, the test word in TAQOZ which is simply called MODZ at the high level.
' MODZ ( n -- set ) 
_MODZ			altd	tos,#0
			modz	0 wz
			mov	tos,#0
		if_z	mov	tos,#1
			ret
So it will return a 1 if the z is set.
Now the test....
TAQOZ# 0 16 ADO CRLF I .B SPACE I MODZ . LOOP  
$00 0
$01 1
$02 0
$03 1
$04 0
$05 1
$06 0
$07 1
$08 0
$09 1
$0A 0
$0B 1
$0C 0
$0D 1
$0E 0
$0F 1 ok
Only the lsb was recognized.

Using %0001 means z = !c & !z.

Peter Jakacki · 2018-05-11 07:11

Ahhh, so it does...

TAQOZ# 1 MODZ . 0 ok
TAQOZ# 1 MODZ . 0 ok
TAQOZ# 1 MODZ 1 MODZ . 1 ok
TAQOZ# 1 MODZ 1 MODZ 1 MODZ . 0 ok

But that's kinda scary that we are using all these instructions that haven't been documented yet!

ozpropdev · 2018-05-11 07:30

From "instructions_v32.txt"

---------------
MODCZ constants
---------------

_CLR                    =       %0000
_NC_AND_NZ              =       %0001
_NZ_AND_NC              =       %0001
_GT                     =       %0001
_NC_AND_Z               =       %0010
_Z_AND_NC               =       %0010
_NC                     =       %0011
_GE                     =       %0011
_C_AND_NZ               =       %0100
_NZ_AND_C               =       %0100
_NZ                     =       %0101
_NE                     =       %0101
_C_NE_Z                 =       %0110
_Z_NE_C                 =       %0110
_NC_OR_NZ               =       %0111
_NZ_OR_NC               =       %0111
_C_AND_Z                =       %1000
_Z_AND_C                =       %1000
_C_EQ_Z                 =       %1001
_Z_EQ_C                 =       %1001
_Z                      =       %1010
_E                      =       %1010
_NC_OR_Z                =       %1011
_Z_OR_NC                =       %1011
_C                      =       %1100
_LT                     =       %1100
_C_OR_NZ                =       %1101
_NZ_OR_C                =       %1101
_C_OR_Z                 =       %1110
_Z_OR_C                 =       %1110
_LE                     =       %1110
_SET                    =       %1111

Peter Jakacki · 2018-05-11 07:37

So I also need instructions on where to look for instructions too! Thanks.

So modz %1100 wz copies C into Z. But what if we don't use wz?

ozpropdev · 2018-05-11 07:53

Peter Jakacki wrote: »

So modz %1100 wz copies C into Z. But what if we don't use wz?

Z state wold remain unchanged.

Peter Jakacki · 2018-05-11 08:03

Yes, but would there ever be a reason to not specify wz or is it simply mandatory?

Cluso99 · 2018-05-11 08:19

cgracey wrote: »
show previous quotes
Cluso99 wrote: »
Questions...

I presume that this is invalid
    _ret_   call     #label
The _RET_ only executes if not preempted by a branch. This means that '_RET_ DJNZ X,#loop' will only RETurn when the looping is done. In your example, the CALL would always take precedence over the _RET_.

Thanks Chip. Actually if we had though about it earlier, it could have been simple to make it work. Just don't push the call's return address onto the stack. This way, the call's return would bypass this call's return address, and go directly to the previous return address. Saves 2 clocks to boot

Cluso99 · 2018-05-11 09:32

Instruction MODCZ/Z/C

What do you think this instruction does?

                modz    1                       wz      ' =0001

Nope! It inverts Z! So in the test below you will see NZNZNZNZNZ...

Here are the tests...

		mov	lmm_x, #0       wz	' preset "Z"
loop
'		modz    1111	      wz	' set "Z"
'		modz    0000          wz      ' set "NZ"
                modz    1             wz      ' invert "Z" (0001)
	if_z	mov	lmm_x, #"Z"
	if_nz	mov	lmm_x, #"N"
		CALL	#@_HubTx
		jmp	#loop

Postedit:
I have just seen the above discussion with ccodes.
Not sure about anything other than 0000 and 1111 anymore. No time now tho'.

ozpropdev · 2018-05-11 09:46

modz 1 wz

is the equivalent of

modz _nc_and_nz wz

ozpropdev · 2018-05-11 10:28

Confirmation of "modz 1 wz" operation.

	modcz	_clr,_clr wcz	'clear c , clear z
	modz	1 wz	'z = !c & !z, z = 1
	modcz	_set,_clr wcz	'set c, clear z
	modz	1 wz	'z = !c & !z, z = 0
	nop

Returns the following flag states

>(P2 Debugger V3.34_V32c)================================================================================================
00000: FD7C006F              MODCZ   _CLR,_CLR WCZ
Flags (CZ) = __
>*
(P2 Debugger V3.34_V32c)================================================================================================
00001: FD6C026F              MODCZ   _CLR,_NC_AND_NZ WZ
Flags (CZ) = __
>*
(P2 Debugger V3.34_V32c)================================================================================================
00002: FD7DE06F              MODCZ   _SET,_CLR WCZ
Flags (CZ) = _Z
>*
(P2 Debugger V3.34_V32c)================================================================================================
00003: FD6C026F              MODCZ   _CLR,_NC_AND_NZ WZ
Flags (CZ) = C_
>*
(P2 Debugger V3.34_V32c)================================================================================================
00004: 00000000              NOP
Flags (CZ) = C_
>

Questions about the P2 instrution set for anyone but Chip

Comments