Translating MCP251-SPI Engine to P2

Comments

• ke4pjw Posts: 1,183

  2022-12-29 01:16

  MOVD isn't a valid instruction.
• jay_harlow Posts: 43

  2022-12-29 01:37

  hey thanks! need to remember that

  Jay
• evanh Posts: 16,407

  2022-12-29 01:48

  Prop2 is SETD
• JonnyMac Posts: 9,269

  2022-12-29 01:50 edited 2022-12-29 01:55

  You're doing a block transfer from hub to cog -- which is easier in Spin2. Have a look at Ada's online help file here:
  -- https://p2docs.github.io/hubmem.html#block-transfers

  This bit of code seems overwritten for simply reading two longs from the hub and writing them to cog variables arg0 and arg1.

                movd    argP,#arg0                        ' get 2 arguments ; arg0 to arg1
                mov     t2,t1                             '     │
                mov     t3,#2                             ' ───┘
  argP          rdlong  arg0,t2
                add     argP,d0                           ' point to next long in COG (incr COG ptr)
                add     t2,#4                             ' point to next addr in MAIN (incr MAIN ptr)
                djnz    t3,#argP
  

  It seems like it could be simplified to this:

                mov     t2, t1
                rdlong  arg0, t2
                add     t2, #4
                rdlong  arg1, t2 
  

  This accomplishes the same thing, modifies the same variables, and there are no shenanigans. This will also translate to Spin2.

  You could also it this way in the P2 which would be faster but less obvious (not tested -- I have only done this with ptra in the source position for rdlong).

                setq    #2-1                              ' get 2 arguments
                rdlong  arg0, t1