Support for Linux x86-64 x32 ABI

6 files changed, 105 insertions, 17 deletions

diff --git a/binaryninjaapi.h b/binaryninjaapi.h
index ca5b630c..6d5c144c 100644
--- a/binaryninjaapi.h
+++ b/binaryninjaapi.h
@@ -15593,6 +15593,7 @@ namespace BinaryNinja {
 		static void InitViewCallback(void* ctxt, BNBinaryView* view);
 		static uint32_t* GetGlobalRegistersCallback(void* ctxt, size_t* count);
 		static void FreeRegisterListCallback(void* ctxt, uint32_t* regs, size_t count);
+		static size_t GetAddressSizeCallback(void* ctxt);
 		static BNType* GetGlobalRegisterTypeCallback(void* ctxt, uint32_t reg);
 		static void AdjustTypeParserInputCallback(
 			void* ctxt,
@@ -15777,6 +15778,12 @@ namespace BinaryNinja {
 		 */
 		virtual Ref<Type> GetGlobalRegisterType(uint32_t reg);
 
+		/*! Get the address size for this platform
+
+			\return The address size for this platform
+		*/
+		virtual size_t GetAddressSize() const;
+
 		/*! Modify the input passed to the Type Parser with Platform-specific features.
 
 			\param[in] parser Type Parser instance
@@ -15904,6 +15911,7 @@ namespace BinaryNinja {
 			std::vector<std::string>& arguments,
 			std::vector<std::pair<std::string, std::string>>& sourceFiles
 		) override;
+		virtual size_t GetAddressSize() const override;
 	};
 
 	/*!
diff --git a/binaryninjacore.h b/binaryninjacore.h
index ec03c1dd..505e10e2 100644
--- a/binaryninjacore.h
+++ b/binaryninjacore.h
@@ -37,7 +37,7 @@
 // Current ABI version for linking to the core. This is incremented any time
 // there are changes to the API that affect linking, including new functions,
 // new types, or modifications to existing functions or types.
-#define BN_CURRENT_CORE_ABI_VERSION 107
+#define BN_CURRENT_CORE_ABI_VERSION 108
 
 // Minimum ABI version that is supported for loading of plugins. Plugins that
 // are linked to an ABI version less than this will not be able to load and
@@ -1935,6 +1935,8 @@ extern "C"
 
 		BNType* (*getGlobalRegisterType)(void* ctxt, uint32_t reg);
 
+		size_t (*getAddressSize)(void* ctxt);
+
 		void (*adjustTypeParserInput)(
 			void* ctxt,
 			BNTypeParser* parser,
@@ -6972,6 +6974,7 @@ extern "C"
 
 	BINARYNINJACOREAPI uint32_t* BNGetPlatformGlobalRegisters(BNPlatform* platform, size_t* count);
 	BINARYNINJACOREAPI BNType* BNGetPlatformGlobalRegisterType(BNPlatform* platform, uint32_t reg);
+	BINARYNINJACOREAPI size_t BNGetPlatformAddressSize(BNPlatform* platform);
 	BINARYNINJACOREAPI void BNPlatformAdjustTypeParserInput(
 		BNPlatform* platform,
 		BNTypeParser* parser,
diff --git a/platform.cpp b/platform.cpp
index e367f0c2..ef32e469 100644
--- a/platform.cpp
+++ b/platform.cpp
@@ -41,6 +41,7 @@ Platform::Platform(Architecture* arch, const string& name)
 	plat.viewInit = InitViewCallback;
 	plat.getGlobalRegisters = GetGlobalRegistersCallback;
 	plat.freeRegisterList = FreeRegisterListCallback;
+	plat.getAddressSize = GetAddressSizeCallback;
 	plat.getGlobalRegisterType = GetGlobalRegisterTypeCallback;
 	plat.adjustTypeParserInput = AdjustTypeParserInputCallback;
 	plat.freeTypeParserInput = FreeTypeParserInputCallback;
@@ -59,6 +60,7 @@ Platform::Platform(Architecture* arch, const string& name, const string& typeFil
 	plat.getGlobalRegisters = GetGlobalRegistersCallback;
 	plat.freeRegisterList = FreeRegisterListCallback;
 	plat.getGlobalRegisterType = GetGlobalRegisterTypeCallback;
+	plat.getAddressSize = GetAddressSizeCallback;
 	plat.adjustTypeParserInput = AdjustTypeParserInputCallback;
 	plat.freeTypeParserInput = FreeTypeParserInputCallback;
 	plat.getFallbackEnabled = GetFallbackEnabledCallback;
@@ -193,6 +195,14 @@ BNType* Platform::GetGlobalRegisterTypeCallback(void* ctxt, uint32_t reg)
 	return BNNewTypeReference(result->GetObject());
 }
 
+
+size_t Platform::GetAddressSizeCallback(void* ctxt)
+{
+	CallbackRef<Platform> plat(ctxt);
+	return plat->GetAddressSize();
+}
+
+
 bool Platform::GetFallbackEnabledCallback(void* ctxt)
 {
 	CallbackRef<Platform> plat(ctxt);
@@ -424,6 +434,12 @@ bool Platform::GetFallbackEnabled()
 }
 
 
+size_t Platform::GetAddressSize() const
+{
+	return GetArchitecture()->GetAddressSize();
+}
+
+
 std::vector<uint32_t> CorePlatform::GetGlobalRegisters()
 {
 	size_t count;
@@ -448,6 +464,12 @@ Ref<Type> CorePlatform::GetGlobalRegisterType(uint32_t reg)
 }
 
 
+size_t CorePlatform::GetAddressSize() const
+{
+	return BNGetPlatformAddressSize(m_object);
+}
+
+
 void Platform::AdjustTypeParserInput(
 	Ref<TypeParser> parser,
 	vector<string>& arguments,
diff --git a/platform/linux/platform_linux.cpp b/platform/linux/platform_linux.cpp
index 0102fa42..20a258d6 100644
--- a/platform/linux/platform_linux.cpp
+++ b/platform/linux/platform_linux.cpp
@@ -3,6 +3,8 @@
 using namespace BinaryNinja;
 using namespace std;
 
+Ref<Platform> g_linuxX32;
+#define EM_X86_64 62 // AMD x86-64 architecture
 
 class LinuxX86Platform: public Platform
 {
@@ -103,6 +105,50 @@ public:
 };
 
 
+class LinuxX32Platform: public Platform
+{
+	public:
+	LinuxX32Platform(Architecture* arch): Platform(arch, "linux-x32")
+	{
+		Ref<CallingConvention> cc;
+		cc = arch->GetCallingConventionByName("sysv");
+		if (cc)
+		{
+			RegisterDefaultCallingConvention(cc);
+			RegisterCdeclCallingConvention(cc);
+			RegisterFastcallCallingConvention(cc);
+			RegisterStdcallCallingConvention(cc);
+		}
+
+		cc = arch->GetCallingConventionByName("linux-syscall");
+		if (cc)
+			SetSystemCallConvention(cc);
+	}
+
+	virtual size_t GetAddressSize() const override
+	{
+		return 4;
+	}
+
+	static Ref<Platform> Recognize(BinaryView* view, Metadata* metadata)
+	{
+		Ref<Metadata> fileClass = metadata->Get("EI_CLASS");
+
+		if (!fileClass || !fileClass->IsUnsignedInteger())
+			return nullptr;
+
+		Ref<Metadata> machine = metadata->Get("e_machine");
+		if (!machine || !machine->IsUnsignedInteger())
+			return nullptr;
+
+		if (fileClass->GetUnsignedInteger() == 1 && machine->GetUnsignedInteger() == EM_X86_64)
+			return g_linuxX32;
+
+		return nullptr;
+	}
+};
+
+
 class LinuxArmv7Platform: public Platform
 {
 public:
@@ -314,13 +360,20 @@ extern "C"
 		Ref<Architecture> x64 = Architecture::GetByName("x86_64");
 		if (x64)
 		{
-			Ref<Platform> platform;
+			Ref<Platform> x64Platform = new LinuxX64Platform(x64);
+			g_linuxX32 = new LinuxX32Platform(x64);
+
+			Platform::Register("linux", x64Platform);
+			Platform::Register("linux", g_linuxX32);
 
-			platform = new LinuxX64Platform(x64);
-			Platform::Register("linux", platform);
 			// Linux binaries sometimes have an OS identifier of zero, even though 3 is the correct one
-			BinaryViewType::RegisterPlatform("ELF", 0, platform);
-			BinaryViewType::RegisterPlatform("ELF", 3, platform);
+			BinaryViewType::RegisterPlatform("ELF", 0, x64, x64Platform);
+			BinaryViewType::RegisterPlatform("ELF", 3, x64, x64Platform);
+
+
+			Ref<BinaryViewType> elf = BinaryViewType::GetByName("ELF");
+			if (elf)
+				elf->RegisterPlatformRecognizer(EM_X86_64, LittleEndian, LinuxX32Platform::Recognize);
 		}
 
 		Ref<Architecture> armv7 = Architecture::GetByName("armv7");
diff --git a/python/platform.py b/python/platform.py
index d7cdabe8..29eb81b0 100644
--- a/python/platform.py
+++ b/python/platform.py
@@ -113,6 +113,7 @@ class Platform(metaclass=_PlatformMetaClass):
 			self._cb.getGlobalRegisters = self._cb.getGlobalRegisters.__class__(self._get_global_regs)
 			self._cb.freeRegisterList = self._cb.freeRegisterList.__class__(self._free_register_list)
 			self._cb.getGlobalRegisterType = self._cb.getGlobalRegisterType.__class__(self._get_global_reg_type)
+			self._cb.getAddressSize = self._cb.getAddressSize.__class__(self._get_address_size)
 			self._cb.adjustTypeParserInput = self._cb.adjustTypeParserInput.__class__(self._adjust_type_parser_input)
 			self._cb.freeTypeParserInput = self._cb.freeTypeParserInput.__class__(self._free_type_parser_input)
 			self._pending_reg_lists = {}
@@ -149,6 +150,7 @@ class Platform(metaclass=_PlatformMetaClass):
 		self.handle: ctypes.POINTER(core.BNPlatform) = _handle
 		self._arch = _arch
 		self._name = None
+		self._address_size = core.BNGetPlatformAddressSize(_handle)
 
 	def _init(self, ctxt):
 		pass
@@ -196,6 +198,12 @@ class Platform(metaclass=_PlatformMetaClass):
 			log_error(traceback.format_exc())
 			return None
 
+	def _get_address_size(self, ctxt):
+		try:
+			return self.address_size
+		except:
+			return self.arch.address_size
+
 	def _adjust_type_parser_input(
 			self,
 			ctxt,
@@ -328,6 +336,10 @@ class Platform(metaclass=_PlatformMetaClass):
 			self._name = core.BNGetPlatformName(self.handle)
 		return self._name
 
+	@property
+	def address_size(self) -> int:
+		return self._address_size
+
 	@classmethod
 	@property
 	def os_list(cls) -> List[str]:
diff --git a/view/elf/elfview.cpp b/view/elf/elfview.cpp
index 6d19bc5f..bb193495 100644
--- a/view/elf/elfview.cpp
+++ b/view/elf/elfview.cpp
@@ -54,7 +54,7 @@ ElfView::ElfView(BinaryView* data, bool parseOnly): BinaryView("ELF", data->GetF
 	CreateLogger("BinaryView");
 	m_logger = CreateLogger("BinaryView.ElfView");
 	m_elf32 = m_ident.fileClass == 1;
-	m_addressSize = (m_ident.fileClass == 1) ? 4 : 8;
+	m_addressSize = (m_ident.fileClass == 1 || (m_plat && m_plat->GetName() == "linux-32")) ? 4 : 8;
 	m_endian = endian;
 	m_relocatable = m_commonHeader.type == ET_DYN || m_commonHeader.type == ET_REL;
 	m_objectFile = m_commonHeader.type == ET_REL;
@@ -2874,16 +2874,6 @@ uint64_t ElfViewType::ParseHeaders(BinaryView* data, ElfIdent& ident, ElfCommonH
 	commonHeader.arch = reader.Read16();
 	commonHeader.version = reader.Read32();
 
-	// Promote the file class to 64-bit
-	// TODO potentially add a setting to allow the user to override header interpretation
-	if ((commonHeader.type == ET_EXEC) && (commonHeader.arch == EM_X86_64) && (ident.fileClass == 1))
-	{
-		ident.fileClass = 2;
-		m_logger->LogWarn(
-			"Executable file claims to be 32-bit but specifies a 64-bit architecture. It is likely malformed or "
-			"malicious. Treating it as 64-bit.");
-	}
-
 	// parse Elf64Header
 	if (ident.fileClass == 1) // 32-bit ELF
 	{