Add sideset assembly support.

[tannewt]

You must also set sideset_enable when creating the rp2pio.StateMachine
object.

Fixes #16 and fixes #21.

This needs adafruit/circuitpython#5766

[tannewt]

@dannystaple Please try this out. Note that it'll need a new CircuitPython build as well.

[dannystaple]

Which build? Will the 7.1.0-beta-3 build do this or do I need to pick up a CI artefact for it?

[dannystaple]

Ok - not sure this is working. Perhaps I'm missing something.
My code:

.program pwm_prog
.wrap_target
.side_set 1 opt                ; 1 sideset pin, optional.

    set x 5           side 0
    set y 10         
pwmloop:
    jmp x!=y skip           ; skip if they are different
    nop             side 1  ; drive side set pin high (which will be the one time they are the same)

skip:
    jmp y-- pwmloop         ; decrement x and go to pwm loop.
.wrap

(I know we can ignore wrap/wrap target)

The output assembled:

000 111_10000_001_00101
001 111_00000_010_01010
010 000_00000_101_00100
011 101_11000_010_00010
100 000_00000_100_00010

I think I am expecting:

000 111_10001_001_00101 
001 111_00001_010_01010 
010 000_00001_101_00100 
011 101_11000_010_00010 
100 000_00001_100_00010 

I could be wrong though.

I have used the 7.1.0-beta3, and just to be sure I was loading your code, called the module adafruit_pioasm_tst, and imported that. Assembly:

pwm_prog = adafruit_pioasm_tst.assemble(pwm_prog_raw)

FYI - the output I made with this, on the Pico:

instructions = ('{:016b}'.format(inst) for inst in pwm_prog)
formatted = ("{i}_{ss}_{src}_{dst}".format(
    i=inst[0:3],
    ss=inst[3:8],
    src=inst[8:11],
    dst=inst[11:16]
) for inst in instructions)

numbered =  ['{:03b} {}'.format(line, inst) for line, inst in enumerate(formatted)]

print('\n'.join(numbered))

[dannystaple]

Wait - being a donut. I made those expected things based on your unit tests. The last bit shoudln't be set - that was because you had a [1] in the test code I used to get example bits from. I might be getting what I mean.

[dannystaple]

Confirmed - output with the old library:

000 111_00000_001_00101
001 111_00000_010_01010
010 000_00000_101_00100
011 101_10000_010_00010
100 000_00000_100_00010

It's subtle - but it's the 1 for side set 0 and 11 for side set 1 that makes the difference. Time to get the scope out and check that it does what I think it should (it's a really stupid fixed pwm).

[tannewt]

Absolute newest builds should work with this now.

[dannystaple]

Looks good to me. It tests out with the right bytecode, comes with tests and example code.

[jepler]

Thank you!