module sram_srom
  (input	     i_clk,
   input	     i_rst,
   input	     i_dir,
   input [1:0]	     i_act,
   input [23:0]	     i_addr,
   input [31:0]	     i_data,
   output reg [31:0] o_data,
   output	     o_we,
   output reg	     o_stb);
   
   (* ram_style = "block" *) reg [7:0] sram0 [4095:0];
   (* ram_style = "block" *) reg [7:0] sram1 [4095:0];
   (* ram_style = "block" *) reg [7:0] sram2 [4095:0];
   (* ram_style = "block" *) reg [7:0] sram3 [4095:0];
   (* ram_style = "block" *) reg [31:0] srom [4095:0];

   wire	      use_ram;
   wire	      use_rom;

   wire [7:0] data0, data1, data2, data3;
   wire [11:0] waddr;

   reg	       wram0, wram1, wram2, wram3;
   reg [7:0]   wdata1, wdata2, wdata3;
   reg [7:0]   rdata0, rdata1, rdata2, rdata3;
   reg [31:0]  mdata;

   reg	       waiter;
   
   integer     i;
   
   assign data0 = i_data[7:0];
   assign data1 = i_data[15:8];
   assign data2 = i_data[23:16];
   assign data3 = i_data[31:24];
   assign waddr = i_addr[13:2];
   assign use_ram = i_addr >= 24'h004000 && i_addr < 24'h008000;
   assign use_rom = i_addr < 24'h004000;
   //assign o_data = {rdata3, rdata2, rdata1, rdata0};
   assign o_we = !i_dir && i_act != 2'b00;

   initial begin
      $readmemh("/Users/car/Projects/hope/srom.hex", srom);
   end

   always @(posedge i_clk, posedge i_rst) begin
      if (i_rst)
	o_stb <= 0;
      else if (~o_stb)
	o_stb <= (i_act != 2'b00);
      else
	o_stb <= 0;
   end
   
   always @(posedge i_clk, posedge i_rst) begin
      if (i_rst) begin
	 rdata0 <= 8'd0;
	 rdata1 <= 8'd0;
	 rdata2 <= 8'd0;
	 rdata3 <= 8'd0;
	 mdata <= 32'd0;
      end else begin
	 if (wram0) sram0[waddr] <= data0;
	 if (wram1) sram1[waddr] <= wdata1;
	 if (wram2) sram2[waddr] <= wdata2;
	 if (wram3) sram3[waddr] <= wdata3;
	 
	 rdata0 <= sram0[waddr];
	 rdata1 <= sram1[waddr];
	 rdata2 <= sram2[waddr];
	 rdata3 <= sram3[waddr];

	 mdata <= srom[waddr];
      end // else: !if(i_rst)
   end
   
   always @(*) begin
      /*if (i_rst) begin
	 for (i = 0; i < 20; i = i + 1) begin
	    sram0[i] = 0;
	    sram1[i] = 0;
	    sram2[i] = 0;
	    sram3[i] = 0;
	 end
      end
      else */
      if (i_dir && use_ram && i_act != 2'b00) begin
	 case (i_act)
	   2'b01: begin
	      /* write word */
	      {wram3, wram2, wram1, wram0} = 4'b1111;
	      {wdata3, wdata2, wdata1} = {data3, data2, data1};
	   end
	   
	   2'b10: begin
	      /* write half-word */
	      {wdata3, wdata2, wdata1} = {data1, data0, data1};
	      if (i_addr[1]) begin
		 {wram3, wram2, wram1, wram0} = 4'b1100;
	      end else begin
		 {wram3, wram2, wram1, wram0} = 4'b0011;
	      end
	   end
	   
	   2'b11: begin
	      /* write byte */
	      {wdata3, wdata2, wdata1} = {data0, data0, data0};
	      case (i_addr[1:0])
		2'b00: begin
		   {wram3, wram2, wram1, wram0} = 4'b0001;
		end
		
		2'b01: begin
		   {wram3, wram2, wram1, wram0} = 4'b0010;
		end
		
		2'b10: begin
		   {wram3, wram2, wram1, wram0} = 4'b0100;
		end

		2'b11: begin
		   {wram3, wram2, wram1, wram0} = 4'b1000;
		end
	      endcase // case (i_addr[1:0])
	   end // case: 2'b11
	   
	   default: begin
	      {wram3, wram2, wram1, wram0} = 4'b0000;
	      {wdata3, wdata2, wdata1} = 24'd0;
	   end
	 endcase // case (i_act)
      end else begin // if (i_dir && use_ram && i_act != 2'b00)
	 {wram3, wram2, wram1, wram0} = 4'b0000;
	 {wdata3, wdata2, wdata1} = {data0, data0, data0};
      end
   end // always @ (posedge i_clk, posedge i_rst)
   
   always @(*) begin
      if (!i_dir && use_ram) begin
	 /* SRAM */
	 case (i_act)
	   2'b00: begin
	      /* nop */
	      o_data = 32'd0;
	   end
	   
	   2'b01: begin
	      /* read word */
	      o_data = {rdata3, rdata2, rdata1, rdata0};
	   end
	   
	   2'b10: begin
	      /* read half-word */
	      o_data[15:0] = i_addr[1] ?
			      {rdata3, rdata2} :
			      {rdata1, rdata0};
	      o_data[31:16] = 16'd0;
	   end
	   
	   2'b11: begin
	      /* read byte */
	      o_data[7:0] = (i_addr[1:0] == 2'b00) ? rdata0 :
			     (i_addr[1:0] == 2'b01) ? rdata1 :
			     (i_addr[1:0] == 2'b10) ? rdata2 :
			     rdata3;
	      o_data[31:8] = 24'd0;
	   end
	 endcase
      end else if (!i_dir && use_rom) begin
	 case (i_act)
	   2'b00: begin
	      /* nop */
	      o_data = 32'd0;
	   end
	   
	   2'b01: begin
	      /* read word */
	      o_data = mdata;
	   end
	   
	   2'b10: begin
	      /* read half-word */
	      o_data = {16'd0, (i_addr[1] ? mdata[31:16] : mdata[15:0])};
	   end
	   
	   2'b11: begin
	      /* read byte */
	      o_data = {24'd0, ((i_addr[1:0] == 2'b00) ? mdata[7:0] :
		       (i_addr[1:0] == 2'b01) ? mdata[15:8] :
		       (i_addr[1:0] == 2'b10) ? mdata[23:16] :
		       mdata[31:24])};
	   end
	 endcase
      end else o_data = mdata; // if (use_rom)
   end // always @ (posedge i_clk, posedge i_rst)

   initial begin
      // Simple test: PUSHB 0x42, infinite loop
      //mem[0] = 8'h10;  // PUSHB (opcode bits[7:3] = 2 → entry 2 → pushb)
      //mem[1] = 8'h42;  // immediate value
      //mem[2] = 8'h10;  // PUSHB again
      //mem[3] = 8'h33;  // different value
      // After this, will fetch garbage and execute nop/fetch
   end
endmodule