From 54c005754aedbedec44c1c98f135bfbb0de78068 Mon Sep 17 00:00:00 2001
From: Hendiadyoin1 <leon2002.la@gmail.com>
Date: Tue, 13 Jul 2021 17:18:39 +0200
Subject: [PATCH] UserpaceEmulator: Handle possibly unaligned IO

We also should add some UB warnings for that in the future
---
 .../DevTools/UserspaceEmulator/MmapRegion.cpp | 56 +++++++++++-------
 .../UserspaceEmulator/SimpleRegion.cpp        | 58 +++++++++++++------
 Userland/DevTools/UserspaceEmulator/SoftCPU.h | 12 +++-
 3 files changed, 85 insertions(+), 41 deletions(-)

diff --git a/Userland/DevTools/UserspaceEmulator/MmapRegion.cpp b/Userland/DevTools/UserspaceEmulator/MmapRegion.cpp
index a7056e47f0..eff2431999 100644
--- a/Userland/DevTools/UserspaceEmulator/MmapRegion.cpp
+++ b/Userland/DevTools/UserspaceEmulator/MmapRegion.cpp
@@ -75,7 +75,7 @@ ValueWithShadow<u8> MmapRegion::read8(FlatPtr offset)
     }
 
     VERIFY(offset < size());
-    return { *reinterpret_cast<const u8*>(m_data + offset), *reinterpret_cast<const u8*>(m_shadow_data + offset) };
+    return { m_data[offset], m_shadow_data[offset] };
 }
 
 ValueWithShadow<u16> MmapRegion::read16(u32 offset)
@@ -92,7 +92,11 @@ ValueWithShadow<u16> MmapRegion::read16(u32 offset)
     }
 
     VERIFY(offset + 1 < size());
-    return { *reinterpret_cast<const u16*>(m_data + offset), *reinterpret_cast<const u16*>(m_shadow_data + offset) };
+    u16 value, shadow;
+    ByteReader::load(m_data + offset, value);
+    ByteReader::load(m_shadow_data + offset, shadow);
+
+    return { value, shadow };
 }
 
 ValueWithShadow<u32> MmapRegion::read32(u32 offset)
@@ -109,7 +113,11 @@ ValueWithShadow<u32> MmapRegion::read32(u32 offset)
     }
 
     VERIFY(offset + 3 < size());
-    return { *reinterpret_cast<const u32*>(m_data + offset), *reinterpret_cast<const u32*>(m_shadow_data + offset) };
+    u32 value, shadow;
+    ByteReader::load(m_data + offset, value);
+    ByteReader::load(m_shadow_data + offset, shadow);
+
+    return { value, shadow };
 }
 
 ValueWithShadow<u64> MmapRegion::read64(u32 offset)
@@ -126,7 +134,11 @@ ValueWithShadow<u64> MmapRegion::read64(u32 offset)
     }
 
     VERIFY(offset + 7 < size());
-    return { *reinterpret_cast<const u64*>(m_data + offset), *reinterpret_cast<const u64*>(m_shadow_data + offset) };
+    u64 value, shadow;
+    ByteReader::load(m_data + offset, value);
+    ByteReader::load(m_shadow_data + offset, shadow);
+
+    return { value, shadow };
 }
 
 ValueWithShadow<u128> MmapRegion::read128(u32 offset)
@@ -143,7 +155,10 @@ ValueWithShadow<u128> MmapRegion::read128(u32 offset)
     }
 
     VERIFY(offset + 15 < size());
-    return { *reinterpret_cast<const u128*>(m_data + offset), *reinterpret_cast<const u128*>(m_shadow_data + offset) };
+    u128 value, shadow;
+    ByteReader::load(m_data + offset, value);
+    ByteReader::load(m_shadow_data + offset, shadow);
+    return { value, shadow };
 }
 
 ValueWithShadow<u256> MmapRegion::read256(u32 offset)
@@ -160,7 +175,10 @@ ValueWithShadow<u256> MmapRegion::read256(u32 offset)
     }
 
     VERIFY(offset + 31 < size());
-    return { *reinterpret_cast<const u256*>(m_data + offset), *reinterpret_cast<const u256*>(m_shadow_data + offset) };
+    u256 value, shadow;
+    ByteReader::load(m_data + offset, value);
+    ByteReader::load(m_shadow_data + offset, shadow);
+    return { value, shadow };
 }
 
 void MmapRegion::write8(u32 offset, ValueWithShadow<u8> value)
@@ -177,8 +195,8 @@ void MmapRegion::write8(u32 offset, ValueWithShadow<u8> value)
     }
 
     VERIFY(offset < size());
-    *reinterpret_cast<u8*>(m_data + offset) = value.value();
-    *reinterpret_cast<u8*>(m_shadow_data + offset) = value.shadow();
+    m_data[offset] = value.value();
+    m_shadow_data[offset] = value.shadow();
 }
 
 void MmapRegion::write16(u32 offset, ValueWithShadow<u16> value)
@@ -195,8 +213,8 @@ void MmapRegion::write16(u32 offset, ValueWithShadow<u16> value)
     }
 
     VERIFY(offset + 1 < size());
-    *reinterpret_cast<u16*>(m_data + offset) = value.value();
-    *reinterpret_cast<u16*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 
 void MmapRegion::write32(u32 offset, ValueWithShadow<u32> value)
@@ -214,8 +232,8 @@ void MmapRegion::write32(u32 offset, ValueWithShadow<u32> value)
 
     VERIFY(offset + 3 < size());
     VERIFY(m_data != m_shadow_data);
-    *reinterpret_cast<u32*>(m_data + offset) = value.value();
-    *reinterpret_cast<u32*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 
 void MmapRegion::write64(u32 offset, ValueWithShadow<u64> value)
@@ -233,8 +251,8 @@ void MmapRegion::write64(u32 offset, ValueWithShadow<u64> value)
 
     VERIFY(offset + 7 < size());
     VERIFY(m_data != m_shadow_data);
-    *reinterpret_cast<u64*>(m_data + offset) = value.value();
-    *reinterpret_cast<u64*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 
 void MmapRegion::write128(u32 offset, ValueWithShadow<u128> value)
@@ -249,11 +267,10 @@ void MmapRegion::write128(u32 offset, ValueWithShadow<u128> value)
         if (auto* tracer = emulator().malloc_tracer())
             tracer->audit_write(*this, base() + offset, 16);
     }
-
     VERIFY(offset + 15 < size());
     VERIFY(m_data != m_shadow_data);
-    *reinterpret_cast<u128*>(m_data + offset) = value.value();
-    *reinterpret_cast<u128*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 
 void MmapRegion::write256(u32 offset, ValueWithShadow<u256> value)
@@ -268,11 +285,10 @@ void MmapRegion::write256(u32 offset, ValueWithShadow<u256> value)
         if (auto* tracer = emulator().malloc_tracer())
             tracer->audit_write(*this, base() + offset, 32);
     }
-
     VERIFY(offset + 31 < size());
     VERIFY(m_data != m_shadow_data);
-    *reinterpret_cast<u256*>(m_data + offset) = value.value();
-    *reinterpret_cast<u256*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 
 NonnullOwnPtr<MmapRegion> MmapRegion::split_at(VirtualAddress offset)
diff --git a/Userland/DevTools/UserspaceEmulator/SimpleRegion.cpp b/Userland/DevTools/UserspaceEmulator/SimpleRegion.cpp
index 99f035c8c6..e353e94d1f 100644
--- a/Userland/DevTools/UserspaceEmulator/SimpleRegion.cpp
+++ b/Userland/DevTools/UserspaceEmulator/SimpleRegion.cpp
@@ -5,6 +5,7 @@
  */
 
 #include "SimpleRegion.h"
+#include <AK/ByteReader.h>
 #include <string.h>
 
 namespace UserspaceEmulator {
@@ -26,77 +27,98 @@ SimpleRegion::~SimpleRegion()
 ValueWithShadow<u8> SimpleRegion::read8(FlatPtr offset)
 {
     VERIFY(offset < size());
-    return { *reinterpret_cast<const u8*>(m_data + offset), *reinterpret_cast<const u8*>(m_shadow_data + offset) };
+    return { m_data[offset], m_shadow_data[offset] };
 }
 
 ValueWithShadow<u16> SimpleRegion::read16(u32 offset)
 {
     VERIFY(offset + 1 < size());
-    return { *reinterpret_cast<const u16*>(m_data + offset), *reinterpret_cast<const u16*>(m_shadow_data + offset) };
+
+    u16 value, shadow;
+    ByteReader::load<u16>(m_data + offset, value);
+    ByteReader::load<u16>(m_shadow_data + offset, shadow);
+
+    return { value, shadow };
 }
 
 ValueWithShadow<u32> SimpleRegion::read32(u32 offset)
 {
     VERIFY(offset + 3 < size());
-    return { *reinterpret_cast<const u32*>(m_data + offset), *reinterpret_cast<const u32*>(m_shadow_data + offset) };
+
+    u32 value, shadow;
+    ByteReader::load<u32>(m_data + offset, value);
+    ByteReader::load<u32>(m_shadow_data + offset, shadow);
+
+    return { value, shadow };
 }
 
 ValueWithShadow<u64> SimpleRegion::read64(u32 offset)
 {
     VERIFY(offset + 7 < size());
-    return { *reinterpret_cast<const u64*>(m_data + offset), *reinterpret_cast<const u64*>(m_shadow_data + offset) };
+
+    u64 value, shadow;
+    ByteReader::load<u64>(m_data + offset, value);
+    ByteReader::load<u64>(m_shadow_data + offset, shadow);
+
+    return { value, shadow };
 }
 
 ValueWithShadow<u128> SimpleRegion::read128(u32 offset)
 {
     VERIFY(offset + 15 < size());
-    return { *reinterpret_cast<const u128*>(m_data + offset), *reinterpret_cast<const u128*>(m_shadow_data + offset) };
+    u128 value, shadow;
+    ByteReader::load(m_data + offset, value);
+    ByteReader::load(m_shadow_data + offset, shadow);
+    return { value, shadow };
 }
 
 ValueWithShadow<u256> SimpleRegion::read256(u32 offset)
 {
     VERIFY(offset + 31 < size());
-    return { *reinterpret_cast<const u256*>(m_data + offset), *reinterpret_cast<const u256*>(m_shadow_data + offset) };
+    u256 value, shadow;
+    ByteReader::load(m_data + offset, value);
+    ByteReader::load(m_shadow_data + offset, shadow);
+    return { value, shadow };
 }
 
 void SimpleRegion::write8(u32 offset, ValueWithShadow<u8> value)
 {
     VERIFY(offset < size());
-    *reinterpret_cast<u8*>(m_data + offset) = value.value();
-    *reinterpret_cast<u8*>(m_shadow_data + offset) = value.shadow();
+    m_data[offset] = value.value();
+    m_shadow_data[offset] = value.shadow();
 }
 
 void SimpleRegion::write16(u32 offset, ValueWithShadow<u16> value)
 {
     VERIFY(offset + 1 < size());
-    *reinterpret_cast<u16*>(m_data + offset) = value.value();
-    *reinterpret_cast<u16*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 
 void SimpleRegion::write32(u32 offset, ValueWithShadow<u32> value)
 {
     VERIFY(offset + 3 < size());
-    *reinterpret_cast<u32*>(m_data + offset) = value.value();
-    *reinterpret_cast<u32*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 
 void SimpleRegion::write64(u32 offset, ValueWithShadow<u64> value)
 {
     VERIFY(offset + 7 < size());
-    *reinterpret_cast<u64*>(m_data + offset) = value.value();
-    *reinterpret_cast<u64*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 void SimpleRegion::write128(u32 offset, ValueWithShadow<u128> value)
 {
     VERIFY(offset + 15 < size());
-    *reinterpret_cast<u128*>(m_data + offset) = value.value();
-    *reinterpret_cast<u128*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 void SimpleRegion::write256(u32 offset, ValueWithShadow<u256> value)
 {
     VERIFY(offset + 31 < size());
-    *reinterpret_cast<u256*>(m_data + offset) = value.value();
-    *reinterpret_cast<u256*>(m_shadow_data + offset) = value.shadow();
+    ByteReader::store(m_data + offset, value.value());
+    ByteReader::store(m_shadow_data + offset, value.shadow());
 }
 
 u8* SimpleRegion::cacheable_ptr(u32 offset)
diff --git a/Userland/DevTools/UserspaceEmulator/SoftCPU.h b/Userland/DevTools/UserspaceEmulator/SoftCPU.h
index eb8b55d039..c9713cdb3e 100644
--- a/Userland/DevTools/UserspaceEmulator/SoftCPU.h
+++ b/Userland/DevTools/UserspaceEmulator/SoftCPU.h
@@ -8,6 +8,7 @@
 
 #include "Region.h"
 #include "ValueWithShadow.h"
+#include <AK/ByteReader.h>
 #include <LibX86/Instruction.h>
 #include <LibX86/Interpreter.h>
 
@@ -1223,7 +1224,8 @@ ALWAYS_INLINE u16 SoftCPU::read16()
     if (!m_cached_code_region || !m_cached_code_region->contains(m_eip))
         update_code_cache();
 
-    u16 value = *reinterpret_cast<const u16*>(&m_cached_code_base_ptr[m_eip - m_cached_code_region->base()]);
+    u16 value;
+    ByteReader::load<u16>(&m_cached_code_base_ptr[m_eip - m_cached_code_region->base()], value);
     m_eip += 2;
     return value;
 }
@@ -1233,7 +1235,9 @@ ALWAYS_INLINE u32 SoftCPU::read32()
     if (!m_cached_code_region || !m_cached_code_region->contains(m_eip))
         update_code_cache();
 
-    u32 value = *reinterpret_cast<const u32*>(&m_cached_code_base_ptr[m_eip - m_cached_code_region->base()]);
+    u32 value;
+    ByteReader::load<u32>(&m_cached_code_base_ptr[m_eip - m_cached_code_region->base()], value);
+
     m_eip += 4;
     return value;
 }
@@ -1243,7 +1247,9 @@ ALWAYS_INLINE u64 SoftCPU::read64()
     if (!m_cached_code_region || !m_cached_code_region->contains(m_eip))
         update_code_cache();
 
-    auto value = *reinterpret_cast<const u64*>(&m_cached_code_base_ptr[m_eip - m_cached_code_region->base()]);
+    u64 value;
+    ByteReader::load<u64>(&m_cached_code_base_ptr[m_eip - m_cached_code_region->base()], value);
+
     m_eip += 8;
     return value;
 }