FMV.D.X imperas test passes

wally-pipelined/config/rv64ic/wally-config.vh

@@ -31,7 +31,7 @@
 `define XLEN 64
 
 // MISA RISC-V configuration per specification
-`define MISA (32'h00000104 | 0 << 5 | 0 << 3 | 1 << 18 | 1 << 20 | 1 << 12 | 1 << 0)
+`define MISA (32'h00000104 | 0 << 5 | 1 << 3 | 1 << 18 | 1 << 20 | 1 << 12 | 1 << 0)
 `define A_SUPPORTED ((`MISA >> 0) % 2 == 1)
 `define C_SUPPORTED ((`MISA >> 2) % 2 == 1)
 `define D_SUPPORTED ((`MISA >> 3) % 2 == 1)

wally-pipelined/regression/wave-dos/default-waves.do

@@ -49,6 +49,8 @@ add wave -hex /testbench/dut/hart/ieu/dp/ResultW
 add wave -hex /testbench/dut/hart/ieu/dp/RdW
 add wave -divider
 
+add wave -hex -r /testbench/*
+
 # appearance
 TreeUpdate [SetDefaultTree]
 WaveRestoreZoom {0 ps} {100 ps}

wally-pipelined/src/fpu/fctrl.sv

@@ -13,9 +13,12 @@ module fctrl (
   output logic [3:0] OpCtrlD,
   output logic       FmtD,
   output logic [2:0] FrmD,
-  output logic       WriteIntD);
+  output logic [1:0] FMemRWD,
+  output logic       OutputInput2D,
+  output logic       FWriteIntD);
 
 
+  logic IllegalFPUInstr1D, IllegalFPUInstr2D;
   //precision is taken directly from instruction
   assign FmtD = Funct7D[0];
   // *** fix rounding for dynamic rounding
@@ -53,6 +56,7 @@ module fctrl (
 
   always_comb begin
 	//checks all but FMA/store/load
+  IllegalFPUInstr2D = 0;
 	if(OpD == 7'b1010011) begin
   		casez(Funct7D)
 			//compare	
@@ -77,7 +81,7 @@ module fctrl (
                    else if (Funct7D[1] == 0) FResultSelD = 3'b111;
 			//output SrcW
 			7'b111100? : FResultSelD = 3'b110;
-			default    : FResultSelD = 3'bxxx;
+			default    : begin FResultSelD = 3'b0; IllegalFPUInstr2D = 1'b1; end
 		endcase
 	end
 	//FMA/store/load
@@ -92,12 +96,15 @@ module fctrl (
 			7'b0100111 : FResultSelD = 3'b111;
 			//load
 			7'b0000111 : FResultSelD = 3'b111;
-			default    : FResultSelD = 3'bxxx;
+			default    : begin FResultSelD = 3'b0; IllegalFPUInstr2D = 1'b1; end
 		endcase
 	end
   end
 
+  assign OutputInput2D = OpD == 7'b0100111;
 
+  assign FMemRWD[0] = OutputInput2D;
+  assign FMemRWD[1] = OpD == 7'b0000111;
 
 
 
@@ -143,7 +150,7 @@ module fctrl (
 
  
   always_comb begin
-    IllegalFPUInstrD = 0;
+    IllegalFPUInstr1D = 0;
     case (FResultSelD)
       // div/sqrt
       //  fdiv  = ???0
@@ -191,23 +198,24 @@ module fctrl (
       //  fmv.w.x = ???0
       //  fmv.w.d = ???1
       3'b110 : OpCtrlD = {3'b0, Funct7D[0]};
-      // output ReadData1
+      // output Input1
       //  flw       = ?000
       //  fld       = ?001 
-      //  fsw       = ?010
+      //  fsw       = ?010 // output Input2
-      //  fsd       = ?011
+      //  fsd       = ?011 // output Input2
       //  fmv.x.w  = ?100
       //  fmv.d.w  = ?101
       //		   {?, is mv, is store, is double or fcvt.d.w}
       3'b111 : OpCtrlD = {1'b0, OpD[6:5], Funct3D[0] | (OpD[6]&Funct7D[0])};
-      default : begin OpCtrlD = 4'bxxxx; IllegalFPUInstrD = 1'b1; end
+      default : begin OpCtrlD = 4'b0; IllegalFPUInstr1D = 1'b1; end
     endcase
   end
 
+  assign IllegalFPUInstrD = IllegalFPUInstr1D | IllegalFPUInstr2D;
   //write to integer source if conv to int occurs
   //AND of Funct7 for int results 
   //			is add/cvt       and  is to int  or is classify		 or     is cmp	       	and not max/min or is output ReadData1 and is mv
-  assign WriteIntD = ((FResultSelD == 3'b100)&Funct7D[3]) | (FResultSelD == 3'b101) | ((FResultSelD == 3'b001)&~Funct7D[2]) | ((FResultSelD == 3'b001)&OpD[6]);
+  assign FWriteIntD = ((FResultSelD == 3'b100)&Funct7D[3]) | (FResultSelD == 3'b101) | ((FResultSelD == 3'b001)&~Funct7D[2]) | ((FResultSelD == 3'b111)&OpD[6]);
   // 		      if not writting to int reg and not a store function and not move
-  assign FRegWriteD = ~WriteIntD & ~OpD[5] & ~((FResultSelD == 3'b111)&OpD[6]);
+  assign FRegWriteD = ~FWriteIntD & ~OpD[5] & ~((FResultSelD == 3'b111)&OpD[6]) & isFP;
 endmodule

wally-pipelined/src/fpu/fma1.sv

@@ -15,13 +15,13 @@
 //    normalize Normalization shifter
 //    round     Rounding of result
 //    exception Handles exceptional cases
-//    bypass    Handles bypass of result to ReadData1E or ReadData3E inputs
+//    bypass    Handles bypass of result to Input1E or Input3E inputs
 //    sign      One bit sign handling block 
 //    special   Catch special cases (inputs = 0  / infinity /  etc.) 
 //
-//   The FMAC computes FmaResultM=ReadData1E*ReadData2E+ReadData3E, rounded with the mode specified by
+//   The FMAC computes FmaResultM=Input1E*Input2E+Input3E, rounded with the mode specified by
 //   RN, RZ, RM, or RP.  The result is optionally bypassed back to
-//   the ReadData1E or ReadData3E inputs for use on the next cycle.  In addition,  four signals
+//   the Input1E or Input3E inputs for use on the next cycle.  In addition,  four signals
 //   are produced: trap, overflow, underflow, and inexact.  Trap indicates
 //   an infinity, NaN, or denormalized number to be handled in software;
 //   the other three signals are IEEE flags.
@@ -29,15 +29,15 @@
 /////////////////////////////////////////////////////////////////////////////
 
 /////////////////////////////////////////////////////////////////////////////
-module fma1(ReadData1E, ReadData2E, ReadData3E, FrmE,  
+module fma1(Input1E, Input2E, Input3E, FrmE,  
 			rE, sE, tE, bsE, killprodE, sumshiftE, sumshiftzeroE,  aligncntE, aeE
 			, xzeroE, yzeroE, zzeroE, xnanE,ynanE, znanE, xdenormE, ydenormE, zdenormE,
 			xinfE, yinfE, zinfE, nanE, prodinfE);
 /////////////////////////////////////////////////////////////////////////////
  
-	input logic 		[63:0]		ReadData1E;		// input 1
+	input logic 		[63:0]		Input1E;		// input 1
-	input logic		[63:0]		ReadData2E;     // input 2 
+	input logic		[63:0]		Input2E;     // input 2 
-	input logic 		[63:0]		ReadData3E;     // input 3
+	input logic 		[63:0]		Input3E;     // input 3
 	input logic 		[2:0]	 	FrmE;          	// Rounding mode
 	output logic 		[12:0]		aligncntE;    	// status flags
 	output logic 		[105:0]		rE; 				// one result of partial product sum
@@ -45,7 +45,7 @@ module fma1(ReadData1E, ReadData2E, ReadData3E, FrmE,
 	output logic 		[163:0]		tE;				// output logic of alignment shifter	
 	output logic 		[12:0]		aeE; 		// multiplier expoent
 	output logic 					bsE;				// sticky bit of addend
-	output logic 					killprodE; 		// ReadData3E >> product
+	output logic 					killprodE; 		// Input3E >> product
 	output logic					xzeroE;
 	output logic					yzeroE;
 	output logic					zzeroE;
@@ -68,7 +68,7 @@ module fma1(ReadData1E, ReadData2E, ReadData3E, FrmE,
 //	output logic 		[12:0]		aligncntE; 		// shift count for alignment
 
 
-	logic 					prodof; 		// ReadData1E*ReadData2E out of range
+	logic 					prodof; 		// Input1E*Input2E out of range
 
 
 
@@ -84,12 +84,12 @@ module fma1(ReadData1E, ReadData2E, ReadData3E, FrmE,
 
 //   Instantiate fraction datapath
 
-	multiply		multiply(.xman(ReadData1E[51:0]), .yman(ReadData2E[51:0]), .*);
+	multiply		multiply(.xman(Input1E[51:0]), .yman(Input2E[51:0]), .*);
-	align			align(.zman(ReadData3E[51:0]),.*);
+	align			align(.zman(Input3E[51:0]),.*);
 
 // Instantiate exponent datapath
 
-	expgen1			expgen1(.xexp(ReadData1E[62:52]),.yexp(ReadData2E[62:52]),.zexp(ReadData3E[62:52]),.*);
+	expgen1			expgen1(.xexp(Input1E[62:52]),.yexp(Input2E[62:52]),.zexp(Input3E[62:52]),.*);
 // Instantiate special case detection across datapath & exponent path 
 
 	special			special(.*);

wally-pipelined/src/fpu/fma2.sv

@@ -15,13 +15,13 @@
 //    normalize Normalization shifter
 //    round     Rounding of result
 //    exception Handles exceptional cases
-//    bypass    Handles bypass of result to ReadData1M or ReadData3M input logics
+//    bypass    Handles bypass of result to Input1M or Input3M input logics
 //    sign      One bit sign handling block 
 //    special   Catch special cases (input logics = 0  / infinity /  etc.) 
 //
-//   The FMAC computes FmaResultM=ReadData1M*ReadData2M+ReadData3M, rounded with the mode specified by
+//   The FMAC computes FmaResultM=Input1M*Input2M+Input3M, rounded with the mode specified by
 //   RN, RZ, RM, or RP.  The result is optionally bypassed back to
-//   the ReadData1M or ReadData3M input logics for use on the next cycle.  In addition,  four signals
+//   the Input1M or Input3M input logics for use on the next cycle.  In addition,  four signals
 //   are produced: trap, overflow, underflow, and inexact.  Trap indicates
 //   an infinity, NaN, or denormalized number to be handled in software;
 //   the other three signals are IMMM flags.
@@ -29,7 +29,7 @@
 /////////////////////////////////////////////////////////////////////////////
 
 /////////////////////////////////////////////////////////////////////////////
-module fma2(ReadData1M, ReadData2M, ReadData3M, FrmM,
+module fma2(Input1M, Input2M, Input3M, FrmM,
 			FmaResultM, FmaFlagsM, aligncntM, rM, sM,
 			tM,	normcntM, aeM, bsM,killprodM,
 			xzeroM,	yzeroM,zzeroM,xdenormM,ydenormM,
@@ -39,9 +39,9 @@ module fma2(ReadData1M, ReadData2M, ReadData3M, FrmM,
 );
 /////////////////////////////////////////////////////////////////////////////
  
-	input logic 		[63:0]		ReadData1M;		// input logic 1
+	input logic 		[63:0]		Input1M;		// input logic 1
-	input logic		[63:0]		ReadData2M;     // input logic 2 
+	input logic		[63:0]		Input2M;     // input logic 2 
-	input logic 		[63:0]		ReadData3M;     // input logic 3
+	input logic 		[63:0]		Input3M;     // input logic 3
 	input logic 		[2:0]	 	FrmM;          	// Rounding mode
 	input logic 		[12:0]		aligncntM;    	// status flags
 	input logic 		[105:0]		rM; 				// one result of partial product sum
@@ -50,7 +50,7 @@ module fma2(ReadData1M, ReadData2M, ReadData3M, FrmM,
 	input logic 		[8:0]		normcntM; 		// shift count for normalizer
 	input logic 		[12:0]		aeM; 		// multiplier expoent
 	input logic 					bsM;				// sticky bit of addend
-	input logic 					killprodM; 		// ReadData3M >> product
+	input logic 					killprodM; 		// Input3M >> product
 	input logic					prodinfM;
 	input logic					xzeroM;
 	input logic					yzeroM;
@@ -69,7 +69,7 @@ module fma2(ReadData1M, ReadData2M, ReadData3M, FrmM,
 	input logic					sumshiftzeroM;
 
 
-	output logic 		[63:0]		FmaResultM;     // output FmaResultM=ReadData1M*ReadData2M+ReadData3M
+	output logic 		[63:0]		FmaResultM;     // output FmaResultM=Input1M*Input2M+Input3M
 	output logic 		[4:0]		FmaFlagsM;    	// status flags
 	
 
@@ -120,18 +120,18 @@ module fma2(ReadData1M, ReadData2M, ReadData3M, FrmM,
 
 	add				add(.*);
 	lza				lza(.*);
-	normalize		normalize(.zexp(ReadData3M[62:52]),.*); 
+	normalize		normalize(.zexp(Input3M[62:52]),.*); 
-	round			round(.xman(ReadData1M[51:0]), .yman(ReadData2M[51:0]),.zman(ReadData3M[51:0]),.*);
+	round			round(.xman(Input1M[51:0]), .yman(Input2M[51:0]),.zman(Input3M[51:0]),.*);
 
 // Instantiate exponent datapath
 
-	expgen2			expgen2(.xexp(ReadData1M[62:52]),.yexp(ReadData2M[62:52]),.zexp(ReadData3M[62:52]),.*);
+	expgen2			expgen2(.xexp(Input1M[62:52]),.yexp(Input2M[62:52]),.zexp(Input3M[62:52]),.*);
 
 
 // Instantiate control logic
  
-sign				sign(.xsign(ReadData1M[63]),.ysign(ReadData2M[63]),.zsign(ReadData3M[63]),.*); 
+sign				sign(.xsign(Input1M[63]),.ysign(Input2M[63]),.zsign(Input3M[63]),.*); 
-flag2				flag2(.xsign(ReadData1M[63]),.ysign(ReadData2M[63]),.zsign(ReadData3M[63]),.vbits(v[1:0]),.*); 
+flag2				flag2(.xsign(Input1M[63]),.ysign(Input2M[63]),.zsign(Input3M[63]),.vbits(v[1:0]),.*); 
 
 assign FmaResultM = {wsign,wexp,wman};
 

wally-pipelined/src/fpu/fpdiv.sv

@@ -24,7 +24,7 @@
 
 // `timescale 1ps/1ps
 module fpdiv (DivSqrtDone, DivResultM, DivFlagsM, DivDenormM, DivOp1, DivOp2, DivFrm, DivOpType, DivP, DivOvEn, DivUnEn,
-	      DivStart, reset, clk);
+	      DivStart, reset, clk, DivBusyM);
 
    input [63:0] DivOp1;		// 1st input operand (A)
    input [63:0] DivOp2;		// 2nd input operand (B)
@@ -42,6 +42,7 @@ module fpdiv (DivSqrtDone, DivResultM, DivFlagsM, DivDenormM, DivOp1, DivOp2, Di
    output [4:0]  DivFlagsM;   	// IEEE exception flags 
    output 	 DivDenormM;   	// DivDenormM on input or output
    output 	 DivSqrtDone;
+   output    DivBusyM;
 
    supply1 	  vdd;
    supply0 	  vss;   
@@ -139,7 +140,7 @@ module fpdiv (DivSqrtDone, DivResultM, DivFlagsM, DivDenormM, DivOp1, DivOp2, Di
    // FSM : control divider
    fsm control (DivSqrtDone, load_rega, load_regb, load_regc, load_regd, 
 		load_regr, load_regs, sel_muxa, sel_muxb, sel_muxr, 
-		clk, reset, DivStart, DivOpType);
+		clk, reset, DivStart, DivOpType, DivBusyM);
    
    // Round the mantissa to a 52-bit value, with the leading one
    // removed. The rounding units also handles special cases and 

wally-pipelined/src/fpu/fpu.sv

@@ -13,8 +13,13 @@ module fpu (
   input  logic [`XLEN-1:0] SrcAM,       // Integer input being written into fpreg
   input  logic 		         StallE, StallM, StallW,
   input  logic             FlushE, FlushM, FlushW,
+  input  logic             RegWriteD,
   output logic [4:0]       SetFflagsM,
   output logic [31:0]      FSROutW,
+  output logic [1:0]       FMemRWM,
+	output logic             FStallE,
+  output logic             FWriteIntW,
+  output logic [`XLEN-1:0] FWriteDataM,       // Integer input being written into fpreg
   output logic             DivSqrtDoneE,
   output logic             IllegalFPUInstrD,
   output logic [`XLEN-1:0] FPUResultW);
@@ -72,7 +77,16 @@ module fpu (
   logic                    FmtD;
   logic                    DivSqrtStartD;
   logic [3:0]              OpCtrlD;
-  logic                    WriteIntD;
+  logic                    FWriteIntD;
+  logic                    OutputInput2D;
+  logic [1:0]              FMemRWD;
+
+  logic                    DivBusyM;
+	logic [1:0]              Input1MuxD, Input2MuxD;
+  logic                    Input3MuxD;
+  
+  //Hazard unit for FPU
+  fpuhazard hazard(.Adr1(InstrD[19:15]), .Adr2(InstrD[24:20]), .Adr3(InstrD[31:27]), .*);
 
   //top-level controller for FPU
   fctrl ctrl (.Funct7D(InstrD[31:25]), .OpD(InstrD[6:0]), .Rs2D(InstrD[24:20]), .Funct3D(InstrD[14:12]), .*);
@@ -108,10 +122,17 @@ module fpu (
   logic                    FmtE;
   logic                    DivSqrtStartE;
   logic [3:0]              OpCtrlE;
+	logic [1:0]              Input1MuxE, Input2MuxE;
+  logic                    Input3MuxE;
+  logic [63:0]             FPUResultDirE;
+  logic                    FWriteIntE;
+  logic                    OutputInput2E;
+  logic [1:0]              FMemRWE;
 
   //instantiation of E stage regfile signals
   logic [4:0]              RdE;
   logic [`XLEN-1:0]        ReadData1E, ReadData2E, ReadData3E;
+  logic [`XLEN-1:0]        Input1E, Input2E, Input3E, Input1tmpE;
 
   //instantiation of E/M stage div/sqrt signals
   logic                    DivSqrtDone, DivDenormM;
@@ -195,6 +216,13 @@ module fpu (
   flopenrc #(5) DEReg8(clk, reset, PipeClearDE, PipeEnableDE, InstrD[11:7], RdE);
   flopenrc #(4) DEReg9(clk, reset, PipeClearDE, PipeEnableDE, OpCtrlD, OpCtrlE);
   flopenrc #(1) DEReg10(clk, reset, PipeClearDE, PipeEnableDE, DivSqrtStartD, DivSqrtStartE);
+  flopenrc #(2) DEReg11(clk, reset, PipeClearDE, PipeEnableDE, Input1MuxD, Input1MuxE);
+  flopenrc #(2) DEReg12(clk, reset, PipeClearDE, PipeEnableDE, Input2MuxD, Input2MuxE);
+  flopenrc #(1) DEReg13(clk, reset, PipeClearDE, PipeEnableDE, Input3MuxD, Input3MuxE);
+  flopenrc #(64) DEReg14(clk, reset, PipeClearDE, PipeEnableDE, FPUResultDirW, FPUResultDirE);
+  flopenrc #(1) DEReg15(clk, reset, PipeClearDE, PipeEnableDE, FWriteIntD, FWriteIntE);
+  flopenrc #(1) DEReg16(clk, reset, PipeClearDE, PipeEnableDE, OutputInput2D, OutputInput2E);
+  flopenrc #(2) DEReg17(clk, reset, PipeClearDE, PipeEnableDE, FMemRWD, FMemRWE);
 
   //
   //END D/E PIPE
@@ -204,16 +232,27 @@ module fpu (
   //BEGIN EXECUTION STAGE
   //
 
+
+
+  // input muxs for forwarding
+  
+  mux4  #(64)  Input1Emux(ReadData1E, FPUResultDirW, FPUResultDirE, SrcAM, Input1MuxE, Input1tmpE);
+  mux3  #(64)  Input2Emux(ReadData2E, FPUResultDirW, FPUResultDirE, Input2MuxE, Input2E);
+  mux2  #(64)  Input3Emux(ReadData3E, FPUResultDirE, Input3MuxE, Input3E);
+  mux2  #(64)  OutputInput2mux(Input1tmpE, Input2E, OutputInput2E, Input1E);
+
+
+
   fma1 fma1 (.*);
 
   //first and only instance of floating-point divider
   fpdiv fpdivsqrt (.*);
 
   //first of two-stage instance of floating-point add/cvt unit
-  fpuaddcvt1 fpadd1 (AddSumE, AddSumTcE, AddSelInvE, AddExpPostSumE, AddCorrSignE, AddOp1NormE, AddOp2NormE, AddOpANormE, AddOpBNormE, AddInvalidE, AddDenormInE, AddConvertE, AddSwapE, AddNormOvflowE, AddSignAE, AddFloat1E, AddFloat2E, AddExp1DenormE, AddExp2DenormE, AddExponentE, ReadData1E, ReadData2E, FrmE, OpCtrlE, FmtE);
+  fpuaddcvt1 fpadd1 (AddSumE, AddSumTcE, AddSelInvE, AddExpPostSumE, AddCorrSignE, AddOp1NormE, AddOp2NormE, AddOpANormE, AddOpBNormE, AddInvalidE, AddDenormInE, AddConvertE, AddSwapE, AddNormOvflowE, AddSignAE, AddFloat1E, AddFloat2E, AddExp1DenormE, AddExp2DenormE, AddExponentE, Input1E, Input2E, FrmE, OpCtrlE, FmtE);
 
   //first of two-stage instance of floating-point comparator
-  fpucmp1 fpcmp1 (WE, XE, ANaNE, BNaNE, AzeroE, BzeroE, ReadData1E, ReadData2E, OpCtrlE[1:0]);
+  fpucmp1 fpcmp1 (WE, XE, ANaNE, BNaNE, AzeroE, BzeroE, Input1E, Input2E, OpCtrlE[1:0]);
 
   //first and only instance of floating-point sign converter
   fpusgn fpsgn (.*);
@@ -227,25 +266,25 @@ module fpu (
   //truncate to 64 bits
   //(causes warning during compilation - case never reached) 
 //   if(`XLEN > 64) begin // ***KEP this isn't usedand it causes a lint error
-//         DivOp1 = ReadData1E[`XLEN-1:`XLEN-64];
+//         DivOp1 = Input1E[`XLEN-1:`XLEN-64];
-// 	DivOp2 = ReadData2E[`XLEN-1:`XLEN-64];
+// 	DivOp2 = Input2E[`XLEN-1:`XLEN-64];
-//         AddOp1E = ReadData1E[`XLEN-1:`XLEN-64];
+//         AddOp1E = Input1E[`XLEN-1:`XLEN-64];
-// 	AddOp2E = ReadData2E[`XLEN-1:`XLEN-64];
+// 	AddOp2E = Input2E[`XLEN-1:`XLEN-64];
-//         CmpOp1E = ReadData1E[`XLEN-1:`XLEN-64];
+//         CmpOp1E = Input1E[`XLEN-1:`XLEN-64];
-// 	CmpOp2E = ReadData2E[`XLEN-1:`XLEN-64];
+// 	CmpOp2E = Input2E[`XLEN-1:`XLEN-64];
-//         SgnOp1E = ReadData1E[`XLEN-1:`XLEN-64];
+//         SgnOp1E = Input1E[`XLEN-1:`XLEN-64];
-// 	SgnOp2E = ReadData2E[`XLEN-1:`XLEN-64];
+// 	SgnOp2E = Input2E[`XLEN-1:`XLEN-64];
 //   end
 //   //zero extend to 64 bits
 //   else begin
-//         DivOp1 = {ReadData1E,{64-`XLEN{1'b0}}};
+//         DivOp1 = {Input1E,{64-`XLEN{1'b0}}};
-// 	DivOp2 = {ReadData2E,{64-`XLEN{1'b0}}};
+// 	DivOp2 = {Input2E,{64-`XLEN{1'b0}}};
-//         AddOp1E = {ReadData1E,{64-`XLEN{1'b0}}};
+//         AddOp1E = {Input1E,{64-`XLEN{1'b0}}};
-// 	AddOp2E = {ReadData2E,{64-`XLEN{1'b0}}};
+// 	AddOp2E = {Input2E,{64-`XLEN{1'b0}}};
-//         CmpOp1E = {ReadData1E,{64-`XLEN{1'b0}}};
+//         CmpOp1E = {Input1E,{64-`XLEN{1'b0}}};
-// 	CmpOp2E = {ReadData2E,{64-`XLEN{1'b0}}};
+// 	CmpOp2E = {Input2E,{64-`XLEN{1'b0}}};
-//         SgnOp1E = {ReadData1E,{64-`XLEN{1'b0}}};
+//         SgnOp1E = {Input1E,{64-`XLEN{1'b0}}};
-// 	SgnOp2E = {ReadData2E,{64-`XLEN{1'b0}}};
+// 	SgnOp2E = {Input2E,{64-`XLEN{1'b0}}};
 //   end
 
   //assign op codes
@@ -273,6 +312,7 @@ module fpu (
   logic [2:0]              FrmM;
   logic                    FmtM;
   logic [3:0]              OpCtrlM;
+  logic                    FWriteIntM;
 
   //instantiate M stage FMA signals here ***rename fma signals and resize for XLEN
   logic [63:0]		FmaResultM;
@@ -305,7 +345,7 @@ module fpu (
 
   //instantiation of M stage regfile signals
   logic [4:0]              RdM;
-  logic [`XLEN-1:0]        ReadData1M, ReadData2M, ReadData3M;
+  logic [`XLEN-1:0]        Input1M, Input2M, Input3M;
 
   //instantiation of M stage add/cvt signals
   logic [63:0]             AddResultM;
@@ -333,6 +373,14 @@ module fpu (
   logic [63:0]             CmpOp1M, CmpOp2M;
   logic [1:0]              CmpSelM;
 
+
+  //*****************
+  //fpregfile D/E pipe registers
+  //*****************
+  flopenrc #(64) EMFpReg1(clk, reset, PipeClearEM, PipeEnableEM, Input1E, Input1M);
+  flopenrc #(64) EMFpReg2(clk, reset, PipeClearEM, PipeEnableEM, Input2E, Input2M);
+  flopenrc #(64) EMFpReg3(clk, reset, PipeClearEM, PipeEnableEM, Input3E, Input3M);
+
   //*****************
   //fma E/M pipe registers
   //*****************  
@@ -423,6 +471,8 @@ module fpu (
   flopenrc #(1) EMReg4(clk, reset, PipeClearEM, PipeEnableEM, FmtE, FmtM);
   flopenrc #(5) EMReg5(clk, reset, PipeClearEM, PipeEnableEM, RdE, RdM);
   flopenrc #(4) EMReg6(clk, reset, PipeClearEM, PipeEnableEM, OpCtrlE, OpCtrlM);
+  flopenrc #(1) EMReg7(clk, reset, PipeClearEM, PipeEnableEM, FWriteIntE, FWriteIntM);
+  flopenrc #(2) EMReg8(clk, reset, PipeClearEM, PipeEnableEM, FMemRWE, FMemRWM);
 
   //
   //END E/M PIPE
@@ -432,6 +482,9 @@ module fpu (
   //BEGIN MEMORY STAGE
   //
 
+  
+  assign FWriteDataM = Input1M;
+
   fma2 fma2(.*);
 
   //second instance of two-stage floating-point add/cvt unit
@@ -466,7 +519,7 @@ module fpu (
   logic [4:0]             SgnFlagsW;
 
   //instantiation of W stage regfile signals
-  logic [`XLEN-1:0]        ReadData1W, ReadData2W, ReadData3W;
+  logic [`XLEN-1:0]        Input1W;
   logic [`XLEN-1:0]        SrcAW;
 
   //instantiation of W stage add/cvt signals
@@ -523,6 +576,8 @@ module fpu (
   flopenrc #(1) MWReg3(clk, reset, PipeClearMW, PipeEnableMW, FmtM, FmtW);
   flopenrc #(5) MWReg4(clk, reset, PipeClearMW, PipeEnableMW, RdM, RdW);
   flopenrc #(`XLEN) MWReg5(clk, reset, PipeClearMW, PipeEnableMW, SrcAM, SrcAW);
+  flopenrc #(64) MWReg6(clk, reset, PipeClearMW, PipeEnableMW, Input1M, Input1W);
+  flopenrc #(1) MWReg7(clk, reset, PipeClearMW, PipeEnableMW, FWriteIntM, FWriteIntW);
 
   ////END M/W PIPE
   //*****************************************
@@ -590,7 +645,7 @@ module fpu (
 		// output SrcAW
 		3'b110 : FPUResultDirW = SrcAW;
 		// output ReadData1
-		3'b111 : FPUResultDirW = ReadData1W;
+		3'b111 : FPUResultDirW = Input1W;
 		default : FPUResultDirW = {64{1'bx}};
 	endcase
   end

wally-pipelined/src/fpu/fpuhazard.sv

@@ -0,0 +1,70 @@
+///////////////////////////////////////////
+// fpuhazard.sv
+//
+// Written: me@KatherineParry.com 19 May 2021
+// Modified: 
+//
+// Purpose: Determine forwarding, stalls and flushes for the FPU
+// 
+// A component of the Wally configurable RISC-V project.
+// 
+// Copyright (C) 2021 Harvey Mudd College & Oklahoma State University
+//
+// Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation
+// files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, 
+// modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software 
+// is furnished to do so, subject to the following conditions:
+//
+// The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+//
+// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES 
+// OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS 
+// BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT 
+// OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+///////////////////////////////////////////
+
+`include "wally-config.vh"
+
+module fpuhazard(
+    input logic [4:0] Adr1, Adr2, Adr3,
+    input logic  FRegWriteE, FRegWriteM, FRegWriteW, 
+	  input logic [4:0] RdE, RdM, RdW,
+	  input logic  DivBusyM,
+	  input logic	RegWriteD,
+    input logic [2:0] FResultSelD, FResultSelE,
+  // Stall outputs
+	  output logic FStallE,
+    output logic [1:0] Input1MuxD, Input2MuxD, 
+    output logic Input3MuxD
+);
+
+
+  always_comb begin
+    // set ReadData as default
+    Input1MuxD = 2'b00; 
+    Input2MuxD = 2'b00;
+    Input3MuxD = 1'b0;
+    FStallE = DivBusyM;
+
+    if      ((Adr1 == RdE) & (FRegWriteE | ((FResultSelE == 3'b110) & RegWriteD))) 
+	if (FResultSelE == 3'b110) Input1MuxD = 2'b11; // choose SrcAM
+        else FStallE = 1'b1;                           // otherwise stall
+
+    else if ((Adr1 == RdM) & FRegWriteM) Input1MuxD = 2'b01; // choose FPUResultDirW
+    else if ((Adr1 == RdW) & FRegWriteW) Input1MuxD = 2'b11; // choose FPUResultDirE
+ 
+
+ 
+    else if      ((Adr2 == RdE) & FRegWriteE) FStallE = 1'b1;//***add a signals saying whether input 1, 2 or 3 are used
+    else if ((Adr2 == RdM) & FRegWriteM) Input2MuxD = 2'b01; // choose FPUResultDirW
+    else if ((Adr2 == RdW) & FRegWriteW) Input2MuxD = 2'b10; // choose FPUResultDirE
+
+
+
+
+    else if      ((Adr3 == RdE) & FRegWriteE) FStallE = 1'b1;
+    else if ((Adr3 == RdM) & FRegWriteM) FStallE = 1'b1;
+    else if ((Adr3 == RdW) & FRegWriteW) Input3MuxD = 1'b1; // choose FPUResultDirE
+  end 
+
+endmodule

wally-pipelined/src/fpu/fsm.sv

@@ -1,7 +1,7 @@
 module fsm (done, load_rega, load_regb, load_regc, 
 	    load_regd, load_regr, load_regs,
 	    sel_muxa, sel_muxb, sel_muxr, 
-	    clk, reset, start, op_type);
+	    clk, reset, start, op_type, divBusy);
 
    input 	clk;
    input 	reset;
@@ -20,6 +20,7 @@ module fsm (done, load_rega, load_regb, load_regc,
    output [2:0] sel_muxa;
    output [2:0] sel_muxb;
    output 	sel_muxr;
+   output logic	divBusy;
 
    reg 		done;      // End of cycles
    reg 		load_rega; // enable for regA
@@ -63,6 +64,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	       if (start==1'b0)
 		 begin
 		    done = 1'b0;
+		    divBusy = 1'b0;	
 		    load_rega = 1'b0;
 		    load_regb = 1'b0;
 		    load_regc = 1'b0;
@@ -77,6 +79,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	       else if (start==1'b1 && op_type==1'b0) 
 		 begin
 		    done = 1'b0;
+		    divBusy = 1'b1;	
 		    load_rega = 1'b0;
 		    load_regb = 1'b1;
 		    load_regc = 1'b0;
@@ -91,6 +94,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	       else if (start==1'b1 && op_type==1'b1) 
 		 begin
 		    done = 1'b0;
+		    divBusy = 1'b1;
 		    load_rega = 1'b0;
 		    load_regb = 1'b1;
 		    load_regc = 1'b0;
@@ -106,6 +110,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S1:
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b1;
 	       load_regb = 1'b0;
 	       load_regc = 1'b1;
@@ -120,6 +125,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S2: // iteration 1
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b1;
 	       load_regc = 1'b0;
@@ -134,6 +140,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S3:
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b1;
 	       load_regb = 1'b0;
 	       load_regc = 1'b1;
@@ -148,6 +155,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S4: // iteration 2
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b1;
 	       load_regc = 1'b0;
@@ -162,6 +170,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S5:
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b1;
 	       load_regb = 1'b0;
 	       load_regc = 1'b1;
@@ -176,6 +185,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S6: // iteration 3
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b1;
 	       load_regc = 1'b0;
@@ -190,6 +200,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S7:
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b1;
 	       load_regb = 1'b0;
 	       load_regc = 1'b1;
@@ -204,6 +215,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S8: // q,qm,qp
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -218,6 +230,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S9:  // rem
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -232,6 +245,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S10:  // done
 	    begin
 	       done = 1'b1;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -246,6 +260,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S13:  // start of sqrt path
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -260,6 +275,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S14:  
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b1;
 	       load_regb = 1'b0;
 	       load_regc = 1'b1;
@@ -274,6 +290,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S15:  // iteration 1
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b1;
 	       load_regc = 1'b0;
@@ -288,6 +305,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S16:  
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -302,6 +320,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S17:  
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b1;
 	       load_regb = 1'b0;
 	       load_regc = 1'b1;
@@ -316,6 +335,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S18:  // iteration 2
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b1;
 	       load_regc = 1'b0;
@@ -330,6 +350,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S19:  
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -344,6 +365,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S20:  
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b1;
 	       load_regb = 1'b0;
 	       load_regc = 1'b1;
@@ -358,6 +380,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S21:  // iteration 3
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b1;
 	       load_regc = 1'b0;
@@ -372,6 +395,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S22:  
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -386,6 +410,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S23:  
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b1;
 	       load_regb = 1'b0;
 	       load_regc = 1'b1;
@@ -400,6 +425,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S24: // q,qm,qp
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -414,6 +440,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S25:  // rem
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b1;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -428,6 +455,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  S26:  // done
 	    begin
 	       done = 1'b1;
+		   divBusy = 1'b0;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;
@@ -442,6 +470,7 @@ module fsm (done, load_rega, load_regb, load_regc,
 	  default: 
 	    begin
 	       done = 1'b0;
+		   divBusy = 1'b0;
 	       load_rega = 1'b0;
 	       load_regb = 1'b0;
 	       load_regc = 1'b0;

wally-pipelined/src/fpu/special.sv

@@ -10,46 +10,46 @@
 /////////////////////////////////////////////////////////////////////////////
 
 /////////////////////////////////////////////////////////////////////////////
-module special(ReadData1E, ReadData2E, ReadData3E, xzeroE, yzeroE, zzeroE,
+module special(Input1E, Input2E, Input3E, xzeroE, yzeroE, zzeroE,
 				xnanE, ynanE, znanE, xdenormE, ydenormE, zdenormE, xinfE, yinfE, zinfE);
 /////////////////////////////////////////////////////////////////////////////
 
-	input logic   	[63:0]     	ReadData1E;              // Input ReadData1E
+	input logic   	[63:0]     	Input1E;              // Input Input1E
-	input logic     	[63:0]     	ReadData2E;           	// Input ReadData2E
+	input logic     	[63:0]     	Input2E;           	// Input Input2E
-	input logic      	[63:0]    	ReadData3E;            	// Input ReadData3E 
+	input logic      	[63:0]    	Input3E;            	// Input Input3E 
-	output logic				xzeroE;		// Input ReadData1E = 0
+	output logic				xzeroE;		// Input Input1E = 0
-	output logic				yzeroE;		// Input ReadData2E = 0
+	output logic				yzeroE;		// Input Input2E = 0
-	output logic				zzeroE;		// Input ReadData3E = 0
+	output logic				zzeroE;		// Input Input3E = 0
-	output logic				xnanE;		// ReadData1E is NaN
+	output logic				xnanE;		// Input1E is NaN
-	output logic				ynanE;		// ReadData2E is NaN
+	output logic				ynanE;		// Input2E is NaN
-	output logic				znanE;		// ReadData3E is NaN
+	output logic				znanE;		// Input3E is NaN
-	output logic				xdenormE;	// ReadData1E is denormalized
+	output logic				xdenormE;	// Input1E is denormalized
-	output logic				ydenormE;	// ReadData2E is denormalized
+	output logic				ydenormE;	// Input2E is denormalized
-	output logic				zdenormE;	// ReadData3E is denormalized
+	output logic				zdenormE;	// Input3E is denormalized
-	output logic				xinfE;		// ReadData1E is infinity
+	output logic				xinfE;		// Input1E is infinity
-	output logic				yinfE;		// ReadData2E is infinity
+	output logic				yinfE;		// Input2E is infinity
-	output logic				zinfE;		// ReadData3E is infinity
+	output logic				zinfE;		// Input3E is infinity
 
 	// In the actual circuit design, the gates looking at bits
 	// 51:0 and at bits 62:52 should be shared among the various detectors.
 
 	// Check if input is NaN
 
-	assign xnanE = &ReadData1E[62:52] && |ReadData1E[51:0]; 
+	assign xnanE = &Input1E[62:52] && |Input1E[51:0]; 
-	assign ynanE = &ReadData2E[62:52] && |ReadData2E[51:0]; 
+	assign ynanE = &Input2E[62:52] && |Input2E[51:0]; 
-	assign znanE = &ReadData3E[62:52] && |ReadData3E[51:0];
+	assign znanE = &Input3E[62:52] && |Input3E[51:0];
 
 	// Check if input is denormalized
 
-	assign xdenormE = ~(|ReadData1E[62:52]) && |ReadData1E[51:0]; 
+	assign xdenormE = ~(|Input1E[62:52]) && |Input1E[51:0]; 
-	assign ydenormE = ~(|ReadData2E[62:52]) && |ReadData2E[51:0]; 
+	assign ydenormE = ~(|Input2E[62:52]) && |Input2E[51:0]; 
-	assign zdenormE = ~(|ReadData3E[62:52]) && |ReadData3E[51:0];
+	assign zdenormE = ~(|Input3E[62:52]) && |Input3E[51:0];
 
 	// Check if input is infinity
 
-	assign xinfE = &ReadData1E[62:52] && ~(|ReadData1E[51:0]); 
+	assign xinfE = &Input1E[62:52] && ~(|Input1E[51:0]); 
-	assign yinfE = &ReadData2E[62:52] && ~(|ReadData2E[51:0]); 
+	assign yinfE = &Input2E[62:52] && ~(|Input2E[51:0]); 
-	assign zinfE = &ReadData3E[62:52] && ~(|ReadData3E[51:0]);
+	assign zinfE = &Input3E[62:52] && ~(|Input3E[51:0]);
 
 	// Check if inputs are all zero
 	// Also forces denormalized inputs to zero.
@@ -57,11 +57,11 @@ module special(ReadData1E, ReadData2E, ReadData3E, xzeroE, yzeroE, zzeroE,
 	// to just check if the exponent is zero.
 	
 	// KATHERINE - commented following (21/01/11)
-	// assign xzeroE = ~(|ReadData1E[62:0]) || xdenormE;
+	// assign xzeroE = ~(|Input1E[62:0]) || xdenormE;
-	// assign yzeroE = ~(|ReadData2E[62:0]) || ydenormE;
+	// assign yzeroE = ~(|Input2E[62:0]) || ydenormE;
-	// assign zzeroE = ~(|ReadData3E[62:0]) || zdenormE;
+	// assign zzeroE = ~(|Input3E[62:0]) || zdenormE;
 	// KATHERINE - removed denorm to prevent output logicing zero when computing with a denormalized number
-	assign xzeroE = ~(|ReadData1E[62:0]);
+	assign xzeroE = ~(|Input1E[62:0]);
-	assign yzeroE = ~(|ReadData2E[62:0]);
+	assign yzeroE = ~(|Input2E[62:0]);
-	assign zzeroE = ~(|ReadData3E[62:0]);
+	assign zzeroE = ~(|Input3E[62:0]);
  endmodule

wally-pipelined/src/hazard/hazard.sv

@@ -32,6 +32,7 @@ module hazard(
 	      input logic  BPPredWrongE, CSRWritePendingDEM, RetM, TrapM,
 	      input logic  LoadStallD, MulDivStallD, CSRRdStallD,
 	      input logic  DataStall, ICacheStallF,
+        input logic  FStallE,
 	      input logic  DivBusyE,
   // Stall & flush outputs
 	      output logic StallF, StallD, StallE, StallM, StallW,
@@ -60,7 +61,7 @@ module hazard(
   assign StallFCause = CSRWritePendingDEM & ~(BranchFlushDE);
   assign StallDCause = (LoadStallD | MulDivStallD | CSRRdStallD) & ~(BranchFlushDE);    // stall in decode if instruction is a load/mul/csr dependent on previous
 //  assign StallDCause = LoadStallD | MulDivStallD | CSRRdStallD;    // stall in decode if instruction is a load/mul/csr dependent on previous
-  assign StallECause = DivBusyE;
+  assign StallECause = DivBusyE | FStallE;
   assign StallMCause = 0; 
   assign StallWCause = DataStall | ICacheStallF;
 

wally-pipelined/src/ieu/controller.sv

@@ -33,6 +33,7 @@ module controller(
   output logic [2:0] ImmSrcD,
   input  logic       IllegalIEUInstrFaultD, 
   output logic       IllegalBaseInstrFaultD,
+  output logic       RegWriteD,
   // Execute stage control signals
   input logic 	     StallE, FlushE, 
   input logic  [2:0] FlagsE, 
@@ -68,7 +69,7 @@ module controller(
   `define CTRLW 23
 
   // pipelined control signals
-  logic 	    RegWriteD, RegWriteE;
+  logic 	    RegWriteE;
   logic [2:0] ResultSrcD, ResultSrcE, ResultSrcM;
   logic [1:0] MemRWD, MemRWE;
   logic		    JumpD;
@@ -105,6 +106,7 @@ module controller(
       // RegWrite_ImmSrc_ALUSrc_MemRW_ResultSrc_Branch_ALUOp_Jump_TargetSrc_W64_CSRRead_Privileged_MulDiv_Atomic_Illegal
         7'b0000000:   ControlsD = `CTRLW'b0_000_00_00_000_0_00_0_0_0_0_0_0_00_1; // illegal instruction
         7'b0000011:   ControlsD = `CTRLW'b1_000_01_10_001_0_00_0_0_0_0_0_0_00_0; // lw
+        7'b0000111:   ControlsD = `CTRLW'b0_000_01_10_001_0_00_0_0_0_0_0_0_00_0; // flw
         7'b0001111:   ControlsD = `CTRLW'b0_000_00_00_000_0_00_0_0_0_0_0_0_00_0; // fence = nop
         7'b0010011:   ControlsD = `CTRLW'b1_000_01_00_000_0_10_0_0_0_0_0_0_00_0; // I-type ALU
         7'b0010111:   ControlsD = `CTRLW'b1_100_11_00_000_0_00_0_0_0_0_0_0_00_0; // auipc
@@ -113,6 +115,7 @@ module controller(
                     else
                       ControlsD = `CTRLW'b0_000_00_00_000_0_00_0_0_0_0_0_0_00_1; // non-implemented instruction
         7'b0100011:   ControlsD = `CTRLW'b0_001_01_01_000_0_00_0_0_0_0_0_0_00_0; // sw
+        7'b0100111:   ControlsD = `CTRLW'b0_001_01_01_000_0_00_0_0_0_0_0_0_00_0; // fsw
         7'b0101111: if (`A_SUPPORTED) begin
                       if (InstrD[31:27] == 5'b00010)
                         ControlsD = `CTRLW'b1_000_00_10_001_0_00_0_0_0_0_0_0_01_0; // lr

wally-pipelined/src/ieu/datapath.sv

@@ -48,6 +48,8 @@ module datapath (
   output logic [`XLEN-1:0] WriteDataM, MemAdrM,
   // Writeback stage signals
   input  logic             StallW, FlushW,
+  input  logic             FWriteIntW,
+  input  logic [`XLEN-1:0] FPUResultW,
   input  logic             RegWriteW, 
   input  logic             SquashSCW,
   input  logic [2:0]       ResultSrcW,
@@ -77,13 +79,18 @@ module datapath (
   // Writeback stage signals
   logic [`XLEN-1:0] SCResultW;
   logic [`XLEN-1:0] ALUResultW;
+  logic [`XLEN-1:0] WriteDataW;
   logic [`XLEN-1:0] ResultW;
 
   // Decode stage
   assign Rs1D      = InstrD[19:15];
   assign Rs2D      = InstrD[24:20];
   assign RdD       = InstrD[11:7];
-  regfile regf(clk, reset, RegWriteW, Rs1D, Rs2D, RdW, ResultW, RD1D, RD2D);
+
+  //Mux for writting floating point
+  mux2  #(`XLEN)  writedatamux(ResultW, FPUResultW, FWriteIntW, WriteDataW);  
+  
+  regfile regf(clk, reset, {RegWriteW | FWriteIntW}, Rs1D, Rs2D, RdW, WriteDataW, RD1D, RD2D);
   extend ext(.InstrD(InstrD[31:7]), .*);
  
   // Execute stage pipeline register and logic

wally-pipelined/src/ieu/ieu.sv

@@ -31,6 +31,7 @@ module ieu (
   input logic [31:0] 	   InstrD,
   input logic 		   IllegalIEUInstrFaultD, 
   output logic 		   IllegalBaseInstrFaultD,
+  output logic       RegWriteD,
   // Execute Stage interface
   input logic [`XLEN-1:0]  PCE, 
   input logic [`XLEN-1:0]  PCLinkE,
@@ -49,6 +50,8 @@ module ieu (
   output logic [2:0] 	   Funct3M,
   // Writeback stage
   input logic [`XLEN-1:0]  CSRReadValW, ReadDataW, MulDivResultW,
+  input logic             FWriteIntW,
+  input logic [`XLEN-1:0] FPUResultW,
   // input  logic [`XLEN-1:0] PCLinkW,
   output logic 		   InstrValidW,
   // hazards

wally-pipelined/src/privileged/csru.sv

@@ -44,7 +44,7 @@ module csru #(parameter
 
   // Floating Point CSRs in User Mode only needed if Floating Point is supported
   generate  
-    if (`F_SUPPORTED) begin
+    if (`F_SUPPORTED | `D_SUPPORTED) begin
       logic [4:0] FFLAGS_REGW;
       logic WriteFFLAGSM, WriteFRMM, WriteFCSRM;
       logic [2:0] NextFRMM;

wally-pipelined/src/wally/wallypipelinedhart.sv

@@ -93,7 +93,12 @@ module wallypipelinedhart (
   logic [4:0] SetFflagsM;
   logic [2:0] FRM_REGW;
   logic       FloatRegWriteW;
+  logic [1:0] FMemRWM;
+  logic       RegWriteD;
+  logic [`XLEN-1:0] FWriteDataM;
   logic       SquashSCW;
+  logic       FStallE;
+  logic       FWriteIntW;
   logic [31:0]      FSROutW;
   logic             DivSqrtDoneE;
   logic             IllegalFPUInstrD, IllegalFPUInstrE;
@@ -140,17 +145,23 @@ module wallypipelinedhart (
   logic 	    RASPredPCWrongM;
   logic 	    BPPredClassNonCFIWrongM;
 
+  logic[`XLEN-1:0] WriteDatatmpM;
+
   logic [4:0]       InstrClassM;
   
            
   ifu ifu(.InstrInF(InstrRData), .*); // instruction fetch unit: PC, branch prediction, instruction cache
 
   ieu ieu(.*); // integer execution unit: integer register file, datapath and controller
-  dmem dmem(.*); // data cache unit
+
+  
+  mux2  #(`XLEN)  OutputInput2mux(WriteDataM, FWriteDataM, FMemRWM[0], WriteDatatmpM);
+  dmem dmem(.MemRWM(MemRWM|FMemRWM), .WriteDataM(WriteDatatmpM),.*); // data cache unit
 
   ahblite ebu( 
     //.InstrReadF(1'b0),
     //.InstrRData(InstrF), // hook up InstrF later
+    .WriteDataM(WriteDatatmpM),
     .MemSizeM(Funct3M[1:0]), .UnsignedLoadM(Funct3M[2]),
     .Funct7M(InstrM[31:25]),
     .*);

wally-pipelined/testbench/testbench-imperas.sv

@@ -120,38 +120,38 @@ string tests32f[] = '{
 
 
   string tests64d[] = '{
-    "rv64d/I-FMV-X-D-01", "2000",
-    "rv64d/I-FADD-D-01", "2000",
-    "rv64d/I-FCLASS-D-01", "2000",
-    "rv64d/I-FCVT-D-L-01", "2000",
-    "rv64d/I-FCVT-D-LU-01", "2000",
-    "rv64d/I-FCVT-D-S-01", "2000",
-    "rv64d/I-FCVT-D-W-01", "2000",
-    "rv64d/I-FCVT-D-WU-01", "2000",
-    "rv64d/I-FCVT-L-D-01", "2000",
-    "rv64d/I-FCVT-LU-D-01", "2000",
-    "rv64d/I-FCVT-S-D-01", "2000",
-    "rv64d/I-FCVT-W-D-01", "2000",
-    "rv64d/I-FCVT-WU-D-01", "2000",
-    "rv64d/I-FDIV-D-01", "2000",
-    "rv64d/I-FEQ-D-01", "2000",
-    "rv64d/I-FLD-D-01", "2000",
-    "rv64d/I-FLE-D-01", "2000",
-    "rv64d/I-FLT-D-01", "2000",
-    "rv64d/I-FMADD-D-01", "2000",
-    "rv64d/I-FMAX-D-01", "2000",
-    "rv64d/I-FMIN-D-01", "2000",
-    "rv64d/I-FMSUB-D-01", "2000",
-    "rv64d/I-FMUL-D-01", "2000",
     "rv64d/I-FMV-D-X-01", "2000",
-    "rv64d/I-FNMADD-D-01", "2000",
+    // "rv64d/I-FADD-D-01", "2000",
-    "rv64d/I-FNMSUB-D-01", "2000",
+    // "rv64d/I-FCLASS-D-01", "2000",
-    "rv64d/I-FSD-01", "2000",
+    // "rv64d/I-FCVT-D-L-01", "2000",
-    "rv64d/I-FSGNJ-D-01", "2000",
+    // "rv64d/I-FCVT-D-LU-01", "2000",
-    "rv64d/I-FSGNJN-D-01", "2000",
+    // "rv64d/I-FCVT-D-S-01", "2000",
-    "rv64d/I-FSGNJX-D-01", "2000",
+    // "rv64d/I-FCVT-D-W-01", "2000",
-    "rv64d/I-FSQRTD-01", "2000",
+    // "rv64d/I-FCVT-D-WU-01", "2000",
-    "rv64d/I-FSUB-D-01", "2000"
+    // "rv64d/I-FCVT-L-D-01", "2000",
+    // "rv64d/I-FCVT-LU-D-01", "2000",
+    // "rv64d/I-FCVT-S-D-01", "2000",
+    // "rv64d/I-FCVT-W-D-01", "2000",
+    // "rv64d/I-FCVT-WU-D-01", "2000",
+    // "rv64d/I-FDIV-D-01", "2000",
+    // "rv64d/I-FEQ-D-01", "2000",
+    // "rv64d/I-FLD-D-01", "2000",
+    // "rv64d/I-FLE-D-01", "2000",
+    // "rv64d/I-FLT-D-01", "2000",
+    // "rv64d/I-FMADD-D-01", "2000",
+    // "rv64d/I-FMAX-D-01", "2000",
+    // "rv64d/I-FMIN-D-01", "2000",
+    // "rv64d/I-FMSUB-D-01", "2000",
+    // "rv64d/I-FMUL-D-01", "2000",
+    // "rv64d/I-FMV-X-D-01", "2000",
+    // "rv64d/I-FNMADD-D-01", "2000",
+    // "rv64d/I-FNMSUB-D-01", "2000",
+    // "rv64d/I-FSD-01", "2000",
+    // "rv64d/I-FSGNJ-D-01", "2000",
+    // "rv64d/I-FSGNJN-D-01", "2000",
+    // "rv64d/I-FSGNJX-D-01", "2000",
+    // "rv64d/I-FSQRTD-01", "2000",
+    // "rv64d/I-FSUB-D-01", "2000"
   };
 
 
@@ -528,7 +528,7 @@ string tests32f[] = '{
         if (`M_SUPPORTED) tests = {tests, tests64m};
         if (`A_SUPPORTED) tests = {tests, tests64a};
         if (`MEM_VIRTMEM) tests = {tests, tests64mmu};
-        if (`F_SUPPORTED) tests = {tests64f, tests};
+        // if (`F_SUPPORTED) tests = {tests64f, tests};
         if (`D_SUPPORTED) tests = {tests64d, tests};
       end
       //tests = {tests64a, tests};
@@ -655,7 +655,7 @@ string tests32f[] = '{
               errors = errors+1;
               $display("  Error on test %s result %d: adr = %h sim = %h, signature = %h", 
                     tests[test], i, (testadr+i)*`XLEN/8, dut.uncore.dtim.RAM[testadr+i], signature[i]);
-              // $stop;//***debug
+              $stop;//***debug
             end
           end
           i = i + 1;
@@ -923,8 +923,8 @@ module instrNameDecTB(
                        else                              name = "ILLEGAL";
       10'b0000111_010: name = "FLW";
       10'b0100111_010: name = "FSW";
-      10'b0000111_010: name = "FLD";
+      10'b0000111_011: name = "FLD";
-      10'b0100111_010: name = "FSD";
+      10'b0100111_011: name = "FSD";
       default:         name = "ILLEGAL";
     endcase
 endmodule