What does a setcommand statement do?

Comments

• ozpropdev Posts: 2,793

  2020-07-06 23:17

  setcommand is simply a method that would have been defined elsewhere in the SPI object.
  Look for something like this in the code.

  PUB setcommand(c,p)
  
• JonnyMac Posts: 9,510

  2020-07-07 00:45 edited 2020-07-07 04:00

  setcommand is simply a method that would have been defined elsewhere in the SPI object.

  Indeed, it is.

  PRI setcommand(cmd, argptr)        
      command := cmd << 16 + argptr  
      repeat while command  
  

  You're right, Jon: @Dpin is passing the hub address of the parameters which are used by the PASM cog. A non-zero value in the cmd parameter of setcommand() triggers the operation.
  
  I haven't yet developed a comfort with the SPI modes of the smart pins, so I'm working on an inline P2ASM program. This is the shiftout() function from that code. I really like that the P2 allows us to blend Spin2 and P2ASM without invoking a cog. I still need to wring-out my shiftin() and shiftio() functions -- then I will send to Parallax.

  pub shiftout(outval, bits, mode) | x, clk, do, di, tix
  
  '' TX bits from outval using defined SPI pins
  '' -- mode determines bit order and clock phase
  ''    * 0 = LSBPRE, 1 = MSBPRE, 2 = LSBPOST, 3 = MSBPOST
  ''    * modes 0..1 are CPHS0; modes 2..3 ard CPHS1
  
    longmove(@clk, @sck, 4)                                       ' copy pins & timing
  
    org
                  drvl      do                                    ' make sdo output if shared
  
                  test      mode, #%01                    wc      ' check for lsb mode
      if_nc       rev       outval                                ' lsbfirst -- reverse the bits
      if_c        mov       x, #32
      if_c        sub       x, bits
      if_c        shl       outval, x                             ' msbfirst -- slide into position
                             
                  test      mode, #%10                    wz      ' check clock mode
  
                  rep       #9, bits                              ' shift out the bits
                   rol      outval, #1                    wc      ' c <- outval.[31]
      if_z         drvc     do                                    ' c -> sdo (CPHS0)
                   waitx    tix                                   ' let sdo settle
                   drvnot   clk                                   ' toggle sclk
                   waitx    tix
      if_nz        drvc     do                                    ' c -> sdo (CPHS1)        
                   waitx    tix
                   drvnot   clk
                   waitx    tix
                 
  so_exit         drvl      do                                    ' leave sdo low
                             
    end
  
• Ariba Posts: 2,694

  2020-07-07 00:54

  Please don't just port P1 objects to P2 without rethinking it. Starting another cog for PASM code is not necessary for the P2. Spin2 is nearly as fast as PASM on the P1, and if you need the speed, you can use inline PASM, or smart pins.
  
  Andy
  
• JonnyMac Posts: 9,510

  2020-07-07 01:04 edited 2020-07-07 01:05

  +1. I love inline P2ASM for things like this. Last week I ported by PASM code for 1-Wire coms to inline P2ASM; works great, and I save a cog.
• JonTitus Posts: 193

  2020-07-07 02:50

  I agree, Jon. Nice to use the Smart-Pin capabilities, too, and free software from a lot of bit bashing and overhead. Are you working on an SPI object for Prop-2? --Jon
• JonnyMac Posts: 9,510

  2020-07-07 03:51 edited 2020-07-07 03:54

  Are you working on an SPI object for Prop-2? --Jon

  Yes; that inline P2ASM code is from my P2 SPI object. The methods are:
  -- start() (setup pins, clock mode, and clock timing)
  -- shiftout() (as above, works in all modes)
  -- shiftin() (needs testing -- waiting on some shift registers)
  -- shiftio() (needs testing)
  
  At the moment, I'm not convinced using smart pins is the best way for me to do SPI. That said, I reserve the right to be wrong and to change my mind down the line.