How to wait on a pin state?

Dave Hein

I was updating my P1Spin bytecode interpreter this morning, and I'm trying to figure out how to handle the WAITPEQ and WAITPNE bytecodes. When I did this a few years ago I use SETPAE, SETPBA, SETPAN or SETPBN followed by a WAITPAT. However, those SETXXX instructions no longer exists. It seems now I must do a MODCZ, DECOD, SETPAT and then WAITPAT. Is that the way best way to wait on a pin where the pin number is specified in a register, or is there an easier way to do it that requires fewer instructions?

evanh

For a single pin SETSEx/WAITSEx pair would be the way to go. Or TESTP is another option.

evanh

Now you mention it I did have a fiddle with your old serial code that I've been using. You'll see I've used jnse1 instead of waitse1 to maintain the non-IRQ-blocking arrangement you already had:

'*******************************************************************************
'  COG Code
'*******************************************************************************

'smartpin version
getch
		testp   #rx_pin         wz
if_nz		jmp     #getch                'wait while Smartpin is idle
		rdpin   char, #rx_pin
_ret_		shr     char, #32-8           'shift the data to bottom of register


putch
		rqpin   temp1, #tx_pin  wc    'transmiting
		testp   #tx_pin         wz    'buffer free
if_nz_and_c	jmp     #putch                'wait while Smartpin is both full (nz) and transmitting (c)
_ret_		wypin   char, #tx_pin


'non-smartpin version
'getch
'		setse1  #%100_000000|rx_pin   'reset selectable event #1 (SE1) - needs an instruction time to take effect
'		mov     char, #0
'.getc0		jnse1   #.getc0               'wait while rx_pin is idle (HIGH) using SE1
'		getct   temp2                 '"Start the clock!"
'		add     temp2, bitcentre      'first interval aligns bit centred sampling on bit0, skipping the start bit
'		mov     temp1, #8             '8 bits to read in
'.getc1
'		addct1  temp2, bitcycles
'.getc2		jnct1   #.getc2               'wait a bit interval
'		testp   #rx_pin     wc        'sample the data bit (single sampling only)
'		rcr     char, #1              'log it
'		djnz    temp1, #.getc1
'
'_ret_		shr     char, #32-8           'shift the data to bottom of register


'bitcycles	long    (clock_freq + baud_rate/2) / baud_rate   'rounded to nearest
'bitcentre	long    clock_freq / baud_rate / 2 - 20          'an estimated 20 clocks for combined event and instruction lags present in the start-bit trigger event


'putch
'		or      char, #$100
'		shl     char, #1
'		mov     temp1, #10
'		getct   temp2
'.putc0
'		shr     char, #1   wc
'		drvc    #tx_pin
'		addct1  temp2, bitcycles
'.putc1		jnct1   #.putc1
'		djnz    temp1, #.putc0
'		ret

parm		long    0
char		long    0
temp1		long    0
temp2		long    0

evanh

And the reason I used branch on non-event instruction is because it can loop on itself, making for a tighter loop than testp + branch can achieve.

Dave Hein

After looking at is some more I found that I needed to use MODZ along with SETPAT. The C flag is set by another instruction, and I didn't need DECOD since WAITPXX uses a mask for multiple bits. It turned out to be the same number of instructions as I had been using before.