CoCoIO Buffer math outline

by patrick ulland

No matter what you do with CoCoIO, you'll be managing buffer addresses, based on a few hints from the WizNet chip. This includes starting and ending addresses and possible wraparound mid segment. Don't forget to reset the next read location when you are done! 

The overall process:

  • Use IRQ or periodically poll the Received Size Register. If >0 ...
  • Get the read or write pointer, AND to the buffer mask (buff size -1)
  • Add to the buffer base address. This is the start location.
  • Get the rx/tx data size.
  • Subtract from end of buffer. This is sizeof first segment.
  • Read that many bytes from the dataport.
  • Reset the rx/tx pointer to the start of the buffer.
  • Read the remaining bytes from the dataport.
  • Set the read pointer one address past the last one read
  • Send RECV ($40) to the socket command register

If you'd rather think in code, here is a Basic09 procedure to do the above. It's assuming we will always use socket one (everywhere $FF69,$04 appears could also be $05,$06,or $07).



PROCEDURE bufmath
(* bare outline of buffer math 8/19/21 ricku
(*
DIM raddr,rbase,rend,rmask,roff,rsize,rptr,rmax,segment:INTEGER
DIM int1,int2,int3,int4:INTEGER

(* rbase   - base address of read buffer
(* rmask   - buffer size mask
(* rptr    - raw ptr from Wizzy. AND with mask to get roff
(* roff    - read offset from start of buffer
(* rend    - end of read buffer
(* raddr   - absolute read address
(* rsize   - total bytes to injest
(* segment - bytes to injest (to EOB or after wraparound)


(* At this point you have sent a request to the server 
(* and are waiting for data 

     (* ____________________________________________ start of read page
     LOOP 
       (* FIXME _____ POLLS say IRQs are better!
       REPEAT 
           POKE $FF69,$04 \ POKE $FF6A,$26 \ (* get Sn_RX_RSR recv size
         rsize:=PEEK($FF6B)*256+PEEK($FF6B)
       UNTIL rsize<>0
       segment:=rsize \ rbase:=$6000
         POKE $FF69,$04 \ POKE $FF6A,$1E \ (* Sn_RXBUF_SIZE
         int1:=PEEK($FF6B)*256+PEEK($FF6B)
       rmask:=1024*int1-1
         POKE $FF69,$04 \ POKE $FF6A,$26 \ (* Sn_RX_RSR
       rsize:=PEEK($FF6B)*256+PEEK($FF6B)
         POKE $FF69,$04 \ POKE $FF6A,$28 \ (* Sn_RX_RD
       rptr:=PEEK($FF6B)*256+PEEK($FF6B)
         int1:=INT(rptr/256) \ int2:=rptr-256*int1
         int3:=INT(rmask/256) \ int4:=rmask-256*int3
         int1:=LAND(int1,int3) \ int2:=LAND(int2,int4)
       roff:=int1*256+int2+rbase

       (* _____________________________________________ parse the frame 
       raddr:=rbase+roff
       IF roff+rsize>rend THEN 
         segment:=rmask+1-roff
         (* GOSUB - process segment bytes of data
         raddr:=rbase \ segment:=rsize-rend
         (* GOSUB - process segment bytes of data 
       ELSE 
         segment:=rsize 
         (* GOSUB - process segment bytes of data
       ENDIF       
       (* __________________ end of a frame. increase Sn_RX_RD by rsize
         POKE $FF69,$04 \ POKE $FF6A,$28
       rptr:=PEEK($FF6B)*256+PEEK($FF6B)+rsize
         POKE $FF69,$04 \ POKE $FF6A,$28
         int1:=INT(rptr/256) \ int2:=rptr-256*int1
       POKE $FF6B,int1 \ POKE $FF6B,int2
       
       POKE $FF69,$04 \ POKE $FF6A,$01 \ POKE $FF6B,$40 \ (*Sn_CR to RECV
     ENDLOOP
Back to News