65816 JSR (a,x) - NESdev BBS

65816 JSR (a,x)
by Anes on 2015-10-28 (#157890)

My CPU is buggy (maybe very)...

Can't understand how JSR (a,x) or "JSR Absolute Indexed Indirect" works.
It's supposed for making a "Subroutine Table". It's the only opcode that goes out of my scope.
i have read the best docs, but still without understand.

Can somebody explain me?

Re: 65816 JSR (a,x)
by adam_smasher on 2015-10-28 (#157891)

IIRC

Code:

Proc0: ...

Proc1: ...

Proc2: ...

Table: DW Proc0, Proc1, Proc2

LDX #$02
JSR (Table, X) ; jumps to Proc1

Note that both Table and the Procs need to be part of the current program bank.

That clear things up?

Re: 65816 JSR (a,x)
by Anes on 2015-10-28 (#157893)

Yes, i understand it from the point of view of writing it in an 65816 asm.
The thing is that i don't know how to implement it in my cpu emu.

Taking into account this:

Code:

   WriteSNesMem(Cpu.SP, LOBYTE(Cpu.PC + 1));
   Cpu.SP--;
   WriteSNesMem(Cpu.SP, HIBYTE(Cpu.PC + 1));
   Cpu.SP--;   
   unsigned char PCL = ReadSNesMem(FULLPC);
   unsigned char PCH = ReadSNesMem(FULLPC + 1);
   //here i should do something with X!!
   MAKEPC(PCL, PCH);

Well the code is self much explicatory.
When do i hell add the X Register?

Re: 65816 JSR (a,x)
by tokumaru on 2015-10-28 (#157894)

Add X to the address in the instruction, and the result is the memory location from where the final address will be pulled from.

Re: 65816 JSR (a,x)
by tepples on 2015-10-28 (#157895)

Add X to ~~FULLPC~~ the provided address before reading memory. Source says it wraps within a 64K bank, and the first byte is written before the return address is pushed.

EDIT: thinko

Re: 65816 JSR (a,x)
by Anes on 2015-10-28 (#157897)

Thanks tepples... and wow what a good reference page you threw me...

Re: 65816 JSR (a,x)
by tokumaru on 2015-10-28 (#157899)

tepples wrote:

Add it to FULLPC

Wait, isn't FULLPC pointing to the instruction itself? As I see it, he has to read the address pointed by FULLPC (i.e. the bytes after the opcode), add X to that, and finally use the result to fetch the new PC value... isn't it?

Re: 65816 JSR (a,x)
by tepples on 2015-10-28 (#157900)

My bad. Corrected.

Re: 65816 JSR (a,x)
by Anes on 2015-10-28 (#157928)

EDIT: FULLDBRPC is FULLPBRPC;

So:

Code:

   WriteSNesMem(Cpu.SP, LOBYTE(Cpu.PC + 1));
   Cpu.SP--;
   WriteSNesMem(Cpu.SP, HIBYTE(Cpu.PC + 1));
   Cpu.SP--;   
   Cpu.address = ReadSNesMem(FULLPBRPC);
   Cpu.address |= ReadSNesMem(FULLPBRPC + 1) << 8;
   Cpu.address |= Cpu.PBR << 16;
   *(pu16)&Cpu.address += FULLX;
   Cpu.lo = ReadSNesMem(Cpu.address);
   Cpu.hi = ReadSNesMem(Cpu.address+1);
   MAKEPC(Cpu.lo, Cpu.hi);

pu16 = * unsigned short

The line: *(pu16)&Cpu.address += FULLX; is for handling wrap;
Is this ok?
Sorry i just want to be completly that what im doing is right....