A32/A64: Allow std::nullopt from MemoryReadCode

Raise a fault at runtime if this block is executed

src/dynarmic/frontend/A32/translate/translate_arm.cpp

@@ -28,20 +28,23 @@ IR::Block TranslateArm(LocationDescriptor descriptor, TranslateCallbacks* tcb, c
     bool should_continue = true;
     do {
         const u32 arm_pc = visitor.ir.current_location.PC();
-        const u32 arm_instruction = tcb->MemoryReadCode(arm_pc);
         visitor.current_instruction_size = 4;
 
+        if (const auto arm_instruction = tcb->MemoryReadCode(arm_pc)) {
             tcb->PreCodeTranslationHook(false, arm_pc, visitor.ir);
 
-        if (const auto vfp_decoder = DecodeVFP<TranslatorVisitor>(arm_instruction)) {
+            if (const auto vfp_decoder = DecodeVFP<TranslatorVisitor>(*arm_instruction)) {
-            should_continue = vfp_decoder->get().call(visitor, arm_instruction);
+                should_continue = vfp_decoder->get().call(visitor, *arm_instruction);
-        } else if (const auto asimd_decoder = DecodeASIMD<TranslatorVisitor>(arm_instruction)) {
+            } else if (const auto asimd_decoder = DecodeASIMD<TranslatorVisitor>(*arm_instruction)) {
-            should_continue = asimd_decoder->get().call(visitor, arm_instruction);
+                should_continue = asimd_decoder->get().call(visitor, *arm_instruction);
-        } else if (const auto decoder = DecodeArm<TranslatorVisitor>(arm_instruction)) {
+            } else if (const auto decoder = DecodeArm<TranslatorVisitor>(*arm_instruction)) {
-            should_continue = decoder->get().call(visitor, arm_instruction);
+                should_continue = decoder->get().call(visitor, *arm_instruction);
             } else {
                 should_continue = visitor.arm_UDF();
             }
+        } else {
+            should_continue = visitor.RaiseException(Exception::NoExecuteFault);
+        }
 
         if (visitor.cond_state == ConditionalState::Break) {
             break;

src/dynarmic/frontend/A32/translate/translate_callbacks.h

@@ -4,18 +4,19 @@
  */
 #pragma once
 
-#include <mcl/stdint.hpp>
+#include <cstdint>
+#include <optional>
 
 namespace Dynarmic::A32 {
 
-using VAddr = u32;
+using VAddr = std::uint32_t;
 
 class IREmitter;
 
 struct TranslateCallbacks {
     // All reads through this callback are 4-byte aligned.
     // Memory must be interpreted as little endian.
-    virtual std::uint32_t MemoryReadCode(VAddr vaddr) = 0;
+    virtual std::optional<std::uint32_t> MemoryReadCode(VAddr vaddr) = 0;
 
     // Thus function is called before the instruction at pc is interpreted.
     // IR code can be emitted by the callee prior to translation of the instruction.

src/dynarmic/frontend/A32/translate/translate_thumb.cpp

@@ -44,28 +44,40 @@ bool IsUnconditionalInstruction(bool is_thumb_16, u32 instruction) {
     return false;
 }
 
-std::tuple<u32, ThumbInstSize> ReadThumbInstruction(u32 arm_pc, TranslateCallbacks* tcb) {
+std::optional<std::tuple<u32, ThumbInstSize>> ReadThumbInstruction(u32 arm_pc, TranslateCallbacks* tcb) {
-    u32 first_part = tcb->MemoryReadCode(arm_pc & 0xFFFFFFFC);
+    u32 instruction;
-    if ((arm_pc & 0x2) != 0) {
-        first_part >>= 16;
-    }
-    first_part &= 0xFFFF;
 
-    if (IsThumb16(static_cast<u16>(first_part))) {
+    const std::optional<u32> first_part = tcb->MemoryReadCode(arm_pc & 0xFFFFFFFC);
+    if (!first_part)
+        return std::nullopt;
+
+    if ((arm_pc & 0x2) != 0) {
+        instruction = *first_part >> 16;
+    } else {
+        instruction = *first_part & 0xFFFF;
+    }
+
+    if (IsThumb16(static_cast<u16>(instruction))) {
         // 16-bit thumb instruction
-        return std::make_tuple(first_part, ThumbInstSize::Thumb16);
+        return std::make_tuple(instruction, ThumbInstSize::Thumb16);
     }
 
     // 32-bit thumb instruction
     // These always start with 0b11101, 0b11110 or 0b11111.
 
-    u32 second_part = tcb->MemoryReadCode((arm_pc + 2) & 0xFFFFFFFC);
+    instruction <<= 16;
-    if (((arm_pc + 2) & 0x2) != 0) {
-        second_part >>= 16;
-    }
-    second_part &= 0xFFFF;
 
-    return std::make_tuple(static_cast<u32>((first_part << 16) | second_part), ThumbInstSize::Thumb32);
+    const std::optional<u32> second_part = tcb->MemoryReadCode((arm_pc + 2) & 0xFFFFFFFC);
+    if (!second_part)
+        return std::nullopt;
+
+    if (((arm_pc + 2) & 0x2) != 0) {
+        instruction |= *second_part >> 16;
+    } else {
+        instruction |= *second_part & 0xFFFF;
+    }
+
+    return std::make_tuple(instruction, ThumbInstSize::Thumb32);
 }
 
 // Convert from thumb ASIMD format to ARM ASIMD format.
@@ -97,7 +109,8 @@ IR::Block TranslateThumb(LocationDescriptor descriptor, TranslateCallbacks* tcb,
     bool should_continue = true;
     do {
         const u32 arm_pc = visitor.ir.current_location.PC();
-        const auto [thumb_instruction, inst_size] = ReadThumbInstruction(arm_pc, tcb);
+        if (const auto maybe_instruction = ReadThumbInstruction(arm_pc, tcb)) {
+            const auto [thumb_instruction, inst_size] = *maybe_instruction;
             const bool is_thumb_16 = inst_size == ThumbInstSize::Thumb16;
             visitor.current_instruction_size = is_thumb_16 ? 2 : 4;
 
@@ -128,12 +141,16 @@ IR::Block TranslateThumb(LocationDescriptor descriptor, TranslateCallbacks* tcb,
                     }
                 }
             }
+        } else {
+            visitor.current_instruction_size = 2;
+            should_continue = visitor.RaiseException(Exception::NoExecuteFault);
+        }
 
         if (visitor.cond_state == ConditionalState::Break) {
             break;
         }
 
-        visitor.ir.current_location = visitor.ir.current_location.AdvancePC(is_thumb_16 ? 2 : 4).AdvanceIT();
+        visitor.ir.current_location = visitor.ir.current_location.AdvancePC(static_cast<int>(visitor.current_instruction_size)).AdvanceIT();
         block.CycleCount()++;
     } while (should_continue && CondCanContinue(visitor.cond_state, visitor.ir) && !single_step);
 

src/dynarmic/frontend/A64/translate/a64_translate.cpp

@@ -22,13 +22,16 @@ IR::Block Translate(LocationDescriptor descriptor, MemoryReadCodeFuncType memory
     bool should_continue = true;
     do {
         const u64 pc = visitor.ir.current_location->PC();
-        const u32 instruction = memory_read_code(pc);
 
-        if (auto decoder = Decode<TranslatorVisitor>(instruction)) {
+        if (const auto instruction = memory_read_code(pc)) {
-            should_continue = decoder->get().call(visitor, instruction);
+            if (auto decoder = Decode<TranslatorVisitor>(*instruction)) {
+                should_continue = decoder->get().call(visitor, *instruction);
             } else {
                 should_continue = visitor.InterpretThisInstruction();
             }
+        } else {
+            should_continue = visitor.RaiseException(Exception::NoExecuteFault);
+        }
 
         visitor.ir.current_location = visitor.ir.current_location->AdvancePC(4);
         block.CycleCount()++;

src/dynarmic/frontend/A64/translate/a64_translate.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <functional>
+#include <optional>
 
 #include <mcl/stdint.hpp>
 
@@ -18,7 +19,7 @@ namespace A64 {
 
 class LocationDescriptor;
 
-using MemoryReadCodeFuncType = std::function<u32(u64 vaddr)>;
+using MemoryReadCodeFuncType = std::function<std::optional<u32>(u64 vaddr)>;
 
 struct TranslationOptions {
     /// This changes what IR we emit when we translate an unpredictable instruction.

src/dynarmic/interface/A32/config.h

@@ -9,6 +9,7 @@
 #include <cstddef>
 #include <cstdint>
 #include <memory>
+#include <optional>
 
 #include "dynarmic/frontend/A32/translate/translate_callbacks.h"
 #include "dynarmic/interface/A32/arch_version.h"
@@ -51,6 +52,9 @@ enum class Exception {
     PreloadDataWithIntentToWrite,
     /// A PLI instruction was executed. (Hint instruction.)
     PreloadInstruction,
+    /// Attempted to execute a code block at an address for which MemoryReadCode returned std::nullopt.
+    /// (Intended to be used to emulate memory protection faults.)
+    NoExecuteFault,
 };
 
 /// These function pointers may be inserted into compiled code.
@@ -59,7 +63,7 @@ struct UserCallbacks : public TranslateCallbacks {
 
     // All reads through this callback are 4-byte aligned.
     // Memory must be interpreted as little endian.
-    std::uint32_t MemoryReadCode(VAddr vaddr) override { return MemoryRead32(vaddr); }
+    std::optional<std::uint32_t> MemoryReadCode(VAddr vaddr) override { return MemoryRead32(vaddr); }
 
     // Thus function is called before the instruction at pc is interpreted.
     // IR code can be emitted by the callee prior to translation of the instruction.

src/dynarmic/interface/A64/config.h

@@ -9,6 +9,7 @@
 #include <cstddef>
 #include <cstdint>
 #include <memory>
+#include <optional>
 
 #include "dynarmic/interface/optimization_flags.h"
 
@@ -45,6 +46,9 @@ enum class Exception {
     Yield,
     /// A BRK instruction was executed. (Hint instruction.)
     Breakpoint,
+    /// Attempted to execute a code block at an address for which MemoryReadCode returned std::nullopt.
+    /// (Intended to be used to emulate memory protection faults.)
+    NoExecuteFault,
 };
 
 enum class DataCacheOperation {
@@ -82,7 +86,7 @@ struct UserCallbacks {
 
     // All reads through this callback are 4-byte aligned.
     // Memory must be interpreted as little endian.
-    virtual std::uint32_t MemoryReadCode(VAddr vaddr) { return MemoryRead32(vaddr); }
+    virtual std::optional<std::uint32_t> MemoryReadCode(VAddr vaddr) { return MemoryRead32(vaddr); }
 
     // Reads through these callbacks may not be aligned.
     virtual std::uint8_t MemoryRead8(VAddr vaddr) = 0;

src/dynarmic/ir/opt/a64_merge_interpret_blocks.cpp

@@ -16,10 +16,12 @@ namespace Dynarmic::Optimization {
 
 void A64MergeInterpretBlocksPass(IR::Block& block, A64::UserCallbacks* cb) {
     const auto is_interpret_instruction = [cb](A64::LocationDescriptor location) {
-        const u32 instruction = cb->MemoryReadCode(location.PC());
+        const auto instruction = cb->MemoryReadCode(location.PC());
+        if (!instruction)
+            return false;
 
         IR::Block new_block{location};
-        A64::TranslateSingleInstruction(new_block, location, instruction);
+        A64::TranslateSingleInstruction(new_block, location, *instruction);
 
         if (!new_block.Instructions().empty())
             return false;

tests/A32/testenv.h

@@ -48,7 +48,7 @@ public:
         return vaddr < sizeof(InstructionType) * code_mem.size();
     }
 
-    std::uint32_t MemoryReadCode(u32 vaddr) override {
+    std::optional<std::uint32_t> MemoryReadCode(u32 vaddr) override {
         if (IsInCodeMem(vaddr)) {
             u32 value;
             std::memcpy(&value, &code_mem[vaddr / sizeof(InstructionType)], sizeof(u32));
@@ -95,11 +95,11 @@ public:
         MemoryWrite32(vaddr + 4, static_cast<u32>(value >> 32));
     }
 
-    void InterpreterFallback(u32 pc, size_t num_instructions) override { ASSERT_MSG(false, "InterpreterFallback({:08x}, {}) code = {:08x}", pc, num_instructions, MemoryReadCode(pc)); }
+    void InterpreterFallback(u32 pc, size_t num_instructions) override { ASSERT_MSG(false, "InterpreterFallback({:08x}, {}) code = {:08x}", pc, num_instructions, *MemoryReadCode(pc)); }
 
     void CallSVC(std::uint32_t swi) override { ASSERT_MSG(false, "CallSVC({})", swi); }
 
-    void ExceptionRaised(u32 pc, Dynarmic::A32::Exception /*exception*/) override { ASSERT_MSG(false, "ExceptionRaised({:08x}) code = {:08x}", pc, MemoryReadCode(pc)); }
+    void ExceptionRaised(u32 pc, Dynarmic::A32::Exception /*exception*/) override { ASSERT_MSG(false, "ExceptionRaised({:08x}) code = {:08x}", pc, *MemoryReadCode(pc)); }
 
     void AddTicks(std::uint64_t ticks) override {
         if (ticks > ticks_left) {
@@ -135,7 +135,7 @@ public:
         memcpy(backing_memory + vaddr, &value, sizeof(T));
     }
 
-    std::uint32_t MemoryReadCode(std::uint32_t vaddr) override {
+    std::optional<std::uint32_t> MemoryReadCode(std::uint32_t vaddr) override {
         return read<std::uint32_t>(vaddr);
     }
 

tests/A64/testenv.h

@@ -30,7 +30,7 @@ public:
         return vaddr >= code_mem_start_address && vaddr < code_mem_start_address + code_mem.size() * 4;
     }
 
-    std::uint32_t MemoryReadCode(u64 vaddr) override {
+    std::optional<std::uint32_t> MemoryReadCode(u64 vaddr) override {
         if (!IsInCodeMem(vaddr)) {
             return 0x14000000;  // B .
         }
@@ -145,7 +145,7 @@ public:
         memcpy(backing_memory + vaddr, &value, sizeof(T));
     }
 
-    std::uint32_t MemoryReadCode(u64 vaddr) override {
+    std::optional<std::uint32_t> MemoryReadCode(u64 vaddr) override {
         return read<std::uint32_t>(vaddr);
     }
 

tests/print_info.cpp

@@ -157,7 +157,7 @@ public:
     }
 
     void InterpreterFallback(u32 pc, size_t num_instructions) override {
-        fmt::print("> InterpreterFallback({:08x}, {}) code = {:08x}\n", pc, num_instructions, MemoryReadCode(pc));
+        fmt::print("> InterpreterFallback({:08x}, {}) code = {:08x}\n", pc, num_instructions, *MemoryReadCode(pc));
     }
     void CallSVC(std::uint32_t swi) override {
         fmt::print("> CallSVC({})\n", swi);

tests/unicorn_emu/a32_unicorn.cpp

@@ -52,7 +52,7 @@ void A32Unicorn<TestEnvironment>::Run() {
             return;
         }
         if (auto cerr_ = uc_emu_start(uc, pc, END_ADDRESS, 0, 1)) {
-            fmt::print("uc_emu_start failed @ {:08x} (code = {:08x}) with error {} ({})", pc, testenv.MemoryReadCode(pc), cerr_, uc_strerror(cerr_));
+            fmt::print("uc_emu_start failed @ {:08x} (code = {:08x}) with error {} ({})", pc, *testenv.MemoryReadCode(pc), cerr_, uc_strerror(cerr_));
             throw "A32Unicorn::Run() failure";
         }
         testenv.ticks_left--;