about summary refs log tree commit diff stats
path: root/miasm2/arch
diff options
context:
space:
mode:
Diffstat (limited to 'miasm2/arch')
-rw-r--r--miasm2/arch/msp430/ctype.py68
-rw-r--r--miasm2/arch/x86/arch.py110
-rw-r--r--miasm2/arch/x86/regs.py3
-rw-r--r--miasm2/arch/x86/sem.py113
4 files changed, 270 insertions, 24 deletions
diff --git a/miasm2/arch/msp430/ctype.py b/miasm2/arch/msp430/ctype.py
new file mode 100644
index 00000000..adb0a953
--- /dev/null
+++ b/miasm2/arch/msp430/ctype.py
@@ -0,0 +1,68 @@
+from miasm2.core.objc import CLeafTypes, ObjCDecl, PADDING_TYPE_NAME
+from miasm2.core.ctypesmngr import CTypeId, CTypePtr
+
+
+class CTypeMSP430_unk(CLeafTypes):
+    """Define C types sizes/alignement for msp430 architecture"""
+
+    obj_pad = ObjCDecl(PADDING_TYPE_NAME, 1, 1) # __padding__ is size 1/align 1
+
+    obj_char = ObjCDecl("char", 1, 1)
+    obj_short = ObjCDecl("short", 2, 2)
+    obj_int = ObjCDecl("int", 2, 2)
+    obj_long = ObjCDecl("long", 2, 2)
+
+    obj_uchar = ObjCDecl("uchar", 1, 1)
+    obj_ushort = ObjCDecl("ushort", 2, 2)
+    obj_uint = ObjCDecl("uint", 2, 2)
+    obj_ulong = ObjCDecl("ulong", 2, 2)
+
+    obj_void = ObjCDecl("void", 1, 1)
+
+    obj_enum = ObjCDecl("enum", 2, 2)
+
+    obj_float = ObjCDecl("float", 4, 4)
+    obj_double = ObjCDecl("double", 8, 8)
+    obj_ldouble = ObjCDecl("ldouble", 16, 16)
+
+    def __init__(self):
+        self.types = {
+            CTypeId(PADDING_TYPE_NAME): self.obj_pad,
+
+            CTypeId('char'): self.obj_char,
+            CTypeId('short'): self.obj_short,
+            CTypeId('int'): self.obj_int,
+            CTypeId('void'): self.obj_void,
+            CTypeId('long',): self.obj_long,
+            CTypeId('float'): self.obj_float,
+            CTypeId('double'): self.obj_double,
+
+            CTypeId('signed', 'char'): self.obj_char,
+            CTypeId('unsigned', 'char'): self.obj_uchar,
+
+            CTypeId('short', 'int'): self.obj_short,
+            CTypeId('signed', 'short'): self.obj_short,
+            CTypeId('signed', 'short', 'int'): self.obj_short,
+            CTypeId('unsigned', 'short'): self.obj_ushort,
+            CTypeId('unsigned', 'short', 'int'): self.obj_ushort,
+
+            CTypeId('unsigned', ): self.obj_uint,
+            CTypeId('unsigned', 'int'): self.obj_uint,
+            CTypeId('signed', 'int'): self.obj_int,
+
+            CTypeId('long', 'int'): self.obj_long,
+            CTypeId('long', 'long'): self.obj_long,
+            CTypeId('long', 'long', 'int'): self.obj_long,
+            CTypeId('signed', 'long', 'long'): self.obj_long,
+            CTypeId('unsigned', 'long', 'long'): self.obj_ulong,
+            CTypeId('signed', 'long', 'long', 'int'): self.obj_long,
+            CTypeId('unsigned', 'long', 'long', 'int'): self.obj_ulong,
+
+            CTypeId('signed', 'long'): self.obj_long,
+            CTypeId('unsigned', 'long'): self.obj_ulong,
+            CTypeId('signed', 'long', 'int'): self.obj_long,
+            CTypeId('unsigned', 'long', 'int'): self.obj_ulong,
+
+            CTypeId('long', 'double'): self.obj_ldouble,
+            CTypePtr(CTypeId('void')): self.obj_uint,
+        }
diff --git a/miasm2/arch/x86/arch.py b/miasm2/arch/x86/arch.py
index 6725f5bc..13c06ae6 100644
--- a/miasm2/arch/x86/arch.py
+++ b/miasm2/arch/x86/arch.py
@@ -134,7 +134,7 @@ RBRACK = Suppress("]")
 dbreg = Group(gpregs16.parser | gpregs32.parser | gpregs64.parser)
 gpreg = (gpregs08.parser | gpregs08_64.parser | gpregs16.parser   |
          gpregs32.parser | gpregs64.parser    | gpregs_xmm.parser |
-         gpregs_mm.parser)
+         gpregs_mm.parser | gpregs_bnd.parser)
 
 
 def reg2exprid(r):
@@ -304,7 +304,8 @@ rmarg = Group(gpregs08.parser |
               gpregs32.parser |
               gpregs64.parser |
               gpregs_mm.parser |
-              gpregs_xmm.parser
+              gpregs_xmm.parser |
+              gpregs_bnd.parser
               ).setParseAction(getreg)
 
 rmarg |= deref_mem
@@ -1712,8 +1713,10 @@ def test_addr_size(ptr, size):
 
 SIZE2XMMREG = {64:gpregs_mm,
                128:gpregs_xmm}
+SIZE2BNDREG = {64:gpregs_mm,
+               128:gpregs_bnd}
 
-def parse_mem(expr, parent, w8, sx=0, xmm=0, mm=0):
+def parse_mem(expr, parent, w8, sx=0, xmm=0, mm=0, bnd=0):
     dct_expr = {}
     opmode = parent.v_opmode()
     if expr.is_mem_segm() and expr.arg.args[0].is_int():
@@ -1780,17 +1783,23 @@ def parse_mem(expr, parent, w8, sx=0, xmm=0, mm=0):
         out = [dct_expr]
     return out, segm, True
 
-def expr2modrm(expr, parent, w8, sx=0, xmm=0, mm=0):
+def expr2modrm(expr, parent, w8, sx=0, xmm=0, mm=0, bnd=0):
     dct_expr = {f_isad : False}
 
-    if mm or xmm:
+    if mm or xmm or bnd:
         if mm and expr.size != 64:
             return None, None, False
         elif xmm and expr.size != 128:
             return None, None, False
+        elif bnd and expr.size != 128:
+            return None, None, False
 
         if isinstance(expr, ExprId):
-            selreg = SIZE2XMMREG[expr.size]
+            if bnd:
+                size2reg = SIZE2BNDREG
+            else:
+                size2reg = SIZE2XMMREG
+            selreg = size2reg[expr.size]
             if not expr in selreg.expr:
                 return None, None, False
             i = selreg.expr.index(expr)
@@ -1828,6 +1837,13 @@ def expr2modrm(expr, parent, w8, sx=0, xmm=0, mm=0):
                 return [dct_expr], None, True
             else:
                 return None, None, False
+        if bnd:
+            if expr in gpregs_bnd.expr:
+                i = gpregs_bnd.expr.index(expr)
+                dct_expr[i] = 1
+                return [dct_expr], None, True
+            else:
+                return None, None, False
         if mm:
             if expr in gpregs_mm.expr:
                 i = gpregs_mm.expr.index(expr)
@@ -1858,9 +1874,9 @@ def expr2modrm(expr, parent, w8, sx=0, xmm=0, mm=0):
                 return None, None, False
         dct_expr[i] = 1
         return [dct_expr], None, True
-    return parse_mem(expr, parent, w8, sx, xmm, mm)
+    return parse_mem(expr, parent, w8, sx, xmm, mm, bnd)
 
-def modrm2expr(modrm, parent, w8, sx=0, xmm=0, mm=0):
+def modrm2expr(modrm, parent, w8, sx=0, xmm=0, mm=0, bnd=0):
     o = []
     if not modrm[f_isad]:
         modrm_k = [x[0] for x in modrm.iteritems() if x[1] == 1]
@@ -1879,6 +1895,8 @@ def modrm2expr(modrm, parent, w8, sx=0, xmm=0, mm=0):
             expr = gpregs_xmm.expr[modrm_k]
         elif mm:
             expr = gpregs_mm.expr[modrm_k]
+        elif bnd:
+            expr = gpregs_bnd.expr[modrm_k]
         elif opmode == 8 and (parent.v_opmode() == 64 or parent.rex_p.value == 1):
             expr = gpregs08_64.expr[modrm_k]
         else:
@@ -1907,6 +1925,8 @@ def modrm2expr(modrm, parent, w8, sx=0, xmm=0, mm=0):
         opmode = 128
     elif mm:
         opmode = 64
+    elif bnd:
+        opmode = 128
 
     expr = ExprMem(expr, size=opmode)
     return expr
@@ -2326,11 +2346,12 @@ class x86_rm_mm(x86_rm_m80):
     msize = 64
     is_mm = True
     is_xmm = False
+    is_bnd = False
 
     def decode(self, v):
         p = self.parent
         xx = self.get_modrm()
-        expr = modrm2expr(xx, p, 0, 0, self.is_xmm, self.is_mm)
+        expr = modrm2expr(xx, p, 0, 0, self.is_xmm, self.is_mm, self.is_bnd)
         if isinstance(expr, ExprMem):
             if self.msize is None:
                 return False
@@ -2356,7 +2377,8 @@ class x86_rm_mm(x86_rm_m80):
             elif self.is_mm:
                 expr = ExprMem(expr.arg, 64)
 
-        v_cand, segm, ok = expr2modrm(expr, p, 0, 0, self.is_xmm, self.is_mm)
+        v_cand, segm, ok = expr2modrm(expr, p, 0, 0, self.is_xmm, self.is_mm,
+                                      self.is_bnd)
         for x in self.gen_cand(v_cand, p.v_admode()):
             yield x
 
@@ -2382,6 +2404,11 @@ class x86_rm_xmm_m64(x86_rm_mm):
     is_mm = False
     is_xmm = True
 
+class x86_rm_xmm_m128(x86_rm_mm):
+    msize = 128
+    is_mm = False
+    is_xmm = True
+
 
 class x86_rm_xmm_reg(x86_rm_mm):
     msize = None
@@ -2393,6 +2420,35 @@ class x86_rm_mm_reg(x86_rm_mm):
     is_mm = True
     is_xmm = False
 
+
+class x86_rm_bnd(x86_rm_mm):
+    msize = 128
+    is_mm = False
+    is_xmm = False
+    is_bnd = True
+
+
+class x86_rm_bnd_reg(x86_rm_mm):
+    msize = None
+    is_mm = False
+    is_xmm = False
+    is_bnd = True
+
+
+class x86_rm_bnd_m64(x86_rm_mm):
+    msize = 64
+    is_mm = False
+    is_xmm = False
+    is_bnd = True
+
+
+class x86_rm_bnd_m128(x86_rm_mm):
+    msize = 128
+    is_mm = False
+    is_xmm = False
+    is_bnd = True
+
+
 class x86_rm_reg_noarg(object):
     prio = default_prio + 1
 
@@ -2511,6 +2567,9 @@ class x86_rm_reg_mm(x86_rm_reg_noarg, m_arg):
 class x86_rm_reg_xmm(x86_rm_reg_mm):
     selreg = gpregs_xmm
 
+class x86_rm_reg_bnd(x86_rm_reg_mm):
+    selreg = gpregs_bnd
+
 class x86_rm_reg(x86_rm_reg_noarg, m_arg):
     pass
 
@@ -3196,6 +3255,7 @@ drreg = bs(l=3, cls=(x86_rm_dr, ), order =1, fname = "reg")
 
 mm_reg = bs(l=3, cls=(x86_rm_reg_mm, ), order =1, fname = "reg")
 xmm_reg = bs(l=3, cls=(x86_rm_reg_xmm, ), order =1, fname = "reg")
+bnd_reg = bs(l=3, cls=(x86_rm_reg_bnd, ), order =1, fname = "reg")
 
 
 fltreg = bs(l=3, cls=(x86_rm_flt, ), order =1, fname = "reg")
@@ -3226,8 +3286,15 @@ rm_arg_mm_reg = bs(l=0, cls=(x86_rm_mm_reg,), fname='rmarg')
 rm_arg_xmm = bs(l=0, cls=(x86_rm_xmm,), fname='rmarg')
 rm_arg_xmm_m32 = bs(l=0, cls=(x86_rm_xmm_m32,), fname='rmarg')
 rm_arg_xmm_m64 = bs(l=0, cls=(x86_rm_xmm_m64,), fname='rmarg')
+rm_arg_xmm_m128 = bs(l=0, cls=(x86_rm_xmm_m128,), fname='rmarg')
 rm_arg_xmm_reg = bs(l=0, cls=(x86_rm_xmm_reg,), fname='rmarg')
 
+rm_arg_bnd = bs(l=0, cls=(x86_rm_bnd,), fname='rmarg')
+rm_arg_bnd_m64 = bs(l=0, cls=(x86_rm_bnd_m64,), fname='rmarg')
+rm_arg_bnd_m128 = bs(l=0, cls=(x86_rm_bnd_m128,), fname='rmarg')
+rm_arg_bnd_reg = bs(l=0, cls=(x86_rm_bnd_reg,), fname='rmarg')
+
+
 swapargs = bs_swapargs(l=1, fname="swap", mn_mod=range(1 << 1))
 
 
@@ -3353,6 +3420,17 @@ addop("and", [bs("100000"), se, w8] + rmmod(d4, rm_arg_w8) + [d_imm])
 addop("and", [bs("001000"), swapargs, w8] +
       rmmod(rmreg, rm_arg_w8), [rm_arg_w8, rmreg])
 
+addop("bndmov", [bs8(0x0f), bs8(0x1a), pref_66, bs_modeno64] +
+      rmmod(bnd_reg, rm_arg_bnd_m64), [bnd_reg, rm_arg_bnd_m64])
+addop("bndmov", [bs8(0x0f), bs8(0x1a), pref_66, bs_mode64] +
+      rmmod(bnd_reg, rm_arg_bnd_m128), [bnd_reg, rm_arg_bnd_m128])
+addop("bndmov", [bs8(0x0f), bs8(0x1b), pref_66, bs_modeno64] +
+      rmmod(bnd_reg, rm_arg_bnd_m64), [rm_arg_bnd_m64, bnd_reg])
+addop("bndmov", [bs8(0x0f), bs8(0x1b), pref_66, bs_mode64] +
+      rmmod(bnd_reg, rm_arg_bnd_m128), [rm_arg_bnd_m128, bnd_reg])
+
+
+
 addop("bsf", [bs8(0x0f), bs8(0xbc)] + rmmod(rmreg))
 addop("bsr", [bs8(0x0f), bs8(0xbd), mod,
     rmreg, rm, sib_scale, sib_index, sib_base, disp, rm_arg])
@@ -3645,6 +3723,8 @@ addop("lgdt", [bs8(0x0f), bs8(0x01)] + rmmod(d2, modrm=mod_mem))
 addop("lidt", [bs8(0x0f), bs8(0x01)] + rmmod(d3, modrm=mod_mem))
 
 addop("lfence", [bs8(0x0f), bs8(0xae), bs8(0xe8)])
+addop("mfence", [bs8(0x0f), bs8(0xae), bs8(0xf0)])
+addop("sfence", [bs8(0x0f), bs8(0xae), bs8(0xf8)])
 
 addop("leave", [bs8(0xc9), stk])
 
@@ -4138,6 +4218,11 @@ addop("cvttsd2si",[bs8(0x0f), bs8(0x2c), pref_f2]
 addop("cvttss2si",[bs8(0x0f), bs8(0x2c), pref_f3]
       + rmmod(reg, rm_arg_xmm_m32))
 
+addop("palignr", [bs8(0x0f), bs8(0x73), bs8(0x0f), no_xmm_pref] +
+      rmmod(mm_reg, rm_arg_mm_m64) + [u08], [mm_reg, rm_arg_mm_m64, u08])
+addop("palignr", [bs8(0x0f), bs8(0x3a), bs8(0x0f), pref_66] +
+      rmmod(xmm_reg, rm_arg_xmm_m128) + [u08], [xmm_reg, rm_arg_xmm_m128, u08])
+
 addop("psrlq", [bs8(0x0f), bs8(0x73), no_xmm_pref] +
       rmmod(d2, rm_arg_mm) + [u08], [rm_arg_mm, u08])
 addop("psrlq", [bs8(0x0f), bs8(0x73), pref_66] +
@@ -4249,6 +4334,11 @@ addop("pcmpeqd", [bs8(0x0f), bs8(0x76), no_xmm_pref] +
 addop("pcmpeqd", [bs8(0x0f), bs8(0x76), pref_66] +
       rmmod(xmm_reg, rm_arg_xmm))
 
+addop("pcmpgtb", [bs8(0x0f), bs8(0x64), no_xmm_pref] +
+      rmmod(mm_reg, rm_arg_mm))
+addop("pcmpgtb", [bs8(0x0f), bs8(0x64), pref_66] +
+      rmmod(xmm_reg, rm_arg_xmm))
+
 addop("pcmpgtd", [bs8(0x0f), bs8(0x66), no_xmm_pref] +
       rmmod(mm_reg, rm_arg_mm))
 addop("pcmpgtd", [bs8(0x0f), bs8(0x66), pref_66] +
diff --git a/miasm2/arch/x86/regs.py b/miasm2/arch/x86/regs.py
index 5db75e37..cb7e0d7b 100644
--- a/miasm2/arch/x86/regs.py
+++ b/miasm2/arch/x86/regs.py
@@ -40,6 +40,8 @@ regs_xmm_expr = [ExprId(x, 128) for x in regs_xmm_str]
 regs_mm_str = ["MM%d" % i for i in xrange(16)]
 regs_mm_expr = [ExprId(x, 64) for x in regs_mm_str]
 
+regs_bnd_str = ["BND%d" % i for i in xrange(4)]
+regs_bnd_expr = [ExprId(x, 128) for x in regs_bnd_str]
 
 gpregs08 = reg_info(regs08_str, regs08_expr)
 gpregs08_64 = reg_info(regs08_64_str, regs08_64_expr)
@@ -49,6 +51,7 @@ gpregs64 = reg_info(regs64_str, regs64_expr)
 
 gpregs_xmm = reg_info(regs_xmm_str, regs_xmm_expr)
 gpregs_mm = reg_info(regs_mm_str, regs_mm_expr)
+gpregs_bnd = reg_info(regs_bnd_str, regs_bnd_expr)
 
 r08_eax = reg_info([regs08_str[0]], [regs08_expr[0]])
 r16_eax = reg_info([regs16_str[0]], [regs16_expr[0]])
diff --git a/miasm2/arch/x86/sem.py b/miasm2/arch/x86/sem.py
index 81da8107..56aca1c2 100644
--- a/miasm2/arch/x86/sem.py
+++ b/miasm2/arch/x86/sem.py
@@ -2626,11 +2626,34 @@ def nop(_, instr, a=None):
     return [], []
 
 
+def prefetch0(_, instr, src=None):
+    # see 4-198 on this documentation
+    # https://www-ssl.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf
+    return [], []
+
+
+def prefetch1(_, instr, src=None):
+    # see 4-198 on this documentation
+    # https://www-ssl.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf
+    return [], []
+
+
+def prefetch2(_, instr, src=None):
+    # see 4-198 on this documentation
+    # https://www-ssl.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf
+    return [], []
+
+
 def prefetchw(_, instr, src=None):
     # see 4-201 on this documentation
     # https://www-ssl.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf
     return [], []
 
+def prefetchnta(_, instr, src=None):
+    # see 4-201 on this documentation
+    # https://www-ssl.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf
+    return [], []
+
 
 def lfence(_, instr, src=None):
     # see 3-485 on this documentation
@@ -2638,6 +2661,18 @@ def lfence(_, instr, src=None):
     return [], []
 
 
+def mfence(_, instr, src=None):
+    # see 3-516 on this documentation
+    # https://www-ssl.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf
+    return [], []
+
+
+def sfence(_, instr, src=None):
+    # see 3-356 on this documentation
+    # https://www-ssl.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf
+    return [], []
+
+
 def ud2(_, instr, src=None):
     e = [m2_expr.ExprAff(exception_flags, m2_expr.ExprInt(
         EXCEPT_ILLEGAL_INSN, exception_flags.size))]
@@ -3245,23 +3280,17 @@ def xorps(_, instr, dst, src):
 
 
 def rdmsr(ir, instr):
-    msr_addr = m2_expr.ExprId('MSR') + m2_expr.ExprInt(
-        0,
-        8) * mRCX[instr.mode][:32]
+    msr_addr = m2_expr.ExprId('MSR', 64) + m2_expr.ExprInt(8, 64) * mRCX[32].zeroExtend(64)
     e = []
-    e.append(
-        m2_expr.ExprAff(mRAX[instr.mode][:32], ir.ExprMem(msr_addr, 32)))
-    e.append(m2_expr.ExprAff(mRDX[instr.mode][:32], m2_expr.ExprMem(
-        msr_addr + m2_expr.ExprInt(4, msr_addr.size), 32)))
+    e.append(m2_expr.ExprAff(mRAX[32], ir.ExprMem(msr_addr, 32)))
+    e.append(m2_expr.ExprAff(mRDX[32], ir.ExprMem(msr_addr + m2_expr.ExprInt(4, 64), 32)))
     return e, []
 
 
 def wrmsr(ir, instr):
-    msr_addr = m2_expr.ExprId('MSR') + m2_expr.ExprInt(
-        8,
-        32) * mRCX[instr.mode][:32]
+    msr_addr = m2_expr.ExprId('MSR', 64) + m2_expr.ExprInt(8, 64) * mRCX[32].zeroExtend(64)
     e = []
-    src = m2_expr.ExprCompose(mRAX[instr.mode][:32], mRDX[instr.mode][:32])
+    src = m2_expr.ExprCompose(mRAX[32], mRDX[32])
     e.append(m2_expr.ExprAff(ir.ExprMem(msr_addr, 64), src))
     return e, []
 
@@ -3647,6 +3676,21 @@ def ucomiss(_, instr, src1, src2):
 
     return e, []
 
+def ucomisd(_, instr, src1, src2):
+    e = []
+    e.append(m2_expr.ExprAff(zf, m2_expr.ExprOp(
+        'ucomisd_zf', src1[:64], src2[:64])))
+    e.append(m2_expr.ExprAff(pf, m2_expr.ExprOp(
+        'ucomisd_pf', src1[:64], src2[:64])))
+    e.append(m2_expr.ExprAff(cf, m2_expr.ExprOp(
+        'ucomisd_cf', src1[:64], src2[:64])))
+
+    e.append(m2_expr.ExprAff(of, m2_expr.ExprInt(0, 1)))
+    e.append(m2_expr.ExprAff(af, m2_expr.ExprInt(0, 1)))
+    e.append(m2_expr.ExprAff(nf, m2_expr.ExprInt(0, 1)))
+
+    return e, []
+
 
 def pshufb(_, instr, dst, src):
     e = []
@@ -3749,6 +3793,16 @@ def pslldq(_, instr, dst, src):
         return [m2_expr.ExprAff(dst, dst << m2_expr.ExprInt(8 * count, dst.size))], []
 
 
+def psrldq(_, instr, dst, src):
+    assert src.is_int()
+    e = []
+    count = int(src)
+    if count > 15:
+        return [m2_expr.ExprAff(dst, m2_expr.ExprInt(0, dst.size))], []
+    else:
+        return [m2_expr.ExprAff(dst, dst >> m2_expr.ExprInt(8 * count, dst.size))], []
+
+
 def iret(ir, instr):
     """IRET implementation
     XXX: only support "no-privilege change"
@@ -4090,6 +4144,29 @@ def smsw(ir, instr, dst):
     return e, []
 
 
+def bndmov(ir, instr, dst, src):
+    # Implemented as a NOP, because BND side effects are not yet supported
+    return [], []
+
+def palignr(ir, instr, dst, src, imm):
+    # dst.src >> imm * 8 [:dst.size]
+
+    shift = int(imm) * 8
+    if shift == 0:
+        result = src
+    elif shift == src.size:
+        result = dst
+    elif shift > src.size:
+        result = dst >> m2_expr.ExprInt(shift - src.size, dst.size)
+    else:
+        # shift < src.size
+        result = m2_expr.ExprCompose(
+            src[shift:],
+            dst[:shift],
+        )
+
+    return [m2_expr.ExprAff(dst, result)], []
+
 
 mnemo_func = {'mov': mov,
               'xchg': xchg,
@@ -4263,8 +4340,14 @@ mnemo_func = {'mov': mov,
               'fcomip': fcomip,
               'nop': nop,
               'ud2': ud2,
+              'prefetch0': prefetch0,
+              'prefetch1': prefetch1,
+              'prefetch2': prefetch2,
               'prefetchw': prefetchw,
+              'prefetchnta': prefetchnta,
               'lfence': lfence,
+              'mfence': mfence,
+              'sfence': sfence,
               'fnop': nop,  # XXX
               'hlt': hlt,
               'rdtsc': rdtsc,
@@ -4439,9 +4522,7 @@ mnemo_func = {'mov': mov,
               "cvttss2si": cvttss2si,
 
 
-
-
-
+              "bndmov": bndmov,
 
 
 
@@ -4449,6 +4530,7 @@ mnemo_func = {'mov': mov,
               "movss": movss,
 
               "ucomiss": ucomiss,
+              "ucomisd": ucomisd,
 
               #
               # MMX/AVX/SSE operations
@@ -4520,6 +4602,9 @@ mnemo_func = {'mov': mov,
               "pslld": pslld,
               "psllq": psllq,
               "pslldq": pslldq,
+              "psrldq": psrldq,
+
+              "palignr": palignr,
 
               "pmaxub": pmaxub,
               "pmaxuw": pmaxuw,