X86: Implement the versions of PUNPCKLBW that use XMM registers.

src/arch/x86/isa/decoder/two_byte_opcodes.isa

@@ -562,7 +562,7 @@
                 }
                 // operand size (0x66)
                 0x1: decode OPCODE_OP_BOTTOM3 {
-                    0x0: punpcklbw_Vo_Wq();
+                    0x0: Inst::PUNPCKLBW(Vo,Wq);
                     0x1: punpcklwd_Vo_Wq();
                     0x2: punpckldq_Vo_Wq();
                     0x3: packsswb_Vo_Wo();

src/arch/x86/isa/insts/simd128/integer/data_reordering/unpack_and_interleave.py

@@ -54,12 +54,28 @@
 # Authors: Gabe Black
 
 microcode = '''
+def macroop PUNPCKLBW_XMM_XMM {
+    unpack xmmh, xmml, xmmlm, sel=1, size=1
+    unpack xmml, xmml, xmmlm, sel=0, size=1
+};
+
+def macroop PUNPCKLBW_XMM_M {
+    ldfp ufp1, seg, sib, disp, dataSize=8
+    unpack xmmh, xmml, ufp1, sel=1, size=1
+    unpack xmml, xmml, ufp1, sel=0, size=1
+};
+
+def macroop PUNPCKLBW_XMM_P {
+    rdip t7
+    ldfp ufp1, seg, riprel, disp, dataSize=8
+    unpack xmmh, xmml, ufp1, sel=1, size=1
+    unpack xmml, xmml, ufp1, sel=0, size=1
+};
+'''
 # PUNPCKHBW
 # PUNPCKHWD
 # PUNPCKHDQ
 # PUNPCKHQDQ
-# PUNPCKLBW
 # PUNPCKLWD
 # PUNPCKLDQ
 # PUNPCKLQDQ
-'''