Issue 115 (avast#422)

* capstone2llvmir: check preconditions in translation routines.

* capstone2llvmir: add all registers for all architectures.

* capstone2llvmir: fix problems with regression tests.

* capstone2llvmir/x86: remove initializeRegistersParentMapCommon().

* capstone2llvmir: refactor.

* bin2llvmir/decoder: catch capstone2llvmir errors and exit.

* capstone2llvmir: provide API to query pseudo asm functions.

* capstone2llvmir: only one implementation of translatePseudoAsmGeneric().

* capstone2llvmir/powerpc: handle some crazy bcctrl (bdzctrl) variant.

* capstone2llvmir/x86: create Kx registers.

* capstone2llvmir/ppc: modify comment.

include/retdec/bin2llvmir/optimizations/decoder/decoder.h

@@ -317,8 +317,6 @@ class Decoder : public llvm::ModulePass
 		/// label is not in jump table).
 		std::map<utils::Address, std::set<llvm::SwitchInst*>> _switchTableStarts;
 
-		std::map<llvm::CallInst*, llvm::Instruction*> _pseudoCalls;
-
 		// We create helper BBs (without name and address) to handle MIPS
 		// likely branches. For convenience, we map them to real BBs they will
 		// eventually jump to.

include/retdec/bin2llvmir/providers/config.h

@@ -171,6 +171,12 @@ class Config
 		llvm::CallInst* isLlvmX87StorePseudoFunctionCall(llvm::Value* c);
 		llvm::CallInst* isLlvmX87LoadPseudoFunctionCall(llvm::Value* c);
 
+		// Assembly pseudo-functions.
+		//
+		void addPseudoAsmFunction(llvm::Function* f);
+		bool isPseudoAsmFunction(llvm::Function* f);
+		llvm::CallInst* isPseudoAsmFunctionCall(llvm::Value* c);
+
 		// Other
 		//
 		llvm::GlobalVariable* getGlobalDummy();
@@ -200,6 +206,7 @@ class Config
 		llvm::Function* _x87TagLoadFunction = nullptr; // i2 (i3)
 
 		std::map<IntrinsicFunctionCreatorPtr, llvm::Function*> _intrinsicFunctions;
+		std::set<llvm::Function*> _pseudoAsmFunctions;
 };
 
 class ConfigProvider

include/retdec/capstone2llvmir/arm/arm_defs.h

@@ -17,4 +17,28 @@ enum arm_reg_cpsr_flags
 	ARM_REG_CPSR_V,
 };
 
+/**
+ * The problem: ARM uses two sets of registers (two enums), ordinary registers
+ * (enum arm_reg) and system registers (enum arm_sysreg).
+ * Most system registers have enum numbers greater than 256 - these numbers do
+ * not overlap with ordinary register numbers.
+ * But 8 registers from arm_sysreg overlap with ordinary registers.
+ * These are SPSR and CPSR related registers.
+ * We cannot use these Capstone enums, since they would collide in our maps with
+ * ordinary registers.
+ * Moreover, these 8 registers denote flag registers and can be OR combined.
+ * Therefore, if we wanted to capture their full semantics, we would either have
+ * to create registers for all combinations (e.g. C, X, S, F, CX, CS, CF, XS,
+ * XF, SF, CXS, ...) and use appropriate variant depending on asm instruction
+ * as Capstone/IDA does (e.g. msr cpsr_fc, r6), or simulate this on a single
+ * register using bit setting (i.e. and/or operations).
+ * Instead, we ignore work with individual flags - we create only two registers
+ * and use them every time any flag is modified.
+ */
+enum arm_sysreg_extension
+{
+	ARM_SYSREG_SPSR = ARM_REG_CPSR_V + 1,
+	ARM_SYSREG_CPSR,
+};
+
 #endif

include/retdec/capstone2llvmir/capstone2llvmir.h

@@ -252,6 +252,43 @@ class Capstone2LlvmIrTranslator
 		virtual ~Capstone2LlvmIrTranslator();
 //
 //==============================================================================
+// Translator configuration methods.
+//==============================================================================
+//
+		/**
+		 * Should the translator ignore unexpected operands encountered in
+		 * Capstone instructions?
+		 * True -> ignore -> try recover or ignore the problem.
+		 * False -> don't ignore -> throw @c UnexpectedOperandsError.
+		 *
+		 * Default value: true.
+		 */
+		virtual void setIgnoreUnexpectedOperands(bool f) = 0;
+		/**
+		 * Should the translator ignore unhandled instructions?
+		 * True -> ignore.
+		 * False -> don't ignore -> throw @c UnhandledInstructionError when
+		 * instructions without an implemented translation routine is
+		 * encountered.
+		 *
+		 * Default value: true.
+		 */
+		virtual void setIgnoreUnhandledInstructions(bool f) = 0;
+		/**
+		 * Should the translator generate pseudo assembly functions for
+		 * instructions which full semantics is not implemented?
+		 * True -> generate.
+		 * False -> don't generate.
+		 *
+		 * Default value: true.
+		 */
+		virtual void setGeneratePseudoAsmFunctions(bool f) = 0;
+
+		virtual bool isIgnoreUnexpectedOperands() const = 0;
+		virtual bool isIgnoreUnhandledInstructions() const = 0;
+		virtual bool isGeneratePseudoAsmFunctions() const = 0;
+//
+//==============================================================================
 // Mode query & modification methods.
 //==============================================================================
 //
@@ -701,6 +738,22 @@ class Capstone2LlvmIrTranslator
 		 * @c ARM_REG_INVALID, @c MIPS_REG_INVALID).
 		 */
 		virtual uint32_t getCapstoneRegister(llvm::GlobalVariable* gv) const = 0;
+
+		/**
+		 * Is the passed LLVM function @p f any pseudo assembly functions for
+		 * instructions which full semantics is not implemented?
+		 */
+		virtual bool isPseudoAsmFunction(llvm::Function* f) const = 0;
+		/**
+		 * Is the passed LLVM call @p c any kind of pseudo assembly call for
+		 * instructions which full semantics is not implemented?
+		 */
+		virtual bool isPseudoAsmFunctionCall(llvm::CallInst* c) const = 0;
+		/**
+		 * Get all pseudo assembly functions for instructions which full
+		 * semantics is not implemented.
+		 */
+		virtual const std::set<llvm::Function*>& getPseudoAsmFunctions() const = 0;
 };
 
 } // namespace capstone2llvmir

include/retdec/capstone2llvmir/exceptions.h

@@ -18,17 +18,18 @@ namespace capstone2llvmir {
 
 /**
  * Base class for all Capstone2LllvmIr errors.
+ * This class itself should never be thrown.
  */
-class Capstone2LlvmIrBaseError : public std::exception
+class BaseError : public std::exception
 {
 	public:
-		virtual ~Capstone2LlvmIrBaseError();
+		virtual ~BaseError();
 };
 
 /**
- * An exception class encapsulating Capstone errors.
+ * An exception class encapsulating all Capstone errors.
  */
-class CapstoneError : public Capstone2LlvmIrBaseError
+class CapstoneError : public BaseError
 {
 	public:
 		CapstoneError(cs_err e);
@@ -45,7 +46,7 @@ class CapstoneError : public Capstone2LlvmIrBaseError
 /**
  * An exception class related to Capstone mode setting errors.
  */
-class Capstone2LlvmIrModeError : public Capstone2LlvmIrBaseError
+class ModeSettingError : public BaseError
 {
 	public:
 		enum class eType
@@ -60,8 +61,8 @@ class Capstone2LlvmIrModeError : public Capstone2LlvmIrBaseError
 		};
 
 	public:
-		Capstone2LlvmIrModeError(cs_arch a, cs_mode m, eType t);
-		virtual ~Capstone2LlvmIrModeError();
+		ModeSettingError(cs_arch a, cs_mode m, eType t);
+		virtual ~ModeSettingError();
 
 		std::string getMessage() const;
 		virtual const char* what() const noexcept override;
@@ -73,13 +74,63 @@ class Capstone2LlvmIrModeError : public Capstone2LlvmIrBaseError
 };
 
 /**
- * A general exception class for all Capstone2LlvmIr errors.
+ * An exception class thrown when unexpected operand(s) (number, type, etc.)
+ * is(are) encountered.
+ *
+ * These exceptions may be suppressed and/or ignored.
  */
-class Capstone2LlvmIrError : public Capstone2LlvmIrBaseError
+class UnexpectedOperandsError : public BaseError
 {
 	public:
-		Capstone2LlvmIrError(const std::string& message);
-		virtual ~Capstone2LlvmIrError();
+		/**
+		 * @param i       Capstone instruction in which unexpected operand
+		 *                was encountered.
+		 * @param comment Optional comment about the problem.
+		 */
+		UnexpectedOperandsError(cs_insn* i, const std::string& comment = "");
+		virtual ~UnexpectedOperandsError();
+
+		virtual const char* what() const noexcept override;
+
+	private:
+		cs_insn* _insn = nullptr;
+		std::string _comment;
+};
+
+/**
+ * An exception class thrown when unhandled instruction is encountered.
+ *
+ * These exceptions may be suppressed and/or ignored. Not all instructions are
+ * handled, or will be handled in the future.
+ */
+class UnhandledInstructionError : public BaseError
+{
+	public:
+		/**
+		 * @param i       Capstone instruction which is not handled.
+		 * @param comment Optional comment about the problem.
+		 */
+		UnhandledInstructionError(cs_insn* i, const std::string& comment = "");
+		~UnhandledInstructionError();
+
+		virtual const char* what() const noexcept override;
+
+	private:
+		cs_insn* _insn = nullptr;
+		std::string _comment;
+};
+
+/**
+ * A generic exception class for miscellaneous Capstone2LlvmIr errors.
+ *
+ * These exceptions signal some operational problems in Capstone2LlvmIr library.
+ * They should not be ignored. They should be reported to RetDec developers.
+ */
+class GenericError : public BaseError
+{
+	public:
+		GenericError(const std::string& message);
+		virtual ~GenericError();
 
 		virtual const char* what() const noexcept override;
 

src/bin2llvmir/optimizations/decoder/decoder.cpp

@@ -89,10 +89,10 @@ bool Decoder::runCatcher()
 	{
 		return run();
 	}
-	catch (const Capstone2LlvmIrBaseError& e)
+	catch (const BaseError& e)
 	{
-		LOG << "[capstone2llvmir]: " << e.what() << std::endl;
-		return false;
+		std::cerr << "[capstone2llvmir]: " << e.what() << std::endl;
+		exit(1);
 	}
 }
 

src/bin2llvmir/optimizations/decoder/decoder_init.cpp

@@ -1021,6 +1021,11 @@ void Decoder::initConfigFunctions()
 			cf->setSourceFileName(df->getSourceFileName());
 		}
 	}
+
+	for (auto* f : _c2l->getPseudoAsmFunctions())
+	{
+		_config->addPseudoAsmFunction(f);
+	}
 }
 
 } // namespace bin2llvmir

src/bin2llvmir/providers/config.cpp

@@ -716,6 +716,22 @@ llvm::CallInst* Config::isLlvmAnyUncondBranchPseudoFunctionCall(llvm::Value* c)
 	return nullptr;
 }
 
+void Config::addPseudoAsmFunction(llvm::Function* f)
+{
+	_pseudoAsmFunctions.insert(f);
+}
+
+bool Config::isPseudoAsmFunction(llvm::Function* f)
+{
+	return _pseudoAsmFunctions.count(f);
+}
+
+llvm::CallInst* Config::isPseudoAsmFunctionCall(llvm::Value* c)
+{
+	auto* cc = dyn_cast_or_null<CallInst>(c);
+	return isPseudoAsmFunction(cc->getCalledFunction()) ? cc : nullptr;
+}
+
 /**
  * Get crypto pattern information for address \p addr - fill \p name,
  * \p description, and \p type, if there is a pattern on address.