Macintosh-HW/SE-VGA
1
0
Fork 0
mirror of https://github.com/techav-homebrew/SE-VGA.git synced 2025-02-10 13:30:58 +00:00
Code Issues Projects Releases Wiki Activity

CPU cycle state machine

Browse Source
This commit is contained in:
techav 2021-04-17 15:41:53 -05:00
parent bbae9212e8
commit 7a519feab3
5 changed files with 1315 additions and 79 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

174
cpusnoop.sv
View File

@ -24,7 +24,25 @@ module cpusnoop (
output wire nvramWE, // VRAM Write strobe
input logic [2:0] ramSize // CPU RAM size selection
);
wire pendWriteLo; // low byte write to VRAM pending
wire pendWriteHi; // high byte write to VRAM pending
logic [13:0] addrCache; // store address for cpu writes to framebuffer
logic [7:0] dataCacheLo; // store data for cpu writes to low byte
logic [7:0] dataCacheHi; // store data for cpu writes to high byte
wire cpuBufSel; // is CPU accessing frame buffer?
logic [2:0] cycleState; // state machine state
// define state machine states
parameter
S0 = 3'h0,
S1 = 3'h1,
S2 = 3'h2,
S3 = 3'h3,
S4 = 3'h4,
S5 = 3'h5;
// when cpu addresses the framebuffer, set our enable signal
/* framebuffer starts $5900 below the top of RAM
* ramSize is used to mask the cpuAddr bits [21:9] to select the amount
* of memory installed in the computer. Not all possible ramSize selections
@ -40,22 +58,143 @@ module cpusnoop (
* $1 $0fa700 $100000 1.0MB Y [256kB 256kB][256kB 256kB]
* $0 $07a700 $080000 0.5MB Y [256kB 256kB][ --- --- ]
*/
wire pendWriteLo; // low byte write to VRAM pending
wire pendWriteHi; // high byte write to VRAM pending
logic [13:0] addrCache; // store address for cpu writes to framebuffer
logic [7:0] dataCacheLo; // store data for cpu writes to low byte
logic [7:0] dataCacheHi; // store data for cpu writes to high byte
wire cpuBufSel; // is CPU accessing frame buffer?
// when cpu addresses the framebuffer, set our enable signal
always_comb begin
// remember cpuAddr is shifted right by one since 68000 does not output A0
if(ramSize == cpuAddr[20:18] && cpuAddr[22:21] == 2'b00 && cpuAddr[17:14] == 4'b1111) begin
cpuBufSel <= 1'b1;
end else begin
cpuBufSel <= 1'b0;
end
end
// CPU Write to VRAM state machine
always @(negedge pixClock or negedge nReset) begin
if(!nReset) begin
cycleState <= S0;
pendWriteHi <= 0;
pendWriteLo <= 0;
addrCache <= 0;
dataCacheHi <= 0;
dataCacheLo <= 0;
end else begin
case (cycleState)
S0 : begin
// idle state, wait for valid address and ncpuAS asserted
if(ncpuAS == 0 && cpuBufSel == 1 && cpuRnW == 0) begin
cycleState <= S1;
end else begin
cycleState <= S0;
end
end
S1 : begin
// wait for either ncpuUDS or ncpuLDS to assert
// if ncpuAS negates first, then abort back to S0
if(ncpuAS == 1) begin
// cpu aborted cycle
cycleState <= S0;
end else if(ncpuUDS == 0 || ncpuLDS == 0) begin
if (ncpuUDS == 0) begin
pendWriteHi <= 1;
dataCacheHi <= cpuData[15:8];
end
if (ncpuLDS == 0) begin
pendWriteLo <= 1;
dataCacheLo <= cpuData[7:0];
end
// Valid CPU-VRAM cycle, so subtract constant $1380 from the
// cpu address and store the result in addrCache register.
// Constant $1380 corresponds to $2700 shifted right by 1.
// Once the selection bits above are masked out, we're left
// with buffer addresses starting at $2700
// e.g. with 4MB of RAM, fram buffer starts at $3FA700
// buffer address: 0011 1111 1010 0111 0000 0000 = $3FA700
// vram addr mask: 0000 0000 0011 1111 1111 1111 - $003FFF
// vram address: 0000 0000 0010 0111 0000 0000 = $002700
// Since CPU is 16-bit and does not provide A0, our cpuAddr
// signals are shifted right by one, so we need to do the same
// to our offset before subtracting it from cpuAddr
// offset: 0000 0000 0010 0111 0000 0000 = $002700
// shifted offset: 0000 0000 0001 0011 1000 0000 = $001380
addrCache <= cpuAddr[13:0] - 14'h1380;
cycleState <= S2;
end else begin
cycleState <= S1;
end
end
S2 : begin
// wait for sequence
if(pendWriteHi == 1 && pendWriteLo == 1 && seq < 5) begin
// we have enough time to write both before the next VRAM read
cycleState <= S3;
end else if(seq < 6) begin
// we have enough time to write the one pending before next VRAM read
if(pendWriteLo == 0) begin
cycleState <= S4;
end else begin
cycleState <= S3;
end
end else begin
// no time for a write sequence, wait
cycleState <= S2;
end
end
S3 : begin
// write CPU low byte to VRAM
if(pendWriteHi == 1) begin
cycleState <= S4;
end else begin
cycleState <= S5;
end
pendWriteLo <= 0;
end
S4 : begin
// write CPU high byte to VRAM
cycleState <= S5;
pendWriteHi <= 0;
end
S5 : begin
// wait for CPU to negate both ncpuUDS and ncpuLDS
if(ncpuUDS == 1 && ncpuLDS == 1) begin
cycleState <= S0;
end else begin
cycleState <= S5;
end
end
default: begin
// how did we end up here? reset to S0
cycleState <= S0;
end
endcase
end
end
always_comb begin
vramAddr[14:1] <= addrCache[13:0];
if(cycleState == S4) begin
vramAddr[0] <= 1;
end else begin
vramAddr[0] <= 0;
end
if(cycleState == S3 || cycleState == S4) begin
nvramWE <= 0;
end else begin
nvramWE <= 1;
end
if(cycleState == S3) begin
vramDataOut <= dataCacheLo;
end else if(cycleState == S4) begin
vramDataOut <= dataCacheHi;
end else begin
vramDataOut <= 0;
end
end
/*
// when cpu addresses the framebuffer, save the address
always @(negedge ncpuAS or negedge nReset) begin
@ -64,7 +203,8 @@ module cpusnoop (
end else begin
// here we match our ramSize jumpers and constants to confirm
// the CPU is accessing the primary frame buffer
if(cpuBufSel == 1'b1) begin
//if(cpuBufSel == 1'b1) begin
if(ramSize == cpuAddr[20:18] && cpuAddr[22:21] == 2'b00 && cpuAddr[17:14] == 4'b1111) begin
// We have a match, so subtract constant $1380 from the
// cpu address and store the result in addrCache register.
// Constant $1380 corresponds to $2700 shifted right by 1.
@ -107,7 +247,7 @@ module cpusnoop (
end
// set pending flags for cpu accesses & clear when that cycle comes back around
always @(negedge pixClock or negedge nReset) begin
/*always @(negedge pixClock or negedge nReset) begin
if(nReset == 1'b0) begin
pendWriteLo <= 1'b0;
pendWriteHi <= 1'b0;
@ -120,23 +260,24 @@ module cpusnoop (
pendWriteLo <= 1'b1;
end
end else begin
if(seq == 3'h1) begin
if(seq == 1 || seq == 3 || seq == 5) begin
pendWriteLo <= 1'b0;
end
if(seq == 3'h2) begin
if(seq == 2 || seq == 4 || seq == 6) begin
pendWriteHi <= 1'b0;
end
end
end
end
end*/
/*
always_comb begin
vramAddr[14:1] <= addrCache[13:0];
if(pendWriteLo == 1'b1 && seq == 3'h1) begin
if(pendWriteLo == 1'b1 && (seq == 1 || seq == 3 || seq == 5)) begin
vramAddr[0] <= 1'b0;
nvramWE <= 1'b0;
vramDataOut <= dataCacheLo;
end else if(pendWriteHi == 1'b1 && seq == 3'h2) begin
end else if(pendWriteHi == 1'b1 && (seq == 2 || seq == 4 || seq == 6)) begin
vramAddr[0] <= 1'b1;
nvramWE <= 1'b0;
vramDataOut <= dataCacheHi;
@ -146,4 +287,5 @@ module cpusnoop (
vramDataOut <= 8'h0;
end
end
*/
endmodule

2
primitives/primitives.sv
View File

@ -117,7 +117,7 @@ module vidShiftOut (
input wire vidActive,
input logic [2:0] seq,
input logic [7:0] parIn,
output wire out,
output wire out
);
/* Shift register functioning similar to a 74597, with 8-bit input latch
* and 8-bit PISO shift register output stage.

9
se-vga.sv
View File

@ -35,6 +35,7 @@ wire hActive;
wire hSEActive;
wire vActive;
wire vSEActive;
wire nvramWEpre;
logic [14:0] vidVramAddr;
logic [14:0] cpuVramAddr;
@ -83,13 +84,13 @@ cpusnoop cpusnp(
.cpuClk(cpuClk),
.vramAddr(cpuVramAddr),
.vramDataOut(cpuVramData),
.nvramWE(nvramWE),
.nvramWE(nvramWEpre),
.ramSize(ramSize)
);
always_comb begin
// vramAddr muxing
if(nvramWE == 1'b0) begin
if(nvramWEpre == 1'b0) begin
vramAddr <= cpuVramAddr;
end else begin
vramAddr <= vidVramAddr;
@ -97,7 +98,7 @@ always_comb begin
end
always_comb begin
if(nvramWE == 1'b0) begin
if(nvramWEpre == 1'b0) begin
vramData <= cpuVramData;
end else begin
vramData <= 8'bZZZZZZZZ;
@ -105,4 +106,6 @@ always_comb begin
vidVramData <= vramData;
end
assign nvramWE = nvramWEpre | pixClk;
endmodule

1203
sevga.vwf
View File

File diff suppressed because it is too large Load Diff

6
vgaout.sv
View File

@ -22,7 +22,7 @@ module vgaout (
output wire vidOut
);
reg [7:0] rVid;
//reg [7:0] rVid;
wire vidMuxOut;
wire vidActive; // combined active video signal
@ -66,7 +66,7 @@ end
*/
// latch incoming vram data on rising clock and sequence 7
always @(posedge pixClock or negedge nReset) begin
/*always @(posedge pixClock or negedge nReset) begin
if(nReset == 1'b0) begin
rVid <= 8'h0;
end else begin
@ -74,7 +74,7 @@ always @(posedge pixClock or negedge nReset) begin
rVid <= vramData;
end
end
end
end*/
always_comb begin
// combined video active signal