some tiny A64 optimisations

3 files changed, 67 insertions, 39 deletions

diff --git a/src/ARMJIT_A64/ARMJIT_ALU.cpp b/src/ARMJIT_A64/ARMJIT_ALU.cpp
index 6f1f99c..1bd427f 100644
--- a/src/ARMJIT_A64/ARMJIT_ALU.cpp
+++ b/src/ARMJIT_A64/ARMJIT_ALU.cpp
@@ -324,16 +324,29 @@ void Compiler::Comp_Arithmetic(int op, bool S, ARM64Reg rd, ARM64Reg rn, Op2 op2
         UBFX(W2, RCPSR, 29, 1);
         if (S)
         {
-            CVInGPR = true;
-            ADDS(W1, rn, W2);
-            CSET(W2, CC_CS);
-            CSET(W3, CC_VS);
             if (op2.IsImm)
-                ADDSI2R(rd, W1, op2.Imm, W0);
+            {
+                CVInGPR = true;
+                ADDS(W1, rn, W2);
+                CSET(W2, CC_CS);
+                CSET(W3, CC_VS);
+                if (op2.IsImm)
+                    ADDSI2R(rd, W1, op2.Imm, W0);
+                else
+                    ADDS(rd, W1, op2.Reg.Rm, op2.ToArithOption());
+                CSINC(W2, W2, WZR, CC_CC);
+                CSINC(W3, W3, WZR, CC_VC);
+            }
             else
-                ADDS(rd, W1, op2.Reg.Rm, op2.ToArithOption());
-            CSINC(W2, W2, WZR, CC_CC);
-            CSINC(W3, W3, WZR, CC_VC);
+            {
+                if (op2.Reg.ShiftAmount > 0)
+                {
+                    MOV(W0, op2.Reg.Rm, op2.ToArithOption());
+                    op2 = Op2(W0, ST_LSL, 0);
+                }
+                CMP(W2, 1);
+                ADCS(rd, rn, op2.Reg.Rm);
+            }
         }
         else
         {
@@ -346,25 +359,38 @@ void Compiler::Comp_Arithmetic(int op, bool S, ARM64Reg rd, ARM64Reg rn, Op2 op2
         break;
     case 0x6: // SBC
         UBFX(W2, RCPSR, 29, 1);
-        // W1 = -op2 - 1
-        if (op2.IsImm)
-            MOVI2R(W1, ~op2.Imm);
-        else
-            ORN(W1, WZR, op2.Reg.Rm, op2.ToArithOption());
-        if (S)
+        if (S && !op2.IsImm)
         {
-            CVInGPR = true;
-            ADDS(W1, W2, W1);
-            CSET(W2, CC_CS);
-            CSET(W3, CC_VS);
-            ADDS(rd, rn, W1);
-            CSINC(W2, W2, WZR, CC_CC);
-            CSINC(W3, W3, WZR, CC_VC);
+            if (op2.Reg.ShiftAmount > 0)
+            {
+                MOV(W0, op2.Reg.Rm, op2.ToArithOption());
+                op2 = Op2(W0, ST_LSL, 0);
+            }
+            CMP(W2, 1);
+            SBCS(rd, rn, op2.Reg.Rm);
         }
         else
         {
-            ADD(W1, W2, W1);
-            ADD(rd, rn, W1);
+            // W1 = -op2 - 1
+            if (op2.IsImm)
+                MOVI2R(W1, ~op2.Imm);
+            else
+                ORN(W1, WZR, op2.Reg.Rm, op2.ToArithOption());
+            if (S)
+            {
+                CVInGPR = true;
+                ADDS(W1, W2, W1);
+                CSET(W2, CC_CS);
+                CSET(W3, CC_VS);
+                ADDS(rd, rn, W1);
+                CSINC(W2, W2, WZR, CC_CC);
+                CSINC(W3, W3, WZR, CC_VC);
+            }
+            else
+            {
+                ADD(W1, W2, W1);
+                ADD(rd, rn, W1);
+            }
         }
         break;
     case 0x7: // RSC
@@ -533,21 +559,7 @@ void Compiler::A_Comp_ALUMovOp()
         }
         else
         {
-            // ORR with shifted operand has cycles latency
-            if (op2.Reg.ShiftAmount > 0)
-            {
-                switch (op2.Reg.ShiftType)
-                {
-                case ST_LSL: LSL(rd, op2.Reg.Rm, op2.Reg.ShiftAmount); break;
-                case ST_LSR: LSR(rd, op2.Reg.Rm, op2.Reg.ShiftAmount); break;
-                case ST_ASR: ASR(rd, op2.Reg.Rm, op2.Reg.ShiftAmount); break;
-                case ST_ROR: ROR(rd, op2.Reg.Rm, op2.Reg.ShiftAmount); break;
-                }
-            }
-            else
-            {
-                MOV(rd, op2.Reg.Rm, op2.ToArithOption());
-            }
+            MOV(rd, op2.Reg.Rm, op2.ToArithOption());
         }
     }
 
diff --git a/src/dolphin/Arm64Emitter.cpp b/src/dolphin/Arm64Emitter.cpp
index 408411c..47e97b1 100644
--- a/src/dolphin/Arm64Emitter.cpp
+++ b/src/dolphin/Arm64Emitter.cpp
@@ -1607,7 +1607,21 @@ void ARM64XEmitter::BICS(ARM64Reg Rd, ARM64Reg Rn, ARM64Reg Rm, ArithOption Shif
 
 void ARM64XEmitter::MOV(ARM64Reg Rd, ARM64Reg Rm, ArithOption Shift)
 {
-  ORR(Rd, Is64Bit(Rd) ? ZR : WZR, Rm, Shift);
+  if (Shift.GetType() == ArithOption::TYPE_SHIFTEDREG)
+  {
+    switch (Shift.GetShiftType())
+    {
+    case ST_LSL: LSL(Rd, Rm, Shift.GetShiftAmount()); break;
+    case ST_LSR: LSR(Rd, Rm, Shift.GetShiftAmount()); break;
+    case ST_ASR: ASR(Rd, Rm, Shift.GetShiftAmount()); break;
+    case ST_ROR: ROR(Rd, Rm, Shift.GetShiftAmount()); break;
+    default: ASSERT_MSG(DYNA_REC, false, "Invalid shift type"); break;
+    }
+  }
+  else
+  {
+    ORR(Rd, Is64Bit(Rd) ? ZR : WZR, Rm, Shift);
+  }
 }
 
 void ARM64XEmitter::MOV(ARM64Reg Rd, ARM64Reg Rm)
diff --git a/src/dolphin/Arm64Emitter.h b/src/dolphin/Arm64Emitter.h
index 3da3912..0b066de 100644
--- a/src/dolphin/Arm64Emitter.h
+++ b/src/dolphin/Arm64Emitter.h
@@ -469,6 +469,8 @@ public:
   }
   TypeSpecifier GetType() const { return m_type; }
   ARM64Reg GetReg() const { return m_destReg; }
+  ShiftType GetShiftType() const { return m_shifttype; }
+  u32 GetShiftAmount() const { return m_shift; }
   u32 GetData() const
   {
     switch (m_type)