translate_arm/load_store: Handle unpredictable instructions

This necessated handling literal versions of the instructions separately
as they had different requirements. The rationale for detecting
unpredictable instructions is because:

a. they are unlikely to be outputted by a well-behaved compiler
b. their behaviour may change between different processors

I would rather unpredictable instructions fail loudly than silently do
approximately the right thing.

src/frontend/decoder/arm.h

@@ -185,42 +185,48 @@ std::vector<ArmMatcher<V>> GetArmDecodeTable() {
         INST(&V::arm_SWPB,        "SWPB",                "cccc00010100nnnntttt00001001uuuu"), // v2S (v6: Deprecated)
 
         // Load/Store instructions
-        INST(&V::arm_LDR_imm,     "LDR (imm)",           "cccc010pu0w1nnnnddddvvvvvvvvvvvv"),
-        INST(&V::arm_LDR_reg,     "LDR (reg)",           "cccc011pu0w1nnnnddddvvvvvrr0mmmm"),
-        INST(&V::arm_LDRB_imm,    "LDRB (imm)",          "cccc010pu1w1nnnnddddvvvvvvvvvvvv"),
-        INST(&V::arm_LDRB_reg,    "LDRB (reg)",          "cccc011pu1w1nnnnddddvvvvvrr0mmmm"),
         INST(&V::arm_LDRBT,       "LDRBT (A1)",          "----0100-111--------------------"),
         INST(&V::arm_LDRBT,       "LDRBT (A2)",          "----0110-111---------------0----"),
-        INST(&V::arm_LDRD_imm,    "LDRD (imm)",          "cccc000pu1w0nnnnddddvvvv1101vvvv"), // v5E
-        INST(&V::arm_LDRD_reg,    "LDRD (reg)",          "cccc000pu0w0nnnndddd00001101mmmm"), // v5E
-        INST(&V::arm_LDRH_imm,    "LDRH (imm)",          "cccc000pu1w1nnnnddddvvvv1011vvvv"),
-        INST(&V::arm_LDRH_reg,    "LDRH (reg)",          "cccc000pu0w1nnnndddd00001011mmmm"),
         INST(&V::arm_LDRHT,       "LDRHT (A1)",          "----0000-111------------1011----"),
         INST(&V::arm_LDRHT,       "LDRHT (A2)",          "----0000-011--------00001011----"),
-        INST(&V::arm_LDRSB_imm,   "LDRSB (imm)",         "cccc000pu1w1nnnnddddvvvv1101vvvv"),
-        INST(&V::arm_LDRSB_reg,   "LDRSB (reg)",         "cccc000pu0w1nnnndddd00001101mmmm"),
         INST(&V::arm_LDRSBT,      "LDRSBT (A1)",         "----0000-111------------1101----"),
         INST(&V::arm_LDRSBT,      "LDRSBT (A2)",         "----0000-011--------00001101----"),
-        INST(&V::arm_LDRSH_imm,   "LDRSH (imm)",         "cccc000pu1w1nnnnddddvvvv1111vvvv"),
-        INST(&V::arm_LDRSH_reg,   "LDRSH (reg)",         "cccc000pu0w1nnnndddd00001111mmmm"),
         INST(&V::arm_LDRSHT,      "LDRSHT (A1)",         "----0000-111------------1111----"),
         INST(&V::arm_LDRSHT,      "LDRSHT (A2)",         "----0000-011--------00001111----"),
         INST(&V::arm_LDRT,        "LDRT (A1)",           "----0100-011--------------------"),
         INST(&V::arm_LDRT,        "LDRT (A2)",           "----0110-011---------------0----"),
-        INST(&V::arm_STR_imm,     "STR (imm)",           "cccc010pu0w0nnnnddddvvvvvvvvvvvv"),
-        INST(&V::arm_STR_reg,     "STR (reg)",           "cccc011pu0w0nnnnddddvvvvvrr0mmmm"),
-        INST(&V::arm_STRB_imm,    "STRB (imm)",          "cccc010pu1w0nnnnddddvvvvvvvvvvvv"),
-        INST(&V::arm_STRB_reg,    "STRB (reg)",          "cccc011pu1w0nnnnddddvvvvvrr0mmmm"),
         INST(&V::arm_STRBT,       "STRBT (A1)",          "----0100-110--------------------"),
         INST(&V::arm_STRBT,       "STRBT (A2)",          "----0110-110---------------0----"),
-        INST(&V::arm_STRD_imm,    "STRD (imm)",          "cccc000pu1w0nnnnddddvvvv1111vvvv"), // v5E
-        INST(&V::arm_STRD_reg,    "STRD (reg)",          "cccc000pu0w0nnnndddd00001111mmmm"), // v5E
-        INST(&V::arm_STRH_imm,    "STRH (imm)",          "cccc000pu1w0nnnnddddvvvv1011vvvv"),
-        INST(&V::arm_STRH_reg,    "STRH (reg)",          "cccc000pu0w0nnnndddd00001011mmmm"),
         INST(&V::arm_STRHT,       "STRHT (A1)",          "----0000-110------------1011----"),
         INST(&V::arm_STRHT,       "STRHT (A2)",          "----0000-010--------00001011----"),
         INST(&V::arm_STRT,        "STRT (A1)",           "----0100-010--------------------"),
         INST(&V::arm_STRT,        "STRT (A2)",           "----0110-010---------------0----"),
+        INST(&V::arm_LDR_lit,     "LDR (lit)",           "cccc0101u0011111ttttvvvvvvvvvvvv"),
+        INST(&V::arm_LDR_imm,     "LDR (imm)",           "cccc010pu0w1nnnnttttvvvvvvvvvvvv"),
+        INST(&V::arm_LDR_reg,     "LDR (reg)",           "cccc011pu0w1nnnnttttvvvvvrr0mmmm"),
+        INST(&V::arm_LDRB_lit,    "LDRB (lit)",          "cccc0101u1011111ttttvvvvvvvvvvvv"),
+        INST(&V::arm_LDRB_imm,    "LDRB (imm)",          "cccc010pu1w1nnnnttttvvvvvvvvvvvv"),
+        INST(&V::arm_LDRB_reg,    "LDRB (reg)",          "cccc011pu1w1nnnnttttvvvvvrr0mmmm"),
+        INST(&V::arm_LDRD_lit,    "LDRD (lit)",          "cccc0001u1001111ttttvvvv1101vvvv"),
+        INST(&V::arm_LDRD_imm,    "LDRD (imm)",          "cccc000pu1w0nnnnttttvvvv1101vvvv"), // v5E
+        INST(&V::arm_LDRD_reg,    "LDRD (reg)",          "cccc000pu0w0nnnntttt00001101mmmm"), // v5E
+        INST(&V::arm_LDRH_lit,    "LDRH (lit)",          "cccc000pu1w11111ttttvvvv1011vvvv"),
+        INST(&V::arm_LDRH_imm,    "LDRH (imm)",          "cccc000pu1w1nnnnttttvvvv1011vvvv"),
+        INST(&V::arm_LDRH_reg,    "LDRH (reg)",          "cccc000pu0w1nnnntttt00001011mmmm"),
+        INST(&V::arm_LDRSB_lit,   "LDRSB (lit)",         "cccc0001u1011111ttttvvvv1101vvvv"),
+        INST(&V::arm_LDRSB_imm,   "LDRSB (imm)",         "cccc000pu1w1nnnnttttvvvv1101vvvv"),
+        INST(&V::arm_LDRSB_reg,   "LDRSB (reg)",         "cccc000pu0w1nnnntttt00001101mmmm"),
+        INST(&V::arm_LDRSH_lit,   "LDRSH (lit)",         "cccc0001u1011111ttttvvvv1111vvvv"),
+        INST(&V::arm_LDRSH_imm,   "LDRSH (imm)",         "cccc000pu1w1nnnnttttvvvv1111vvvv"),
+        INST(&V::arm_LDRSH_reg,   "LDRSH (reg)",         "cccc000pu0w1nnnntttt00001111mmmm"),
+        INST(&V::arm_STR_imm,     "STR (imm)",           "cccc010pu0w0nnnnttttvvvvvvvvvvvv"),
+        INST(&V::arm_STR_reg,     "STR (reg)",           "cccc011pu0w0nnnnttttvvvvvrr0mmmm"),
+        INST(&V::arm_STRB_imm,    "STRB (imm)",          "cccc010pu1w0nnnnttttvvvvvvvvvvvv"),
+        INST(&V::arm_STRB_reg,    "STRB (reg)",          "cccc011pu1w0nnnnttttvvvvvrr0mmmm"),
+        INST(&V::arm_STRD_imm,    "STRD (imm)",          "cccc000pu1w0nnnnttttvvvv1111vvvv"), // v5E
+        INST(&V::arm_STRD_reg,    "STRD (reg)",          "cccc000pu0w0nnnntttt00001111mmmm"), // v5E
+        INST(&V::arm_STRH_imm,    "STRH (imm)",          "cccc000pu1w0nnnnttttvvvv1011vvvv"),
+        INST(&V::arm_STRH_reg,    "STRH (reg)",          "cccc000pu0w0nnnntttt00001011mmmm"),
 
         // Load/Store Multiple instructions
         INST(&V::arm_LDM,         "LDM",                 "cccc100010w1nnnnxxxxxxxxxxxxxxxx"), // all

src/frontend/disassembler/disassembler_arm.cpp

@@ -337,6 +337,10 @@ public:
     std::string arm_YIELD() { return "yield <unimplemented>"; }
 
     // Load/Store instructions
+    std::string arm_LDR_lit(Cond cond, bool U, Reg t, Imm12 imm12) {
+        bool P = true, W = false;
+        return arm_LDR_imm(cond, P, U, W, Reg::PC, t, imm12);
+    }
     std::string arm_LDR_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm12 imm12) {
         if (P) {
             return Common::StringFromFormat("ldr%s %s, [%s, #%c%u]%s", CondToString(cond), RegToString(t), RegToString(n), U ? '+' : '-', imm12, W ? "!" : "");
@@ -351,6 +355,10 @@ public:
             return Common::StringFromFormat("ldr%s %s, [%s], %c%s%s%s", CondToString(cond), RegToString(t), RegToString(n), U ? '+' : '-', RegToString(m), ShiftStr(shift, imm5).c_str(), W ? " (err: W == 1!!!)" : "");
         }
     }
+    std::string arm_LDRB_lit(Cond cond, bool U, Reg t, Imm12 imm12) {
+        bool P = true, W = false;
+        return arm_LDRB_imm(cond, P, U, W, Reg::PC, t, imm12);
+    }
     std::string arm_LDRB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm12 imm12) {
         if (P) {
             return Common::StringFromFormat("ldrb%s %s, [%s, #%c%u]%s", CondToString(cond), RegToString(t), RegToString(n), U ? '+' : '-', imm12, W ? "!" : "");
@@ -366,6 +374,10 @@ public:
         }
     }
     std::string arm_LDRBT() { return "ice"; }
+    std::string arm_LDRD_lit(Cond cond, bool U, Reg t, Imm4 imm8a, Imm4 imm8b) {
+        bool P = true, W = false;
+        return arm_LDRD_imm(cond, P, U, W, Reg::PC, t, imm8a, imm8b);
+    }
     std::string arm_LDRD_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b) {
         u32 imm32 = (imm8a << 4) | imm8b;
         if (P) {
@@ -381,6 +393,9 @@ public:
             return Common::StringFromFormat("ldrd%s %s, %s, [%s], %c%s%s", CondToString(cond), RegToString(t), RegToString(t+1), RegToString(n), U ? '+' : '-', RegToString(m), W ? " (err: W == 1!!!)" : "");
         }
     }
+    std::string arm_LDRH_lit(Cond cond, bool P, bool U, bool W, Reg t, Imm4 imm8a, Imm4 imm8b) {
+        return arm_LDRH_imm(cond, P, U, W, Reg::PC, t, imm8a, imm8b);
+    }
     std::string arm_LDRH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b) {
         u32 imm32 = (imm8a << 4) | imm8b;
         if (P) {
@@ -397,6 +412,10 @@ public:
         }
     }
     std::string arm_LDRHT() { return "ice"; }
+    std::string arm_LDRSB_lit(Cond cond, bool U, Reg t, Imm4 imm8a, Imm4 imm8b) {
+        bool P = true, W = false;
+        return arm_LDRSB_imm(cond, P, U, W, Reg::PC, t, imm8a, imm8b);
+    }
     std::string arm_LDRSB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b) {
         u32 imm32 = (imm8a << 4) | imm8b;
         if (P) {
@@ -413,6 +432,10 @@ public:
         }
     }
     std::string arm_LDRSBT() { return "ice"; }
+    std::string arm_LDRSH_lit(Cond cond, bool U, Reg t, Imm4 imm8a, Imm4 imm8b) {
+        bool P = true, W = false;
+        return arm_LDRSH_imm(cond, P, U, W, Reg::PC, t, imm8a, imm8b);
+    }
     std::string arm_LDRSH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b) {
         u32 imm32 = (imm8a << 4) | imm8b;
         if (P) {

src/frontend/translate/translate_arm/load_store.cpp

@@ -9,6 +9,38 @@
 namespace Dynarmic {
 namespace Arm {
 
+bool ArmTranslatorVisitor::arm_LDRBT() {
+    ASSERT_MSG(false, "System instructions unimplemented");
+}
+
+bool ArmTranslatorVisitor::arm_LDRHT() {
+    ASSERT_MSG(false, "System instructions unimplemented");
+}
+
+bool ArmTranslatorVisitor::arm_LDRSBT() {
+    ASSERT_MSG(false, "System instructions unimplemented");
+}
+
+bool ArmTranslatorVisitor::arm_LDRSHT() {
+    ASSERT_MSG(false, "System instructions unimplemented");
+}
+
+bool ArmTranslatorVisitor::arm_LDRT() {
+    ASSERT_MSG(false, "System instructions unimplemented");
+}
+
+bool ArmTranslatorVisitor::arm_STRBT() {
+    ASSERT_MSG(false, "System instructions unimplemented");
+}
+
+bool ArmTranslatorVisitor::arm_STRHT() {
+    ASSERT_MSG(false, "System instructions unimplemented");
+}
+
+bool ArmTranslatorVisitor::arm_STRT() {
+    ASSERT_MSG(false, "System instructions unimplemented");
+}
+
 static IR::Value GetAddressingMode(IREmitter& ir, bool P, bool U, bool W, Reg n, IR::Value index) {
     IR::Value address;
     if (P) {
@@ -38,11 +70,35 @@ static IR::Value GetAddressingMode(IREmitter& ir, bool P, bool U, bool W, Reg n,
     return address;
 }
 
-bool ArmTranslatorVisitor::arm_LDR_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm12 imm12) {
+bool ArmTranslatorVisitor::arm_LDR_lit(Cond cond, bool U, Reg t, Imm12 imm12) {
+    bool P = true, W = false;
+
+    if (ConditionPassed(cond)) {
+        const auto data = ir.ReadMemory32(GetAddressingMode(ir, P, U, W, Reg::PC, ir.Imm32(imm12)));
+
+        if (t == Reg::PC) {
+            ir.BXWritePC(data);
+            ir.SetTerm(IR::Term::ReturnToDispatch{});
+            return false;
+        }
+
+        ir.SetRegister(t, data);
+    }
+
+    return true;
+}
+
+bool ArmTranslatorVisitor::arm_LDR_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm12 imm12) {
+    if (n == Reg::PC)
+        return UnpredictableInstruction();
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if ((!P || W) && n == t)
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto data = ir.ReadMemory32(GetAddressingMode(ir, P, U, W, n, ir.Imm32(imm12)));
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.BXWritePC(data);
             if (!P && W && n == Reg::R13)
                 ir.SetTerm(IR::Term::PopRSBHint{});
@@ -51,74 +107,167 @@ bool ArmTranslatorVisitor::arm_LDR_imm(Cond cond, bool P, bool U, bool W, Reg n,
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDR_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm5 imm5, ShiftType shift, Reg m) {
+bool ArmTranslatorVisitor::arm_LDR_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm5 imm5, ShiftType shift, Reg m) {
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if (m == Reg::PC)
+        return UnpredictableInstruction();
+    if ((!P || W) && (n == Reg::PC || n == t))
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto shifted = EmitImmShift(ir.GetRegister(m), shift, imm5, ir.GetCFlag());
         const auto data = ir.ReadMemory32(GetAddressingMode(ir, P, U, W, n, shifted.result));
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.BXWritePC(data);
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm12 imm12) {
+bool ArmTranslatorVisitor::arm_LDRB_lit(Cond cond, bool U, Reg t, Imm12 imm12) {
-    if (ConditionPassed(cond)) {
+    if (t == Reg::PC)
-        const auto data = ir.ZeroExtendByteToWord(ir.ReadMemory8(GetAddressingMode(ir, P, U, W, n, ir.Imm32(imm12))));
+        return UnpredictableInstruction();
 
-        if (d == Reg::PC) {
+    bool P = true, W = false;
+    if (ConditionPassed(cond)) {
+        const auto data = ir.ZeroExtendByteToWord(ir.ReadMemory8(GetAddressingMode(ir, P, U, W, Reg::PC, ir.Imm32(imm12))));
+
+        if (t == Reg::PC) {
             ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRB_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm5 imm5, ShiftType shift, Reg m) {
+bool ArmTranslatorVisitor::arm_LDRB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm12 imm12) {
+    if (n == Reg::PC)
+        return UnpredictableInstruction();
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if ((!P || W) && n == t)
+        return UnpredictableInstruction();
+    if (t == Reg::PC)
+        return UnpredictableInstruction();
+
+    if (ConditionPassed(cond)) {
+        const auto data = ir.ZeroExtendByteToWord(ir.ReadMemory8(GetAddressingMode(ir, P, U, W, n, ir.Imm32(imm12))));
+
+        if (t == Reg::PC) {
+            ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
+            ir.SetTerm(IR::Term::ReturnToDispatch{});
+            return false;
+        }
+
+        ir.SetRegister(t, data);
+    }
+
+    return true;
+}
+
+bool ArmTranslatorVisitor::arm_LDRB_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm5 imm5, ShiftType shift, Reg m) {
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if (t == Reg::PC || m == Reg::PC)
+        return UnpredictableInstruction();
+    if ((!P || W) && (n == Reg::PC || n == t))
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto shifted = EmitImmShift(ir.GetRegister(m), shift, imm5, ir.GetCFlag());
         const auto data = ir.ZeroExtendByteToWord(ir.ReadMemory8(GetAddressingMode(ir, P, U, W, n, shifted.result)));
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRBT() {
+bool ArmTranslatorVisitor::arm_LDRD_lit(Cond cond, bool U, Reg t, Imm4 imm8a, Imm4 imm8b) {
-    return InterpretThisInstruction();
+    if (RegNumber(t) % 2 == 1)
+        return UnpredictableInstruction();
+    if (t+1 == Reg::PC)
+        return UnpredictableInstruction();
+
+    bool P = true, W = false;
+    if (ConditionPassed(cond)) {
+        const auto address_a = GetAddressingMode(ir, P, U, W, Reg::PC, ir.Imm32(imm8a << 4 | imm8b));
+        const auto address_b = ir.Add(address_a, ir.Imm32(4));
+        auto data_a = ir.ReadMemory32(address_a);
+        auto data_b = ir.ReadMemory32(address_b);
+
+        switch (t) {
+            case Reg::PC:
+                data_a = ir.Add(data_a, ir.Imm32(4));
+                break;
+            case Reg::LR:
+                data_b = ir.Add(data_b, ir.Imm32(4));
+                break;
+            default:
+                break;
         }
 
-bool ArmTranslatorVisitor::arm_LDRD_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b) {
+        if (t == Reg::PC) {
+            ir.ALUWritePC(data_a);
+        } else {
+            ir.SetRegister(t, data_a);
+        }
+
+        const Reg reg_b = static_cast<Reg>(std::min(t + 1, Reg::R15));
+        if (reg_b == Reg::PC) {
+            ir.ALUWritePC(data_b);
+        } else {
+            ir.SetRegister(reg_b, data_b);
+        }
+
+        if (t == Reg::PC || reg_b == Reg::PC) {
+            ir.SetTerm(IR::Term::ReturnToDispatch{});
+            return false;
+        }
+    }
+
+    return true;
+}
+
+bool ArmTranslatorVisitor::arm_LDRD_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b) {
+    if (n == Reg::PC)
+        return UnpredictableInstruction();
+    if (RegNumber(t) % 2 == 1)
+        return UnpredictableInstruction();
+    if (!P && W)
+        return UnpredictableInstruction();
+    if ((!P || W) && (n == t || n == t+1))
+        return UnpredictableInstruction();
+    if (t+1 == Reg::PC)
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto address_a = GetAddressingMode(ir, P, U, W, n, ir.Imm32(imm8a << 4 | imm8b));
         const auto address_b = ir.Add(address_a, ir.Imm32(4));
         auto data_a = ir.ReadMemory32(address_a);
         auto data_b = ir.ReadMemory32(address_b);
 
-        switch (d) {
+        switch (t) {
         case Reg::PC:
             data_a = ir.Add(data_a, ir.Imm32(4));
             break;
@@ -129,20 +278,20 @@ bool ArmTranslatorVisitor::arm_LDRD_imm(Cond cond, bool P, bool U, bool W, Reg n
             break;
         }
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.ALUWritePC(data_a);
         } else {
-            ir.SetRegister(d, data_a);
+            ir.SetRegister(t, data_a);
         }
 
-        const Reg reg_b = static_cast<Reg>(std::min(d + 1, Reg::R15));
+        const Reg reg_b = static_cast<Reg>(std::min(t + 1, Reg::R15));
         if (reg_b == Reg::PC) {
             ir.ALUWritePC(data_b);
         } else {
             ir.SetRegister(reg_b, data_b);
         }
 
-        if (d == Reg::PC || reg_b == Reg::PC) {
+        if (t == Reg::PC || reg_b == Reg::PC) {
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
@@ -151,14 +300,23 @@ bool ArmTranslatorVisitor::arm_LDRD_imm(Cond cond, bool P, bool U, bool W, Reg n
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRD_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m) {
+bool ArmTranslatorVisitor::arm_LDRD_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Reg m) {
+    if (RegNumber(t) % 2 == 1)
+        return UnpredictableInstruction();
+    if (!P && W)
+        return UnpredictableInstruction();
+    if (t+1 == Reg::PC || m == Reg::PC || m == t || m == t+1)
+        return UnpredictableInstruction();
+    if ((!P || W) && (n == Reg::PC || n == t || n == t+1))
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto address_a = GetAddressingMode(ir, P, U, W, n, ir.GetRegister(m));
         const auto address_b = ir.Add(address_a, ir.Imm32(4));
         auto data_a = ir.ReadMemory32(address_a);
         auto data_b = ir.ReadMemory32(address_b);
 
-        switch (d) {
+        switch (t) {
         case Reg::PC:
             data_a = ir.Add(data_a, ir.Imm32(4));
             break;
@@ -169,20 +327,20 @@ bool ArmTranslatorVisitor::arm_LDRD_reg(Cond cond, bool P, bool U, bool W, Reg n
             break;
         }
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.ALUWritePC(data_a);
         } else {
-            ir.SetRegister(d, data_a);
+            ir.SetRegister(t, data_a);
         }
 
-        const Reg reg_b = static_cast<Reg>(std::min(d + 1, Reg::R15));
+        const Reg reg_b = static_cast<Reg>(std::min(t + 1, Reg::R15));
         if (reg_b == Reg::PC) {
             ir.ALUWritePC(data_b);
         } else {
             ir.SetRegister(reg_b, data_b);
         }
 
-        if (d == Reg::PC || reg_b == Reg::PC) {
+        if (t == Reg::PC || reg_b == Reg::PC) {
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
@@ -191,118 +349,206 @@ bool ArmTranslatorVisitor::arm_LDRD_reg(Cond cond, bool P, bool U, bool W, Reg n
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b) {
+bool ArmTranslatorVisitor::arm_LDRH_lit(Cond cond, bool P, bool U, bool W, Reg t, Imm4 imm8a, Imm4 imm8b) {
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if (P == W)
+        return UnpredictableInstruction();
+    if (t == Reg::PC)
+        return UnpredictableInstruction();
+
+    if (ConditionPassed(cond)) {
+        const auto data = ir.ZeroExtendHalfToWord(ir.ReadMemory16(GetAddressingMode(ir, P, U, W, Reg::PC, ir.Imm32(imm8a << 4 | imm8b))));
+
+        if (t == Reg::PC) {
+            ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
+            ir.SetTerm(IR::Term::ReturnToDispatch{});
+            return false;
+        }
+
+        ir.SetRegister(t, data);
+    }
+
+    return true;
+}
+
+bool ArmTranslatorVisitor::arm_LDRH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b) {
+    if (n == Reg::PC)
+        return UnpredictableInstruction();
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if ((!P || W) && n == t)
+        return UnpredictableInstruction();
+    if (t == Reg::PC)
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto data = ir.ZeroExtendHalfToWord(ir.ReadMemory16(GetAddressingMode(ir, P, U, W, n, ir.Imm32(imm8a << 4 | imm8b))));
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRH_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m) {
+bool ArmTranslatorVisitor::arm_LDRH_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Reg m) {
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if (t == Reg::PC || m == Reg::PC)
+        return UnpredictableInstruction();
+    if ((!P || W) && (n == Reg::PC || n == t))
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto data = ir.ZeroExtendHalfToWord(ir.ReadMemory16(GetAddressingMode(ir, P, U, W, n, ir.GetRegister(m))));
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRHT() {
+bool ArmTranslatorVisitor::arm_LDRSB_lit(Cond cond, bool U, Reg t, Imm4 imm8a, Imm4 imm8b) {
-    return InterpretThisInstruction();
+    if (t == Reg::PC)
+        return UnpredictableInstruction();
+
+    bool P = true, W = false;
+    if (ConditionPassed(cond)) {
+        const auto data = ir.SignExtendByteToWord(ir.ReadMemory8(GetAddressingMode(ir, P, U, W, Reg::PC, ir.Imm32(imm8a << 4 | imm8b))));
+
+        if (t == Reg::PC) {
+            ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
+            ir.SetTerm(IR::Term::ReturnToDispatch{});
+            return false;
         }
 
-bool ArmTranslatorVisitor::arm_LDRSB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b) {
+        ir.SetRegister(t, data);
+    }
+
+    return true;
+}
+
+bool ArmTranslatorVisitor::arm_LDRSB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b) {
+    if (n == Reg::PC)
+        return UnpredictableInstruction();
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if ((!P || W) && n == t)
+        return UnpredictableInstruction();
+    if (t == Reg::PC)
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto data = ir.SignExtendByteToWord(ir.ReadMemory8(GetAddressingMode(ir, P, U, W, n, ir.Imm32(imm8a << 4 | imm8b))));
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRSB_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m) {
+bool ArmTranslatorVisitor::arm_LDRSB_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Reg m) {
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if (t == Reg::PC || m == Reg::PC)
+        return UnpredictableInstruction();
+    if ((!P || W) && (n == Reg::PC || n == t))
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto data = ir.SignExtendByteToWord(ir.ReadMemory8(GetAddressingMode(ir, P, U, W, n, ir.GetRegister(m))));
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRSBT() {
+bool ArmTranslatorVisitor::arm_LDRSH_lit(Cond cond, bool U, Reg t, Imm4 imm8a, Imm4 imm8b) {
-    return InterpretThisInstruction();
+    if (t == Reg::PC)
+        return UnpredictableInstruction();
+
+    bool P = true, W = false;
+    if (ConditionPassed(cond)) {
+        const auto data = ir.SignExtendHalfToWord(ir.ReadMemory16(GetAddressingMode(ir, P, U, W, Reg::PC, ir.Imm32(imm8a << 4 | imm8b))));
+
+        if (t == Reg::PC) {
+            ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
+            ir.SetTerm(IR::Term::ReturnToDispatch{});
+            return false;
         }
 
-bool ArmTranslatorVisitor::arm_LDRSH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b) {
+        ir.SetRegister(t, data);
+    }
+
+    return true;
+}
+
+bool ArmTranslatorVisitor::arm_LDRSH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b) {
+    if (n == Reg::PC)
+        return UnpredictableInstruction();
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if ((!P || W) && n == t)
+        return UnpredictableInstruction();
+    if (t == Reg::PC)
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto data = ir.SignExtendHalfToWord(ir.ReadMemory16(GetAddressingMode(ir, P, U, W, n, ir.Imm32(imm8a << 4 | imm8b))));
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRSH_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m) {
+bool ArmTranslatorVisitor::arm_LDRSH_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Reg m) {
+    ASSERT_MSG(!(!P && W), "T form of instruction unimplemented");
+    if (t == Reg::PC || m == Reg::PC)
+        return UnpredictableInstruction();
+    if ((!P || W) && (n == Reg::PC || n == t))
+        return UnpredictableInstruction();
+
     if (ConditionPassed(cond)) {
         const auto data = ir.SignExtendHalfToWord(ir.ReadMemory16(GetAddressingMode(ir, P, U, W, n, ir.GetRegister(m))));
 
-        if (d == Reg::PC) {
+        if (t == Reg::PC) {
             ir.ALUWritePC(ir.Add(data, ir.Imm32(4)));
             ir.SetTerm(IR::Term::ReturnToDispatch{});
             return false;
         }
 
-        ir.SetRegister(d, data);
+        ir.SetRegister(t, data);
     }
 
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_LDRSHT() {
-    return InterpretThisInstruction();
-}
-
-bool ArmTranslatorVisitor::arm_LDRT() {
-    return InterpretThisInstruction();
-}
-
 bool ArmTranslatorVisitor::arm_STR_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm12 imm12) {
     if (ConditionPassed(cond)) {
         const auto address = GetAddressingMode(ir, P, U, W, n, ir.Imm32(imm12));
@@ -343,10 +589,6 @@ bool ArmTranslatorVisitor::arm_STRB_reg(Cond cond, bool P, bool U, bool W, Reg n
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_STRBT() {
-    return InterpretThisInstruction();
-}
-
 bool ArmTranslatorVisitor::arm_STRD_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b) {
     if (ConditionPassed(cond)) {
         const auto address_a = GetAddressingMode(ir, P, U, W, n, ir.Imm32(imm8a << 4 | imm8b));
@@ -395,14 +637,6 @@ bool ArmTranslatorVisitor::arm_STRH_reg(Cond cond, bool P, bool U, bool W, Reg n
     return true;
 }
 
-bool ArmTranslatorVisitor::arm_STRHT() {
-    return InterpretThisInstruction();
-}
-
-bool ArmTranslatorVisitor::arm_STRT() {
-    return InterpretThisInstruction();
-}
-
 static bool LDMHelper(IREmitter& ir, bool W, Reg n, RegList list, IR::Value start_address, IR::Value writeback_address) {
     auto address = start_address;
     for (size_t i = 0; i <= 14; i++) {

src/frontend/translate/translate_arm/translate_arm.h

@@ -155,35 +155,41 @@ struct ArmTranslatorVisitor final {
     bool arm_WFI() { return true; }
     bool arm_YIELD() { return true; }
 
-    // Load/Store instructions
+    // Load/Store
-    bool arm_LDR_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm12 imm12);
-    bool arm_LDR_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm5 imm5, ShiftType shift, Reg m);
-    bool arm_LDRB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm12 imm12);
-    bool arm_LDRB_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm5 imm5, ShiftType shift, Reg m);
     bool arm_LDRBT();
-    bool arm_LDRD_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b);
-    bool arm_LDRD_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m);
-    bool arm_LDRH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b);
-    bool arm_LDRH_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m);
     bool arm_LDRHT();
-    bool arm_LDRSB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b);
-    bool arm_LDRSB_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m);
     bool arm_LDRSBT();
-    bool arm_LDRSH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b);
-    bool arm_LDRSH_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m);
     bool arm_LDRSHT();
     bool arm_LDRT();
+    bool arm_STRBT();
+    bool arm_STRHT();
+    bool arm_STRT();
+    bool arm_LDR_lit(Cond cond, bool U, Reg t, Imm12 imm12);
+    bool arm_LDR_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm12 imm12);
+    bool arm_LDR_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm5 imm5, ShiftType shift, Reg m);
+    bool arm_LDRB_lit(Cond cond, bool U, Reg t, Imm12 imm12);
+    bool arm_LDRB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm12 imm12);
+    bool arm_LDRB_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm5 imm5, ShiftType shift, Reg m);
+    bool arm_LDRD_lit(Cond cond, bool U, Reg t, Imm4 imm8a, Imm4 imm8b);
+    bool arm_LDRD_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b);
+    bool arm_LDRD_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m);
+    bool arm_LDRH_lit(Cond cond, bool P, bool U, bool W, Reg t, Imm4 imm8a, Imm4 imm8b);
+    bool arm_LDRH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b);
+    bool arm_LDRH_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m);
+    bool arm_LDRSB_lit(Cond cond, bool U, Reg t, Imm4 imm8a, Imm4 imm8b);
+    bool arm_LDRSB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b);
+    bool arm_LDRSB_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Reg m);
+    bool arm_LDRSH_lit(Cond cond, bool U, Reg t, Imm4 imm8a, Imm4 imm8b);
+    bool arm_LDRSH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Imm4 imm8a, Imm4 imm8b);
+    bool arm_LDRSH_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg t, Reg m);
     bool arm_STR_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm12 imm12);
     bool arm_STR_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm5 imm5, ShiftType shift, Reg m);
     bool arm_STRB_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm12 imm12);
     bool arm_STRB_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm5 imm5, ShiftType shift, Reg m);
-    bool arm_STRBT();
     bool arm_STRD_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b);
     bool arm_STRD_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m);
     bool arm_STRH_imm(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Imm4 imm8a, Imm4 imm8b);
     bool arm_STRH_reg(Cond cond, bool P, bool U, bool W, Reg n, Reg d, Reg m);
-    bool arm_STRHT();
-    bool arm_STRT();
 
     // Load/Store multiple instructions
     bool arm_LDM(Cond cond, bool W, Reg n, RegList list);

tests/arm/fuzz_arm.cpp

@@ -643,7 +643,7 @@ TEST_CASE("Fuzz ARM load/store instructions (byte, half-word, word)", "[JitX64]"
             u32 rand = RandInt<u32>(0, 0xFF);
             u32 Rm = RandInt<u32>(0, 14);
 
-            if (W) {
+            if (!P || W) {
                 while (Rn == Rd) {
                     Rn = RandInt<u32>(0, 14);
                     Rd = RandInt<u32>(0, 14);