Merge pull request #853 from serpilliere/fix_xmm_accesses

Fix xmm accesses

11 files changed, 369 insertions, 290 deletions

diff --git a/miasm2/jitter/JitCore.h b/miasm2/jitter/JitCore.h
index 0eaf4939..d85b71d9 100644
--- a/miasm2/jitter/JitCore.h
+++ b/miasm2/jitter/JitCore.h
@@ -36,10 +36,11 @@
 	}								\
 
 
-#define getset_reg_bn(regname)						\
+#define getset_reg_bn(regname, size)					\
 	static PyObject *JitCpu_get_ ## regname  (JitCpu *self, void *closure) \
 	{								\
 		bn_t bn;						\
+		int j;							\
 		PyObject* py_long;					\
 		PyObject* py_long_new;					\
 		PyObject* py_tmp;					\
@@ -48,13 +49,13 @@
 		py_long = PyLong_FromLong(0);				\
 		cst_32 = PyLong_FromLong(32);				\
 		bn = ((vm_cpu_t*)(self->cpu))->  regname;		\
-		while (!bignum_is_zero(bn)) {				\
-			tmp = bignum_to_uint64(bignum_mask(bn, 32)) & 0xffffffff; \
-			bn = bignum_rshift(bn, 32);			\
+		bn = bignum_mask(bn, (size));				\
+		for (j = BN_BYTE_SIZE - 4; j >= 0 ; j -= 4) {		\
+			tmp = bignum_to_uint64(bignum_mask(bignum_rshift(bn, 8 * j), 32)); \
 			py_tmp = PyLong_FromUnsignedLong(tmp);		\
 			py_long_new = PyObject_CallMethod(py_long, "__lshift__", "O", cst_32); \
 			Py_DECREF(py_long);				\
-			py_long = PyObject_CallMethod(py_long_new, "__add__", "O", py_tmp);	\
+			py_long = PyObject_CallMethod(py_long_new, "__add__", "O", py_tmp); \
 			Py_DECREF(py_long_new);				\
 			Py_DECREF(py_tmp);				\
 		}							\
@@ -82,7 +83,7 @@
 			/* Increment ref as we will decement it next */	\
 			Py_INCREF(py_long);				\
 		}							\
-		else{							\
+		else {							\
 			PyErr_SetString(PyExc_TypeError, "Arg must be int"); \
 			return -1;					\
 		}							\
@@ -93,23 +94,21 @@
 									\
 		for (j = 0; j < BN_BYTE_SIZE; j += 4) {			\
 			py_tmp = PyObject_CallMethod(py_long, "__and__", "O", cst_ffffffff); \
-			tmp = PyLong_AsUnsignedLongMask(py_tmp);	\
-			Py_DECREF(py_tmp);				\
-			bn = bignum_lshift(bn, 32);			\
-			bn = bignum_or(bn, bignum_from_uint64(tmp));	\
 			py_long_new = PyObject_CallMethod(py_long, "__rshift__", "O", cst_32); \
 			Py_DECREF(py_long);				\
 			py_long = py_long_new;				\
+			tmp = PyLong_AsUnsignedLongMask(py_tmp);	\
+			Py_DECREF(py_tmp);				\
+			bn = bignum_or(bn, bignum_lshift(bignum_from_uint64(tmp), 8 * j)); \
 		}							\
 									\
-		((vm_cpu_t*)(self->cpu))->  regname   = bn;		\
+		((vm_cpu_t*)(self->cpu))->  regname   = bignum_mask(bn, (size)); \
 		Py_DECREF(py_long);					\
 		Py_DECREF(cst_32);					\
 		Py_DECREF(cst_ffffffff);				\
 		return 0;						\
 	}
 
-
 #define getset_reg_u64(regname)						\
 	static PyObject *JitCpu_get_ ## regname  (JitCpu *self, void *closure) \
 	{								\
@@ -158,8 +157,9 @@
 	} while(0);
 
 
-#define get_reg_bn(reg)  do {						\
+#define get_reg_bn(reg, size)  do {					\
 		bn_t bn;						\
+		int j;							\
 		PyObject* py_long;					\
 		PyObject* py_long_new;					\
 		PyObject* py_tmp;					\
@@ -168,10 +168,10 @@
 		py_long = PyLong_FromLong(0);				\
 		cst_32 = PyLong_FromLong(32);				\
 		bn = ((vm_cpu_t*)(self->cpu))->  reg;			\
-		while (!bignum_is_zero(bn)) {				\
-			tmp = bignum_to_uint64(bignum_mask(bn, 32)) & 0xffffffff; \
-			bn = bignum_rshift(bn, 32);			\
-			py_tmp = PyLong_FromLong(tmp);			\
+		bn = bignum_mask(bn, size);				\
+		for (j = BN_BYTE_SIZE - 4; j >= 0 ; j -= 4) {		\
+			tmp = bignum_to_uint64(bignum_mask(bignum_rshift(bn, 8 * j), 32)); \
+			py_tmp = PyLong_FromUnsignedLong(tmp);		\
 			py_long_new = PyObject_CallMethod(py_long, "__lshift__", "O", cst_32); \
 			Py_DECREF(py_long);				\
 			py_long = PyObject_CallMethod(py_long_new, "__add__", "O", py_tmp); \
diff --git a/miasm2/jitter/arch/JitCore_aarch64.c b/miasm2/jitter/arch/JitCore_aarch64.c
index f5977308..76452a44 100644
--- a/miasm2/jitter/arch/JitCore_aarch64.c
+++ b/miasm2/jitter/arch/JitCore_aarch64.c
@@ -13,48 +13,48 @@
 
 
 reg_dict gpreg_dict[] = {
-	{.name = "X0", .offset = offsetof(vm_cpu_t, X0)},
-	{.name = "X1", .offset = offsetof(vm_cpu_t, X1)},
-	{.name = "X2", .offset = offsetof(vm_cpu_t, X2)},
-	{.name = "X3", .offset = offsetof(vm_cpu_t, X3)},
-	{.name = "X4", .offset = offsetof(vm_cpu_t, X4)},
-	{.name = "X5", .offset = offsetof(vm_cpu_t, X5)},
-	{.name = "X6", .offset = offsetof(vm_cpu_t, X6)},
-	{.name = "X7", .offset = offsetof(vm_cpu_t, X7)},
-	{.name = "X8", .offset = offsetof(vm_cpu_t, X8)},
-	{.name = "X9", .offset = offsetof(vm_cpu_t, X9)},
-	{.name = "X10", .offset = offsetof(vm_cpu_t, X10)},
-	{.name = "X11", .offset = offsetof(vm_cpu_t, X11)},
-	{.name = "X12", .offset = offsetof(vm_cpu_t, X12)},
-	{.name = "X13", .offset = offsetof(vm_cpu_t, X13)},
-	{.name = "X14", .offset = offsetof(vm_cpu_t, X14)},
-	{.name = "X15", .offset = offsetof(vm_cpu_t, X15)},
-	{.name = "X16", .offset = offsetof(vm_cpu_t, X16)},
-	{.name = "X17", .offset = offsetof(vm_cpu_t, X17)},
-	{.name = "X18", .offset = offsetof(vm_cpu_t, X18)},
-	{.name = "X19", .offset = offsetof(vm_cpu_t, X19)},
-	{.name = "X20", .offset = offsetof(vm_cpu_t, X20)},
-	{.name = "X21", .offset = offsetof(vm_cpu_t, X21)},
-	{.name = "X22", .offset = offsetof(vm_cpu_t, X22)},
-	{.name = "X23", .offset = offsetof(vm_cpu_t, X23)},
-	{.name = "X24", .offset = offsetof(vm_cpu_t, X24)},
-	{.name = "X25", .offset = offsetof(vm_cpu_t, X25)},
-	{.name = "X26", .offset = offsetof(vm_cpu_t, X26)},
-	{.name = "X27", .offset = offsetof(vm_cpu_t, X27)},
-	{.name = "X28", .offset = offsetof(vm_cpu_t, X28)},
-	{.name = "X29", .offset = offsetof(vm_cpu_t, X29)},
-	{.name = "LR", .offset = offsetof(vm_cpu_t, LR)},
-
-	{.name = "SP", .offset = offsetof(vm_cpu_t, SP)},
-	{.name = "PC", .offset = offsetof(vm_cpu_t, PC)},
-
-	{.name = "zf", .offset = offsetof(vm_cpu_t, zf)},
-	{.name = "nf", .offset = offsetof(vm_cpu_t, nf)},
-	{.name = "of", .offset = offsetof(vm_cpu_t, of)},
-	{.name = "cf", .offset = offsetof(vm_cpu_t, cf)},
-
-	{.name = "exception_flags", .offset = offsetof(vm_cpu_t, exception_flags)},
-	{.name = "interrupt_num", .offset = offsetof(vm_cpu_t, interrupt_num)},
+	{.name = "X0", .offset = offsetof(vm_cpu_t, X0), .size = 64},
+	{.name = "X1", .offset = offsetof(vm_cpu_t, X1), .size = 64},
+	{.name = "X2", .offset = offsetof(vm_cpu_t, X2), .size = 64},
+	{.name = "X3", .offset = offsetof(vm_cpu_t, X3), .size = 64},
+	{.name = "X4", .offset = offsetof(vm_cpu_t, X4), .size = 64},
+	{.name = "X5", .offset = offsetof(vm_cpu_t, X5), .size = 64},
+	{.name = "X6", .offset = offsetof(vm_cpu_t, X6), .size = 64},
+	{.name = "X7", .offset = offsetof(vm_cpu_t, X7), .size = 64},
+	{.name = "X8", .offset = offsetof(vm_cpu_t, X8), .size = 64},
+	{.name = "X9", .offset = offsetof(vm_cpu_t, X9), .size = 64},
+	{.name = "X10", .offset = offsetof(vm_cpu_t, X10), .size = 64},
+	{.name = "X11", .offset = offsetof(vm_cpu_t, X11), .size = 64},
+	{.name = "X12", .offset = offsetof(vm_cpu_t, X12), .size = 64},
+	{.name = "X13", .offset = offsetof(vm_cpu_t, X13), .size = 64},
+	{.name = "X14", .offset = offsetof(vm_cpu_t, X14), .size = 64},
+	{.name = "X15", .offset = offsetof(vm_cpu_t, X15), .size = 64},
+	{.name = "X16", .offset = offsetof(vm_cpu_t, X16), .size = 64},
+	{.name = "X17", .offset = offsetof(vm_cpu_t, X17), .size = 64},
+	{.name = "X18", .offset = offsetof(vm_cpu_t, X18), .size = 64},
+	{.name = "X19", .offset = offsetof(vm_cpu_t, X19), .size = 64},
+	{.name = "X20", .offset = offsetof(vm_cpu_t, X20), .size = 64},
+	{.name = "X21", .offset = offsetof(vm_cpu_t, X21), .size = 64},
+	{.name = "X22", .offset = offsetof(vm_cpu_t, X22), .size = 64},
+	{.name = "X23", .offset = offsetof(vm_cpu_t, X23), .size = 64},
+	{.name = "X24", .offset = offsetof(vm_cpu_t, X24), .size = 64},
+	{.name = "X25", .offset = offsetof(vm_cpu_t, X25), .size = 64},
+	{.name = "X26", .offset = offsetof(vm_cpu_t, X26), .size = 64},
+	{.name = "X27", .offset = offsetof(vm_cpu_t, X27), .size = 64},
+	{.name = "X28", .offset = offsetof(vm_cpu_t, X28), .size = 64},
+	{.name = "X29", .offset = offsetof(vm_cpu_t, X29), .size = 64},
+	{.name = "LR", .offset = offsetof(vm_cpu_t, LR), .size = 64},
+
+	{.name = "SP", .offset = offsetof(vm_cpu_t, SP), .size = 64},
+	{.name = "PC", .offset = offsetof(vm_cpu_t, PC), .size = 64},
+
+	{.name = "zf", .offset = offsetof(vm_cpu_t, zf), .size = 8},
+	{.name = "nf", .offset = offsetof(vm_cpu_t, nf), .size = 8},
+	{.name = "of", .offset = offsetof(vm_cpu_t, of), .size = 8},
+	{.name = "cf", .offset = offsetof(vm_cpu_t, cf), .size = 8},
+
+	{.name = "exception_flags", .offset = offsetof(vm_cpu_t, exception_flags), .size = 32},
+	{.name = "interrupt_num", .offset = offsetof(vm_cpu_t, interrupt_num), .size = 32},
 
 };
 
diff --git a/miasm2/jitter/arch/JitCore_arm.c b/miasm2/jitter/arch/JitCore_arm.c
index f253c45b..ac7d16bf 100644
--- a/miasm2/jitter/arch/JitCore_arm.c
+++ b/miasm2/jitter/arch/JitCore_arm.c
@@ -12,35 +12,36 @@
 
 
 
-reg_dict gpreg_dict[] = { {.name = "R0", .offset = offsetof(vm_cpu_t, R0)},
-			  {.name = "R1", .offset = offsetof(vm_cpu_t, R1)},
-			  {.name = "R2", .offset = offsetof(vm_cpu_t, R2)},
-			  {.name = "R3", .offset = offsetof(vm_cpu_t, R3)},
-			  {.name = "R4", .offset = offsetof(vm_cpu_t, R4)},
-			  {.name = "R5", .offset = offsetof(vm_cpu_t, R5)},
-			  {.name = "R6", .offset = offsetof(vm_cpu_t, R6)},
-			  {.name = "R7", .offset = offsetof(vm_cpu_t, R7)},
-			  {.name = "R8", .offset = offsetof(vm_cpu_t, R8)},
-			  {.name = "R9", .offset = offsetof(vm_cpu_t, R9)},
-			  {.name = "R10", .offset = offsetof(vm_cpu_t, R10)},
-			  {.name = "R11", .offset = offsetof(vm_cpu_t, R11)},
-			  {.name = "R12", .offset = offsetof(vm_cpu_t, R12)},
-			  {.name = "SP", .offset = offsetof(vm_cpu_t, SP)},
-			  {.name = "LR", .offset = offsetof(vm_cpu_t, LR)},
-			  {.name = "PC", .offset = offsetof(vm_cpu_t, PC)},
-
-			  {.name = "zf", .offset = offsetof(vm_cpu_t, zf)},
-			  {.name = "nf", .offset = offsetof(vm_cpu_t, nf)},
-			  {.name = "of", .offset = offsetof(vm_cpu_t, of)},
-			  {.name = "cf", .offset = offsetof(vm_cpu_t, cf)},
-
-			  {.name = "ge0", .offset = offsetof(vm_cpu_t, ge0)},
-			  {.name = "ge1", .offset = offsetof(vm_cpu_t, ge1)},
-			  {.name = "ge2", .offset = offsetof(vm_cpu_t, ge2)},
-			  {.name = "ge3", .offset = offsetof(vm_cpu_t, ge3)},
-
-        {.name = "exception_flags", .offset = offsetof(vm_cpu_t, exception_flags)},
-        {.name = "interrupt_num", .offset = offsetof(vm_cpu_t, interrupt_num)},
+reg_dict gpreg_dict[] = {
+			 {.name = "R0", .offset = offsetof(vm_cpu_t, R0), .size = 32},
+			 {.name = "R1", .offset = offsetof(vm_cpu_t, R1), .size = 32},
+			 {.name = "R2", .offset = offsetof(vm_cpu_t, R2), .size = 32},
+			 {.name = "R3", .offset = offsetof(vm_cpu_t, R3), .size = 32},
+			 {.name = "R4", .offset = offsetof(vm_cpu_t, R4), .size = 32},
+			 {.name = "R5", .offset = offsetof(vm_cpu_t, R5), .size = 32},
+			 {.name = "R6", .offset = offsetof(vm_cpu_t, R6), .size = 32},
+			 {.name = "R7", .offset = offsetof(vm_cpu_t, R7), .size = 32},
+			 {.name = "R8", .offset = offsetof(vm_cpu_t, R8), .size = 32},
+			 {.name = "R9", .offset = offsetof(vm_cpu_t, R9), .size = 32},
+			 {.name = "R10", .offset = offsetof(vm_cpu_t, R10), .size = 32},
+			 {.name = "R11", .offset = offsetof(vm_cpu_t, R11), .size = 32},
+			 {.name = "R12", .offset = offsetof(vm_cpu_t, R12), .size = 32},
+			 {.name = "SP", .offset = offsetof(vm_cpu_t, SP), .size = 32},
+			 {.name = "LR", .offset = offsetof(vm_cpu_t, LR), .size = 32},
+			 {.name = "PC", .offset = offsetof(vm_cpu_t, PC), .size = 32},
+
+			 {.name = "zf", .offset = offsetof(vm_cpu_t, zf), .size = 8},
+			 {.name = "nf", .offset = offsetof(vm_cpu_t, nf), .size = 8},
+			 {.name = "of", .offset = offsetof(vm_cpu_t, of), .size = 8},
+			 {.name = "cf", .offset = offsetof(vm_cpu_t, cf), .size = 8},
+
+			 {.name = "ge0", .offset = offsetof(vm_cpu_t, ge0), .size = 8},
+			 {.name = "ge1", .offset = offsetof(vm_cpu_t, ge1), .size = 8},
+			 {.name = "ge2", .offset = offsetof(vm_cpu_t, ge2), .size = 8},
+			 {.name = "ge3", .offset = offsetof(vm_cpu_t, ge3), .size = 8},
+
+			 {.name = "exception_flags", .offset = offsetof(vm_cpu_t, exception_flags), .size = 32},
+			 {.name = "interrupt_num", .offset = offsetof(vm_cpu_t, interrupt_num), .size = 32},
 };
 
 /************************** JitCpu object **************************/
diff --git a/miasm2/jitter/arch/JitCore_mep.c b/miasm2/jitter/arch/JitCore_mep.c
index dff65dd5..da070016 100644
--- a/miasm2/jitter/arch/JitCore_mep.c
+++ b/miasm2/jitter/arch/JitCore_mep.c
@@ -14,63 +14,63 @@
 
 
 reg_dict gpreg_dict[] = {
-	{.name = "R0", .offset = offsetof(vm_cpu_t, R0)},
-	{.name = "R1", .offset = offsetof(vm_cpu_t, R1)},
-	{.name = "R2", .offset = offsetof(vm_cpu_t, R2)},
-	{.name = "R3", .offset = offsetof(vm_cpu_t, R3)},
-	{.name = "R4", .offset = offsetof(vm_cpu_t, R4)},
-	{.name = "R5", .offset = offsetof(vm_cpu_t, R5)},
-	{.name = "R6", .offset = offsetof(vm_cpu_t, R6)},
-	{.name = "R7", .offset = offsetof(vm_cpu_t, R7)},
-	{.name = "R8", .offset = offsetof(vm_cpu_t, R8)},
-	{.name = "R9", .offset = offsetof(vm_cpu_t, R9)},
-	{.name = "R10", .offset = offsetof(vm_cpu_t, R10)},
-	{.name = "R11", .offset = offsetof(vm_cpu_t, R11)},
-	{.name = "R12", .offset = offsetof(vm_cpu_t, R12)},
-	{.name = "TP", .offset = offsetof(vm_cpu_t, TP)},
-	{.name = "GP", .offset = offsetof(vm_cpu_t, GP)},
-	{.name = "SP", .offset = offsetof(vm_cpu_t, SP)},
-
-	{.name = "PC", .offset = offsetof(vm_cpu_t, PC)},
-	{.name = "LP", .offset = offsetof(vm_cpu_t, LP)},
-	{.name = "SAR", .offset = offsetof(vm_cpu_t, SAR)},
-	{.name = "S3", .offset = offsetof(vm_cpu_t, S3)},
-	{.name = "RPB", .offset = offsetof(vm_cpu_t, RPB)},
-	{.name = "RPE", .offset = offsetof(vm_cpu_t, RPE)},
-	{.name = "RPC", .offset = offsetof(vm_cpu_t, RPC)},
-	{.name = "HI", .offset = offsetof(vm_cpu_t, HI)},
-	{.name = "LO", .offset = offsetof(vm_cpu_t, LO)},
-	{.name = "S9", .offset = offsetof(vm_cpu_t, S9)},
-	{.name = "S10", .offset = offsetof(vm_cpu_t, S10)},
-	{.name = "S11", .offset = offsetof(vm_cpu_t, S11)},
-	{.name = "MB0", .offset = offsetof(vm_cpu_t, MB0)},
-	{.name = "ME0", .offset = offsetof(vm_cpu_t, ME0)},
-	{.name = "MB1", .offset = offsetof(vm_cpu_t, MB1)},
-	{.name = "ME1", .offset = offsetof(vm_cpu_t, ME1)},
-	{.name = "PSW", .offset = offsetof(vm_cpu_t, PSW)},
-	{.name = "ID", .offset = offsetof(vm_cpu_t, ID)},
-	{.name = "TMP", .offset = offsetof(vm_cpu_t, TMP)},
-	{.name = "EPC", .offset = offsetof(vm_cpu_t, EPC)},
-	{.name = "EXC", .offset = offsetof(vm_cpu_t, EXC)},
-	{.name = "CFG", .offset = offsetof(vm_cpu_t, CFG)},
-	{.name = "S22", .offset = offsetof(vm_cpu_t, S22)},
-	{.name = "NPC", .offset = offsetof(vm_cpu_t, NPC)},
-	{.name = "DBG", .offset = offsetof(vm_cpu_t, DBG)},
-	{.name = "DEPC", .offset = offsetof(vm_cpu_t, DEPC)},
-	{.name = "OPT", .offset = offsetof(vm_cpu_t, OPT)},
-	{.name = "RCFG", .offset = offsetof(vm_cpu_t, RCFG)},
-	{.name = "CCFG", .offset = offsetof(vm_cpu_t, CCFG)},
-	{.name = "S29", .offset = offsetof(vm_cpu_t, S29)},
-	{.name = "S30", .offset = offsetof(vm_cpu_t, S30)},
-	{.name = "S31", .offset = offsetof(vm_cpu_t, S31)},
-	{.name = "S32", .offset = offsetof(vm_cpu_t, S32)},
-	{.name = "take_jmp", .offset = offsetof(vm_cpu_t, take_jmp)},
-	{.name = "last_addr", .offset = offsetof(vm_cpu_t, last_addr)},
-	{.name = "is_repeat_end", .offset = offsetof(vm_cpu_t, is_repeat_end)},
-
-	{.name = "PC_end", .offset = offsetof(vm_cpu_t, PC_end)},
-	{.name = "RPE_instr_count", .offset = offsetof(vm_cpu_t, RPE_instr_count)},
-	{.name = "RPC_current", .offset = offsetof(vm_cpu_t, RPC_current)},
+	{.name = "R0", .offset = offsetof(vm_cpu_t, R0), .size = 32},
+	{.name = "R1", .offset = offsetof(vm_cpu_t, R1), .size = 32},
+	{.name = "R2", .offset = offsetof(vm_cpu_t, R2), .size = 32},
+	{.name = "R3", .offset = offsetof(vm_cpu_t, R3), .size = 32},
+	{.name = "R4", .offset = offsetof(vm_cpu_t, R4), .size = 32},
+	{.name = "R5", .offset = offsetof(vm_cpu_t, R5), .size = 32},
+	{.name = "R6", .offset = offsetof(vm_cpu_t, R6), .size = 32},
+	{.name = "R7", .offset = offsetof(vm_cpu_t, R7), .size = 32},
+	{.name = "R8", .offset = offsetof(vm_cpu_t, R8), .size = 32},
+	{.name = "R9", .offset = offsetof(vm_cpu_t, R9), .size = 32},
+	{.name = "R10", .offset = offsetof(vm_cpu_t, R10), .size = 32},
+	{.name = "R11", .offset = offsetof(vm_cpu_t, R11), .size = 32},
+	{.name = "R12", .offset = offsetof(vm_cpu_t, R12), .size = 32},
+	{.name = "TP", .offset = offsetof(vm_cpu_t, TP), .size = 32},
+	{.name = "GP", .offset = offsetof(vm_cpu_t, GP), .size = 32},
+	{.name = "SP", .offset = offsetof(vm_cpu_t, SP), .size = 32},
+
+	{.name = "PC", .offset = offsetof(vm_cpu_t, PC), .size = 32},
+	{.name = "LP", .offset = offsetof(vm_cpu_t, LP), .size = 32},
+	{.name = "SAR", .offset = offsetof(vm_cpu_t, SAR), .size = 32},
+	{.name = "S3", .offset = offsetof(vm_cpu_t, S3), .size = 32},
+	{.name = "RPB", .offset = offsetof(vm_cpu_t, RPB), .size = 32},
+	{.name = "RPE", .offset = offsetof(vm_cpu_t, RPE), .size = 32},
+	{.name = "RPC", .offset = offsetof(vm_cpu_t, RPC), .size = 32},
+	{.name = "HI", .offset = offsetof(vm_cpu_t, HI), .size = 32},
+	{.name = "LO", .offset = offsetof(vm_cpu_t, LO), .size = 32},
+	{.name = "S9", .offset = offsetof(vm_cpu_t, S9), .size = 32},
+	{.name = "S10", .offset = offsetof(vm_cpu_t, S10), .size = 32},
+	{.name = "S11", .offset = offsetof(vm_cpu_t, S11), .size = 32},
+	{.name = "MB0", .offset = offsetof(vm_cpu_t, MB0), .size = 32},
+	{.name = "ME0", .offset = offsetof(vm_cpu_t, ME0), .size = 32},
+	{.name = "MB1", .offset = offsetof(vm_cpu_t, MB1), .size = 32},
+	{.name = "ME1", .offset = offsetof(vm_cpu_t, ME1), .size = 32},
+	{.name = "PSW", .offset = offsetof(vm_cpu_t, PSW), .size = 32},
+	{.name = "ID", .offset = offsetof(vm_cpu_t, ID), .size = 32},
+	{.name = "TMP", .offset = offsetof(vm_cpu_t, TMP), .size = 32},
+	{.name = "EPC", .offset = offsetof(vm_cpu_t, EPC), .size = 32},
+	{.name = "EXC", .offset = offsetof(vm_cpu_t, EXC), .size = 32},
+	{.name = "CFG", .offset = offsetof(vm_cpu_t, CFG), .size = 32},
+	{.name = "S22", .offset = offsetof(vm_cpu_t, S22), .size = 32},
+	{.name = "NPC", .offset = offsetof(vm_cpu_t, NPC), .size = 32},
+	{.name = "DBG", .offset = offsetof(vm_cpu_t, DBG), .size = 32},
+	{.name = "DEPC", .offset = offsetof(vm_cpu_t, DEPC), .size = 32},
+	{.name = "OPT", .offset = offsetof(vm_cpu_t, OPT), .size = 32},
+	{.name = "RCFG", .offset = offsetof(vm_cpu_t, RCFG), .size = 32},
+	{.name = "CCFG", .offset = offsetof(vm_cpu_t, CCFG), .size = 32},
+	{.name = "S29", .offset = offsetof(vm_cpu_t, S29), .size = 32},
+	{.name = "S30", .offset = offsetof(vm_cpu_t, S30), .size = 32},
+	{.name = "S31", .offset = offsetof(vm_cpu_t, S31), .size = 32},
+	{.name = "S32", .offset = offsetof(vm_cpu_t, S32), .size = 32},
+	{.name = "take_jmp", .offset = offsetof(vm_cpu_t, take_jmp), .size = 32},
+	{.name = "last_addr", .offset = offsetof(vm_cpu_t, last_addr), .size = 32},
+	{.name = "is_repeat_end", .offset = offsetof(vm_cpu_t, is_repeat_end), .size = 32},
+
+	{.name = "PC_end", .offset = offsetof(vm_cpu_t, PC_end), .size = 32},
+	{.name = "RPE_instr_count", .offset = offsetof(vm_cpu_t, RPE_instr_count), .size = 32},
+	{.name = "RPC_current", .offset = offsetof(vm_cpu_t, RPC_current), .size = 32},
 
 };
 
diff --git a/miasm2/jitter/arch/JitCore_mips32.c b/miasm2/jitter/arch/JitCore_mips32.c
index a1bf44f5..2a5dfd0f 100644
--- a/miasm2/jitter/arch/JitCore_mips32.c
+++ b/miasm2/jitter/arch/JitCore_mips32.c
@@ -12,42 +12,42 @@
 
 
 
-reg_dict gpreg_dict[] = { {.name = "ZERO", .offset = offsetof(vm_cpu_t, ZERO)},
-			  {.name = "AT", .offset = offsetof(vm_cpu_t, AT)},
-			  {.name = "V0", .offset = offsetof(vm_cpu_t, V0)},
-			  {.name = "V1", .offset = offsetof(vm_cpu_t, V1)},
-			  {.name = "A0", .offset = offsetof(vm_cpu_t, A0)},
-			  {.name = "A1", .offset = offsetof(vm_cpu_t, A1)},
-			  {.name = "A2", .offset = offsetof(vm_cpu_t, A2)},
-			  {.name = "A3", .offset = offsetof(vm_cpu_t, A3)},
-			  {.name = "T0", .offset = offsetof(vm_cpu_t, T0)},
-			  {.name = "T1", .offset = offsetof(vm_cpu_t, T1)},
-			  {.name = "T2", .offset = offsetof(vm_cpu_t, T2)},
-			  {.name = "T3", .offset = offsetof(vm_cpu_t, T3)},
-			  {.name = "T4", .offset = offsetof(vm_cpu_t, T4)},
-			  {.name = "T5", .offset = offsetof(vm_cpu_t, T5)},
-			  {.name = "T6", .offset = offsetof(vm_cpu_t, T6)},
-			  {.name = "T7", .offset = offsetof(vm_cpu_t, T7)},
-			  {.name = "S0", .offset = offsetof(vm_cpu_t, S0)},
-			  {.name = "S1", .offset = offsetof(vm_cpu_t, S1)},
-			  {.name = "S2", .offset = offsetof(vm_cpu_t, S2)},
-			  {.name = "S3", .offset = offsetof(vm_cpu_t, S3)},
-			  {.name = "S4", .offset = offsetof(vm_cpu_t, S4)},
-			  {.name = "S5", .offset = offsetof(vm_cpu_t, S5)},
-			  {.name = "S6", .offset = offsetof(vm_cpu_t, S6)},
-			  {.name = "S7", .offset = offsetof(vm_cpu_t, S7)},
-			  {.name = "T8", .offset = offsetof(vm_cpu_t, T8)},
-			  {.name = "T9", .offset = offsetof(vm_cpu_t, T9)},
-			  {.name = "K0", .offset = offsetof(vm_cpu_t, K0)},
-			  {.name = "K1", .offset = offsetof(vm_cpu_t, K1)},
-			  {.name = "GP", .offset = offsetof(vm_cpu_t, GP)},
-			  {.name = "SP", .offset = offsetof(vm_cpu_t, SP)},
-			  {.name = "FP", .offset = offsetof(vm_cpu_t, FP)},
-			  {.name = "RA", .offset = offsetof(vm_cpu_t, RA)},
-			  {.name = "PC", .offset = offsetof(vm_cpu_t, PC)},
-			  {.name = "PC_FETCH", .offset = offsetof(vm_cpu_t, PC_FETCH)},
-			  {.name = "R_LO", .offset = offsetof(vm_cpu_t, R_LO)},
-			  {.name = "R_HI", .offset = offsetof(vm_cpu_t, R_HI)},
+reg_dict gpreg_dict[] = { {.name = "ZERO", .offset = offsetof(vm_cpu_t, ZERO), .size = 32},
+			  {.name = "AT", .offset = offsetof(vm_cpu_t, AT), .size = 32},
+			  {.name = "V0", .offset = offsetof(vm_cpu_t, V0), .size = 32},
+			  {.name = "V1", .offset = offsetof(vm_cpu_t, V1), .size = 32},
+			  {.name = "A0", .offset = offsetof(vm_cpu_t, A0), .size = 32},
+			  {.name = "A1", .offset = offsetof(vm_cpu_t, A1), .size = 32},
+			  {.name = "A2", .offset = offsetof(vm_cpu_t, A2), .size = 32},
+			  {.name = "A3", .offset = offsetof(vm_cpu_t, A3), .size = 32},
+			  {.name = "T0", .offset = offsetof(vm_cpu_t, T0), .size = 32},
+			  {.name = "T1", .offset = offsetof(vm_cpu_t, T1), .size = 32},
+			  {.name = "T2", .offset = offsetof(vm_cpu_t, T2), .size = 32},
+			  {.name = "T3", .offset = offsetof(vm_cpu_t, T3), .size = 32},
+			  {.name = "T4", .offset = offsetof(vm_cpu_t, T4), .size = 32},
+			  {.name = "T5", .offset = offsetof(vm_cpu_t, T5), .size = 32},
+			  {.name = "T6", .offset = offsetof(vm_cpu_t, T6), .size = 32},
+			  {.name = "T7", .offset = offsetof(vm_cpu_t, T7), .size = 32},
+			  {.name = "S0", .offset = offsetof(vm_cpu_t, S0), .size = 32},
+			  {.name = "S1", .offset = offsetof(vm_cpu_t, S1), .size = 32},
+			  {.name = "S2", .offset = offsetof(vm_cpu_t, S2), .size = 32},
+			  {.name = "S3", .offset = offsetof(vm_cpu_t, S3), .size = 32},
+			  {.name = "S4", .offset = offsetof(vm_cpu_t, S4), .size = 32},
+			  {.name = "S5", .offset = offsetof(vm_cpu_t, S5), .size = 32},
+			  {.name = "S6", .offset = offsetof(vm_cpu_t, S6), .size = 32},
+			  {.name = "S7", .offset = offsetof(vm_cpu_t, S7), .size = 32},
+			  {.name = "T8", .offset = offsetof(vm_cpu_t, T8), .size = 32},
+			  {.name = "T9", .offset = offsetof(vm_cpu_t, T9), .size = 32},
+			  {.name = "K0", .offset = offsetof(vm_cpu_t, K0), .size = 32},
+			  {.name = "K1", .offset = offsetof(vm_cpu_t, K1), .size = 32},
+			  {.name = "GP", .offset = offsetof(vm_cpu_t, GP), .size = 32},
+			  {.name = "SP", .offset = offsetof(vm_cpu_t, SP), .size = 32},
+			  {.name = "FP", .offset = offsetof(vm_cpu_t, FP), .size = 32},
+			  {.name = "RA", .offset = offsetof(vm_cpu_t, RA), .size = 32},
+			  {.name = "PC", .offset = offsetof(vm_cpu_t, PC), .size = 32},
+			  {.name = "PC_FETCH", .offset = offsetof(vm_cpu_t, PC_FETCH), .size = 32},
+			  {.name = "R_LO", .offset = offsetof(vm_cpu_t, R_LO), .size = 32},
+			  {.name = "R_HI", .offset = offsetof(vm_cpu_t, R_HI), .size = 32},
 };
 
 /************************** JitCpu object **************************/
diff --git a/miasm2/jitter/arch/JitCore_ppc32.c b/miasm2/jitter/arch/JitCore_ppc32.c
index 3acf769a..dfc46c91 100644
--- a/miasm2/jitter/arch/JitCore_ppc32.c
+++ b/miasm2/jitter/arch/JitCore_ppc32.c
@@ -11,7 +11,7 @@
 
 reg_dict gpreg_dict[] = {
 #define JITCORE_PPC_REG_EXPAND(_name, _size)				\
-    { .name = #_name, .offset = offsetof(struct vm_cpu, _name) },
+    { .name = #_name, .offset = offsetof(struct vm_cpu, _name), .size = _size },
 #include "JitCore_ppc32_regs.h"
 #undef JITCORE_PPC_REG_EXPAND
 };
diff --git a/miasm2/jitter/arch/JitCore_x86.c b/miasm2/jitter/arch/JitCore_x86.c
index b8d8270d..1782c5ae 100644
--- a/miasm2/jitter/arch/JitCore_x86.c
+++ b/miasm2/jitter/arch/JitCore_x86.c
@@ -12,72 +12,73 @@
 
 vm_cpu_t ref_arch_regs;
 
-reg_dict gpreg_dict[] = { {.name = "RAX", .offset = offsetof(vm_cpu_t, RAX), .size = 8*sizeof(ref_arch_regs.RAX)},
-			  {.name = "RBX", .offset = offsetof(vm_cpu_t, RBX), .size = 8*sizeof(ref_arch_regs.RBX)},
-			  {.name = "RCX", .offset = offsetof(vm_cpu_t, RCX), .size = 8*sizeof(ref_arch_regs.RCX)},
-			  {.name = "RDX", .offset = offsetof(vm_cpu_t, RDX), .size = 8*sizeof(ref_arch_regs.RDX)},
-			  {.name = "RSI", .offset = offsetof(vm_cpu_t, RSI), .size = 8*sizeof(ref_arch_regs.RSI)},
-			  {.name = "RDI", .offset = offsetof(vm_cpu_t, RDI), .size = 8*sizeof(ref_arch_regs.RDI)},
-			  {.name = "RSP", .offset = offsetof(vm_cpu_t, RSP), .size = 8*sizeof(ref_arch_regs.RSP)},
-			  {.name = "RBP", .offset = offsetof(vm_cpu_t, RBP), .size = 8*sizeof(ref_arch_regs.RBP)},
-
-			  {.name = "R8", .offset = offsetof(vm_cpu_t, R8), .size = 8*sizeof(ref_arch_regs.R8)},
-			  {.name = "R9", .offset = offsetof(vm_cpu_t, R9), .size = 8*sizeof(ref_arch_regs.R9)},
-			  {.name = "R10", .offset = offsetof(vm_cpu_t, R10), .size = 8*sizeof(ref_arch_regs.R10)},
-			  {.name = "R11", .offset = offsetof(vm_cpu_t, R11), .size = 8*sizeof(ref_arch_regs.R11)},
-			  {.name = "R12", .offset = offsetof(vm_cpu_t, R12), .size = 8*sizeof(ref_arch_regs.R12)},
-			  {.name = "R13", .offset = offsetof(vm_cpu_t, R13), .size = 8*sizeof(ref_arch_regs.R13)},
-			  {.name = "R14", .offset = offsetof(vm_cpu_t, R14), .size = 8*sizeof(ref_arch_regs.R14)},
-			  {.name = "R15", .offset = offsetof(vm_cpu_t, R15), .size = 8*sizeof(ref_arch_regs.R15)},
-
-			  {.name = "RIP", .offset = offsetof(vm_cpu_t, RIP), .size = 8*sizeof(ref_arch_regs.RIP)},
-
-			  {.name = "zf", .offset = offsetof(vm_cpu_t, zf), .size = 8*sizeof(ref_arch_regs.zf)},
-			  {.name = "nf", .offset = offsetof(vm_cpu_t, nf), .size = 8*sizeof(ref_arch_regs.nf)},
-			  {.name = "pf", .offset = offsetof(vm_cpu_t, pf), .size = 8*sizeof(ref_arch_regs.pf)},
-			  {.name = "of", .offset = offsetof(vm_cpu_t, of), .size = 8*sizeof(ref_arch_regs.of)},
-			  {.name = "cf", .offset = offsetof(vm_cpu_t, cf), .size = 8*sizeof(ref_arch_regs.cf)},
-			  {.name = "af", .offset = offsetof(vm_cpu_t, af), .size = 8*sizeof(ref_arch_regs.af)},
-			  {.name = "df", .offset = offsetof(vm_cpu_t, df), .size = 8*sizeof(ref_arch_regs.df)},
-
-			  {.name = "ES", .offset = offsetof(vm_cpu_t, ES), .size = 8*sizeof(ref_arch_regs.ES)},
-			  {.name = "CS", .offset = offsetof(vm_cpu_t, CS), .size = 8*sizeof(ref_arch_regs.CS)},
-			  {.name = "SS", .offset = offsetof(vm_cpu_t, SS), .size = 8*sizeof(ref_arch_regs.SS)},
-			  {.name = "DS", .offset = offsetof(vm_cpu_t, DS), .size = 8*sizeof(ref_arch_regs.DS)},
-			  {.name = "FS", .offset = offsetof(vm_cpu_t, FS), .size = 8*sizeof(ref_arch_regs.FS)},
-			  {.name = "GS", .offset = offsetof(vm_cpu_t, GS), .size = 8*sizeof(ref_arch_regs.GS)},
-
-			  {.name = "MM0", .offset = offsetof(vm_cpu_t, MM0), .size = 8*sizeof(ref_arch_regs.MM0)},
-			  {.name = "MM1", .offset = offsetof(vm_cpu_t, MM1), .size = 8*sizeof(ref_arch_regs.MM1)},
-			  {.name = "MM2", .offset = offsetof(vm_cpu_t, MM2), .size = 8*sizeof(ref_arch_regs.MM2)},
-			  {.name = "MM3", .offset = offsetof(vm_cpu_t, MM3), .size = 8*sizeof(ref_arch_regs.MM3)},
-			  {.name = "MM4", .offset = offsetof(vm_cpu_t, MM4), .size = 8*sizeof(ref_arch_regs.MM4)},
-			  {.name = "MM5", .offset = offsetof(vm_cpu_t, MM5), .size = 8*sizeof(ref_arch_regs.MM5)},
-			  {.name = "MM6", .offset = offsetof(vm_cpu_t, MM6), .size = 8*sizeof(ref_arch_regs.MM6)},
-			  {.name = "MM7", .offset = offsetof(vm_cpu_t, MM7), .size = 8*sizeof(ref_arch_regs.MM7)},
-
-			  {.name = "XMM0", .offset = offsetof(vm_cpu_t, XMM0), .size = 128},
-			  {.name = "XMM1", .offset = offsetof(vm_cpu_t, XMM1), .size = 128},
-			  {.name = "XMM2", .offset = offsetof(vm_cpu_t, XMM2), .size = 128},
-			  {.name = "XMM3", .offset = offsetof(vm_cpu_t, XMM3), .size = 128},
-			  {.name = "XMM4", .offset = offsetof(vm_cpu_t, XMM4), .size = 128},
-			  {.name = "XMM5", .offset = offsetof(vm_cpu_t, XMM5), .size = 128},
-			  {.name = "XMM6", .offset = offsetof(vm_cpu_t, XMM6), .size = 128},
-			  {.name = "XMM7", .offset = offsetof(vm_cpu_t, XMM7), .size = 128},
-			  {.name = "XMM8", .offset = offsetof(vm_cpu_t, XMM8), .size = 128},
-			  {.name = "XMM9", .offset = offsetof(vm_cpu_t, XMM9), .size = 128},
-			  {.name = "XMM10", .offset = offsetof(vm_cpu_t, XMM10), .size = 128},
-			  {.name = "XMM11", .offset = offsetof(vm_cpu_t, XMM11), .size = 128},
-			  {.name = "XMM12", .offset = offsetof(vm_cpu_t, XMM12), .size = 128},
-			  {.name = "XMM13", .offset = offsetof(vm_cpu_t, XMM13), .size = 128},
-			  {.name = "XMM14", .offset = offsetof(vm_cpu_t, XMM14), .size = 128},
-			  {.name = "XMM15", .offset = offsetof(vm_cpu_t, XMM15), .size = 128},
-
-			  {.name = "tsc1", .offset = offsetof(vm_cpu_t, tsc1), .size = 8*sizeof(ref_arch_regs.tsc1)},
-			  {.name = "tsc2", .offset = offsetof(vm_cpu_t, tsc2), .size = 8*sizeof(ref_arch_regs.tsc2)},
-
-			  {.name = "exception_flags", .offset = offsetof(vm_cpu_t, exception_flags), .size = 8*sizeof(ref_arch_regs.exception_flags)},
-			  {.name = "interrupt_num", .offset = offsetof(vm_cpu_t, interrupt_num), .size = 8*sizeof(ref_arch_regs.interrupt_num)},
+reg_dict gpreg_dict[] = {
+			 {.name = "RAX", .offset = offsetof(vm_cpu_t, RAX), .size = 64},
+			 {.name = "RBX", .offset = offsetof(vm_cpu_t, RBX), .size = 64},
+			 {.name = "RCX", .offset = offsetof(vm_cpu_t, RCX), .size = 64},
+			 {.name = "RDX", .offset = offsetof(vm_cpu_t, RDX), .size = 64},
+			 {.name = "RSI", .offset = offsetof(vm_cpu_t, RSI), .size = 64},
+			 {.name = "RDI", .offset = offsetof(vm_cpu_t, RDI), .size = 64},
+			 {.name = "RSP", .offset = offsetof(vm_cpu_t, RSP), .size = 64},
+			 {.name = "RBP", .offset = offsetof(vm_cpu_t, RBP), .size = 64},
+
+			 {.name = "R8", .offset = offsetof(vm_cpu_t, R8), .size = 64},
+			 {.name = "R9", .offset = offsetof(vm_cpu_t, R9), .size = 64},
+			 {.name = "R10", .offset = offsetof(vm_cpu_t, R10), .size = 64},
+			 {.name = "R11", .offset = offsetof(vm_cpu_t, R11), .size = 64},
+			 {.name = "R12", .offset = offsetof(vm_cpu_t, R12), .size = 64},
+			 {.name = "R13", .offset = offsetof(vm_cpu_t, R13), .size = 64},
+			 {.name = "R14", .offset = offsetof(vm_cpu_t, R14), .size = 64},
+			 {.name = "R15", .offset = offsetof(vm_cpu_t, R15), .size = 64},
+
+			 {.name = "RIP", .offset = offsetof(vm_cpu_t, RIP), .size = 64},
+
+			 {.name = "zf", .offset = offsetof(vm_cpu_t, zf), .size = 8},
+			 {.name = "nf", .offset = offsetof(vm_cpu_t, nf), .size = 8},
+			 {.name = "pf", .offset = offsetof(vm_cpu_t, pf), .size = 8},
+			 {.name = "of", .offset = offsetof(vm_cpu_t, of), .size = 8},
+			 {.name = "cf", .offset = offsetof(vm_cpu_t, cf), .size = 8},
+			 {.name = "af", .offset = offsetof(vm_cpu_t, af), .size = 8},
+			 {.name = "df", .offset = offsetof(vm_cpu_t, df), .size = 8},
+
+			 {.name = "ES", .offset = offsetof(vm_cpu_t, ES), .size = 16},
+			 {.name = "CS", .offset = offsetof(vm_cpu_t, CS), .size = 16},
+			 {.name = "SS", .offset = offsetof(vm_cpu_t, SS), .size = 16},
+			 {.name = "DS", .offset = offsetof(vm_cpu_t, DS), .size = 16},
+			 {.name = "FS", .offset = offsetof(vm_cpu_t, FS), .size = 16},
+			 {.name = "GS", .offset = offsetof(vm_cpu_t, GS), .size = 16},
+
+			 {.name = "MM0", .offset = offsetof(vm_cpu_t, MM0), .size = 64},
+			 {.name = "MM1", .offset = offsetof(vm_cpu_t, MM1), .size = 64},
+			 {.name = "MM2", .offset = offsetof(vm_cpu_t, MM2), .size = 64},
+			 {.name = "MM3", .offset = offsetof(vm_cpu_t, MM3), .size = 64},
+			 {.name = "MM4", .offset = offsetof(vm_cpu_t, MM4), .size = 64},
+			 {.name = "MM5", .offset = offsetof(vm_cpu_t, MM5), .size = 64},
+			 {.name = "MM6", .offset = offsetof(vm_cpu_t, MM6), .size = 64},
+			 {.name = "MM7", .offset = offsetof(vm_cpu_t, MM7), .size = 64},
+
+			 {.name = "XMM0", .offset = offsetof(vm_cpu_t, XMM0), .size = 128},
+			 {.name = "XMM1", .offset = offsetof(vm_cpu_t, XMM1), .size = 128},
+			 {.name = "XMM2", .offset = offsetof(vm_cpu_t, XMM2), .size = 128},
+			 {.name = "XMM3", .offset = offsetof(vm_cpu_t, XMM3), .size = 128},
+			 {.name = "XMM4", .offset = offsetof(vm_cpu_t, XMM4), .size = 128},
+			 {.name = "XMM5", .offset = offsetof(vm_cpu_t, XMM5), .size = 128},
+			 {.name = "XMM6", .offset = offsetof(vm_cpu_t, XMM6), .size = 128},
+			 {.name = "XMM7", .offset = offsetof(vm_cpu_t, XMM7), .size = 128},
+			 {.name = "XMM8", .offset = offsetof(vm_cpu_t, XMM8), .size = 128},
+			 {.name = "XMM9", .offset = offsetof(vm_cpu_t, XMM9), .size = 128},
+			 {.name = "XMM10", .offset = offsetof(vm_cpu_t, XMM10), .size = 128},
+			 {.name = "XMM11", .offset = offsetof(vm_cpu_t, XMM11), .size = 128},
+			 {.name = "XMM12", .offset = offsetof(vm_cpu_t, XMM12), .size = 128},
+			 {.name = "XMM13", .offset = offsetof(vm_cpu_t, XMM13), .size = 128},
+			 {.name = "XMM14", .offset = offsetof(vm_cpu_t, XMM14), .size = 128},
+			 {.name = "XMM15", .offset = offsetof(vm_cpu_t, XMM15), .size = 128},
+
+			 {.name = "tsc1", .offset = offsetof(vm_cpu_t, tsc1), .size = 32},
+			 {.name = "tsc2", .offset = offsetof(vm_cpu_t, tsc2), .size = 32},
+
+			 {.name = "exception_flags", .offset = offsetof(vm_cpu_t, exception_flags), .size = 32},
+			 {.name = "interrupt_num", .offset = offsetof(vm_cpu_t, interrupt_num), .size = 32},
 };
 
 
@@ -138,22 +139,22 @@ PyObject* cpu_get_gpreg(JitCpu* self)
     get_reg(MM6);
     get_reg(MM7);
 
-    get_reg_bn(XMM0);
-    get_reg_bn(XMM1);
-    get_reg_bn(XMM2);
-    get_reg_bn(XMM3);
-    get_reg_bn(XMM4);
-    get_reg_bn(XMM5);
-    get_reg_bn(XMM6);
-    get_reg_bn(XMM7);
-    get_reg_bn(XMM8);
-    get_reg_bn(XMM9);
-    get_reg_bn(XMM10);
-    get_reg_bn(XMM11);
-    get_reg_bn(XMM12);
-    get_reg_bn(XMM13);
-    get_reg_bn(XMM14);
-    get_reg_bn(XMM15);
+    get_reg_bn(XMM0, 128);
+    get_reg_bn(XMM1, 128);
+    get_reg_bn(XMM2, 128);
+    get_reg_bn(XMM3, 128);
+    get_reg_bn(XMM4, 128);
+    get_reg_bn(XMM5, 128);
+    get_reg_bn(XMM6, 128);
+    get_reg_bn(XMM7, 128);
+    get_reg_bn(XMM8, 128);
+    get_reg_bn(XMM9, 128);
+    get_reg_bn(XMM10, 128);
+    get_reg_bn(XMM11, 128);
+    get_reg_bn(XMM12, 128);
+    get_reg_bn(XMM13, 128);
+    get_reg_bn(XMM14, 128);
+    get_reg_bn(XMM15, 128);
 
     get_reg(tsc1);
     get_reg(tsc2);
@@ -220,6 +221,7 @@ PyObject* cpu_set_gpreg(JitCpu* self, PyObject *args)
 						    py_long = PyLong_FromLong((long)tmp);
 					    } else if (PyLong_Check(py_long)){
 						    /* Already PyLong */
+						    /* Increment ref as we will decement it next */
 						    Py_INCREF(py_long);
 					    }
 					    else{
@@ -227,7 +229,6 @@ PyObject* cpu_set_gpreg(JitCpu* self, PyObject *args)
 					    }
 
 
-
 					    cst_ffffffff = PyLong_FromLong(0xffffffff);
 					    cst_32 = PyLong_FromLong(32);
 					    bn = bignum_from_int(0);
@@ -246,10 +247,7 @@ PyObject* cpu_set_gpreg(JitCpu* self, PyObject *args)
 					    Py_DECREF(py_long);
 					    Py_DECREF(cst_32);
 					    Py_DECREF(cst_ffffffff);
-
-
-
-					    *(bn_t*)(((char*)(self->cpu)) + gpreg_dict[i].offset) = bn;
+					    *(bn_t*)(((char*)(self->cpu)) + gpreg_dict[i].offset) = bignum_mask(bn, 128);
 				    }
 				    break;
 		    }
@@ -619,22 +617,22 @@ getset_reg_u64(MM5);
 getset_reg_u64(MM6);
 getset_reg_u64(MM7);
 
-getset_reg_bn(XMM0);
-getset_reg_bn(XMM1);
-getset_reg_bn(XMM2);
-getset_reg_bn(XMM3);
-getset_reg_bn(XMM4);
-getset_reg_bn(XMM5);
-getset_reg_bn(XMM6);
-getset_reg_bn(XMM7);
-getset_reg_bn(XMM8);
-getset_reg_bn(XMM9);
-getset_reg_bn(XMM10);
-getset_reg_bn(XMM11);
-getset_reg_bn(XMM12);
-getset_reg_bn(XMM13);
-getset_reg_bn(XMM14);
-getset_reg_bn(XMM15);
+getset_reg_bn(XMM0, 128);
+getset_reg_bn(XMM1, 128);
+getset_reg_bn(XMM2, 128);
+getset_reg_bn(XMM3, 128);
+getset_reg_bn(XMM4, 128);
+getset_reg_bn(XMM5, 128);
+getset_reg_bn(XMM6, 128);
+getset_reg_bn(XMM7, 128);
+getset_reg_bn(XMM8, 128);
+getset_reg_bn(XMM9, 128);
+getset_reg_bn(XMM10, 128);
+getset_reg_bn(XMM11, 128);
+getset_reg_bn(XMM12, 128);
+getset_reg_bn(XMM13, 128);
+getset_reg_bn(XMM14, 128);
+getset_reg_bn(XMM15, 128);
 
 getset_reg_u32(tsc1);
 getset_reg_u32(tsc2);
diff --git a/miasm2/jitter/bn.c b/miasm2/jitter/bn.c
index 9462de2c..96e66f4d 100644
--- a/miasm2/jitter/bn.c
+++ b/miasm2/jitter/bn.c
@@ -615,10 +615,9 @@ bn_t bignum_mask(bn_t src, int bits)
 	bn_t dst;
 	bn_t mask;
 
-	mask = bignum_from_int(1);
-	mask = bignum_lshift(mask, bits);
+	mask = bignum_from_int(0);
 	mask = bignum_dec(mask);
-
+	mask = bignum_rshift(mask, BN_BIT_SIZE - bits);
 	dst = bignum_and(src, mask);
 	return dst;
 }
diff --git a/test/arch/x86/unit/mn_getset128.py b/test/arch/x86/unit/mn_getset128.py
new file mode 100644
index 00000000..a084d663
--- /dev/null
+++ b/test/arch/x86/unit/mn_getset128.py
@@ -0,0 +1,52 @@
+#! /usr/bin/env python2
+
+import sys
+
+from asm_test import Asm_Test_32
+
+
+class Test_get_set_128(Asm_Test_32):
+    TXT = '''
+    main:
+       MOVD       XMM0, ESI
+       MOVD       XMM1, EDI
+       PCMPEQQ    XMM0, XMM1
+       JZ         ret
+       MOV        EAX, 1
+
+       PUSH       0x11112222
+       PUSH       0x33334444
+       PUSH       0x55556666
+       PUSH       0x77778888
+       MOVAPS     XMM2, XMMWORD PTR [ESP]
+       ADD        ESP, 0x10
+    ret:
+       RET
+    '''
+
+    def prepare(self):
+        val = 1
+        self.myjit.cpu.ESI = 0x11223344
+        self.myjit.cpu.EDI = 0x11223345
+        self.myjit.cpu.XMM0 = val
+
+        # Check 128 get / set
+        assert self.myjit.cpu.XMM0 == val
+        assert self.myjit.cpu.get_gpreg()['XMM0'] == val
+
+    def check(self):
+        assert self.myjit.cpu.XMM0 == 0xffffffffffffffff0000000000000000L
+        assert self.myjit.cpu.XMM1 == 0x11223345
+
+        # Check 128 get / set
+        assert self.myjit.cpu.get_gpreg()['XMM0'] == 0xffffffffffffffff0000000000000000L
+        assert self.myjit.cpu.get_gpreg()['XMM1'] == 0x11223345
+
+        assert self.myjit.cpu.get_gpreg()['XMM2'] == 0x11112222333344445555666677778888L
+        assert self.myjit.cpu.get_gpreg()['XMM2'] == 0x11112222333344445555666677778888L
+
+
+if __name__ == "__main__":
+    [test(*sys.argv[1:])() for test in [
+        Test_get_set_128,
+    ]]
diff --git a/test/arch/x86/unit/mn_pcmpeq.py b/test/arch/x86/unit/mn_pcmpeq.py
index 22760db4..e934d6b5 100755
--- a/test/arch/x86/unit/mn_pcmpeq.py
+++ b/test/arch/x86/unit/mn_pcmpeq.py
@@ -4,6 +4,7 @@ import sys
 
 from asm_test import Asm_Test_32
 
+
 class Test_PCMPEQB(Asm_Test_32):
     TXT = '''
     main:
@@ -42,7 +43,6 @@ class Test_PCMPEQW(Asm_Test_32):
         assert self.myjit.cpu.MM1 == 0xFFFF0000FFFF0000
 
 
-
 class Test_PCMPEQD(Asm_Test_32):
     TXT = '''
     main:
@@ -62,5 +62,33 @@ class Test_PCMPEQD(Asm_Test_32):
         assert self.myjit.cpu.MM1 == 0x00000000FFFFFFFF
 
 
+class Test_PCMPEQQ(Asm_Test_32):
+    TXT = '''
+    main:
+       MOVD       XMM0, ESI
+       MOVD       XMM1, EDI
+       PCMPEQQ    XMM0, XMM1
+       JZ         ret
+       MOV        EAX, 1
+    ret:
+       RET
+    '''
+
+    def prepare(self):
+        val = 1
+        self.myjit.cpu.ESI = 0x11223344
+        self.myjit.cpu.EDI = 0x11223345
+        self.myjit.cpu.XMM0 = val
+
+    def check(self):
+        assert self.myjit.cpu.XMM0 == 0xffffffffffffffff0000000000000000L
+        assert self.myjit.cpu.XMM1 == 0x11223345
+
+
 if __name__ == "__main__":
-    [test(*sys.argv[1:])() for test in [Test_PCMPEQB, Test_PCMPEQW, Test_PCMPEQD]]
+    [test(*sys.argv[1:])() for test in [
+        Test_PCMPEQB,
+        Test_PCMPEQW,
+        Test_PCMPEQD,
+        Test_PCMPEQQ,
+    ]]
diff --git a/test/test_all.py b/test/test_all.py
index d1ccb19f..77dd04cf 100755
--- a/test/test_all.py
+++ b/test/test_all.py
@@ -88,6 +88,7 @@ for script in ["x86/sem.py",
                "x86/unit/mn_cpuid.py",
                "x86/unit/mn_div.py",
                "x86/unit/test_asm_x86_64.py",
+               "x86/unit/mn_getset128.py",
                "arm/arch.py",
                "arm/sem.py",
                "aarch64/unit/mn_ubfm.py",