x86: fix known issue with prefix by combining with previous prefix instruction. this is not perfect, but good enough for now

arch/X86/X86Mapping.c

@@ -6636,3 +6636,46 @@ unsigned int X86_get_insn_id2(unsigned int id)
 {
 	return insn_reverse_id(insns, ARR_SIZE(insns), id);
 }
+
+// can this instruction combine with prev prefix instruction?
+// this also updates h->pre_prefix if needed
+bool X86_insn_check_combine(cs_struct *h, cs_insn *insn)
+{
+	// is this a prefix instruction?
+	if (insn->id == X86_INS_LOCK || insn->id == X86_INS_REP ||
+		insn->id == X86_INS_REPNE) {
+		// then save this as prev_prefix
+		h->prev_prefix = insn->id;
+		return false;
+	}
+
+	// if the previous instruction is a prefix, then OK to combine with this
+	if (h->prev_prefix) {
+		return true;
+	}
+
+	// cannot combine this with a prefix
+	return false;
+}
+
+// combine this instruction with previous prefix instruction
+void X86_insn_combine(cs_struct *h, cs_insn *insn, cs_insn *prev)
+{
+	// reset prev_prefix
+	h->prev_prefix = 0;
+
+	// copy information from insn to prev
+	prev->id = insn->id;
+	prev->size += insn->size;
+	memmove(prev->bytes+1, insn->bytes, sizeof(insn->bytes) - 1);
+	strlcat(prev->mnemonic, " ", sizeof(insn->mnemonic));
+	strlcat(prev->mnemonic, insn->mnemonic, sizeof(insn->mnemonic));
+	strlcpy(prev->op_str, insn->op_str, sizeof(insn->op_str));
+
+	if (h->detail) {
+		// save old prefix to copy it back later
+		uint8_t prefix = prev->detail->x86.opcode[0];
+		memmove(prev->detail, insn->detail, sizeof(cs_detail));
+		prev->detail->x86.prefix[0] = prefix;
+	}
+}

arch/X86/X86Mapping.h

@@ -38,4 +38,9 @@ unsigned int X86_get_insn_id2(unsigned int insn_id);
 // post printer for X86.
 void X86_post_printer(csh handle, cs_insn *pub_insn, char *insn_asm);
 
+// handle X86 prefixes
+bool X86_insn_check_combine(cs_struct *h, cs_insn *insn);
+
+void X86_insn_combine(cs_struct *h, cs_insn *insn, cs_insn *prev);
+
 #endif

arch/X86/X86Module.c

@@ -19,6 +19,8 @@ static cs_err init(cs_struct *ud)
 	ud->insn_id = X86_get_insn_id;
 	ud->insn_name = X86_insn_name;
 	ud->post_printer = X86_post_printer;
+	ud->check_combine = X86_insn_check_combine;
+	ud->combine = X86_insn_combine;
 
 	return CS_ERR_OK;
 }

cs.c

@@ -268,6 +268,18 @@ cs_err cs_option(csh ud, cs_opt_type type, size_t value)
 	return arch_option[handle->arch](handle, type, value);
 }
 
+// get previous instruction, which can be in the cache, or in total buffer
+static cs_insn *get_prev_insn(cs_insn *cache, unsigned int f, void *total, size_t total_size)
+{
+	if (f == 0) {
+		if (total == NULL)
+			return NULL;
+		// get the trailing insn from total buffer
+		return (cs_insn *)(total + total_size - sizeof(cs_insn));
+	} else
+		return &cache[f - 1];
+}
+
 // dynamicly allocate memory to contain disasm insn
 // NOTE: caller must free() the allocated memory itself to avoid memory leaking
 size_t cs_disasm_ex(csh ud, const uint8_t *buffer, size_t size, uint64_t offset, size_t count, cs_insn **insn)
@@ -275,7 +287,8 @@ size_t cs_disasm_ex(csh ud, const uint8_t *buffer, size_t size, uint64_t offset,
 	cs_struct *handle = (cs_struct *)(uintptr_t)ud;
 	MCInst mci;
 	uint16_t insn_size;
-	size_t c = 0, f = 0;
+	size_t c = 0;
+	unsigned int f = 0;
 	cs_insn insn_cache[64];
 	void *total = NULL;
 	size_t total_size = 0;
@@ -315,25 +328,32 @@ size_t cs_disasm_ex(csh ud, const uint8_t *buffer, size_t size, uint64_t offset,
 
 			fill_insn(handle, &insn_cache[f], ss.buffer, &mci, handle->post_printer, buffer);
 
-			f++;
+			if (!handle->check_combine || !handle->check_combine(handle, &insn_cache[f])) {
+				f++;
 
-			if (f == ARR_SIZE(insn_cache)) {
-				// resize total to contain newly disasm insns
-				total_size += sizeof(insn_cache);
-				void *tmp = cs_mem_realloc(total, total_size);
-				if (tmp == NULL) {	// insufficient memory
-					cs_mem_free(total);
-					handle->errnum = CS_ERR_MEM;
-					return 0;
+				if (f == ARR_SIZE(insn_cache)) {
+					// resize total to contain newly disasm insns
+					total_size += sizeof(insn_cache);
+					void *tmp = cs_mem_realloc(total, total_size);
+					if (tmp == NULL) {	// insufficient memory
+						cs_mem_free(total);
+						handle->errnum = CS_ERR_MEM;
+						return 0;
+					}
+
+					total = tmp;
+					memcpy(total + total_size - sizeof(insn_cache), insn_cache, sizeof(insn_cache));
+					// reset f back to 0
+					f = 0;
 				}
 
-				total = tmp;
-				memcpy(total + total_size - sizeof(insn_cache), insn_cache, sizeof(insn_cache));
-				// reset f back to 0
-				f = 0;
+				c++;
+			} else {
+				// combine this instruction with previous prefix instruction
+				cs_insn *prev = get_prev_insn(insn_cache, f, total, total_size);
+				handle->combine(handle, &insn_cache[f], prev);
 			}
 
-			c++;
 			buffer += insn_size;
 			size -= insn_size;
 			offset += insn_size;

cs_priv.h

@@ -21,6 +21,10 @@ typedef const char *(*GetName_t)(csh handle, unsigned int reg);
 
 typedef void (*GetID_t)(cs_struct *h, cs_insn *insn, unsigned int id);
 
+typedef bool (*CheckCombineInsn_t)(cs_struct *h, cs_insn *insn);
+
+typedef void (*CombineInsn_t)(cs_struct *h, cs_insn *insn, cs_insn *prev);
+
 // for ARM only
 typedef struct ARM_ITStatus {
 	unsigned char ITStates[128];	// FIXME
@@ -45,6 +49,9 @@ struct cs_struct {
 	int syntax;	// asm syntax for simple printer such as PPC
 	bool doing_mem;	// handling memory operand in InstPrinter code
 	unsigned short *insn_cache;	// index caching for mapping.c
+	CheckCombineInsn_t check_combine;
+	CombineInsn_t combine;
+	uint8_t prev_prefix;	// save previous prefix for combining instructions - X86 only.
 };
 
 #define MAX_ARCH 8