diff --git a/lib/Target/ARM/ARMAddressingModes.h b/lib/Target/ARM/ARMAddressingModes.h
index abdd938..0b3a7d4 100644
--- a/lib/Target/ARM/ARMAddressingModes.h
+++ b/lib/Target/ARM/ARMAddressingModes.h
@@ -35,6 +35,10 @@
     add = '+', sub = '-'
   };
 
+  static inline const char *getAddrOpcStr(AddrOpc Op) {
+    return Op == sub ? "-" : "";
+  }
+
   static inline const char *getShiftOpcStr(ShiftOpc Op) {
     switch (Op) {
     default: assert(0 && "Unknown shift opc!");
@@ -127,6 +131,20 @@
     return (Imm >> 8) * 2;
   }
 
+  /// getSOImmValOneRotate - Try to handle Imm with an immediate shifter
+  /// operand, computing the rotate amount to use.  If this immediate value
+  /// cannot be handled with a single shifter-op, return 0.
+  static inline unsigned getSOImmValOneRotate(unsigned Imm) {
+    // A5.2.4 Constants with multiple encodings
+    // The lowest unsigned value of rotation wins!
+    for (unsigned R = 1; R <= 15; ++R)
+      if ((Imm & rotr32(~255U, 2*R)) == 0)
+        return 2*R;
+
+    // Failed to find a suitable rotate amount.
+    return 0;
+  }
+
   /// getSOImmValRotate - Try to handle Imm with an immediate shifter operand,
   /// computing the rotate amount to use.  If this immediate value cannot be
   /// handled with a single shifter-op, determine a good rotate amount that will
@@ -179,7 +197,7 @@
     // of zero.
     if ((Arg & ~255U) == 0) return Arg;
 
-    unsigned RotAmt = getSOImmValRotate(Arg);
+    unsigned RotAmt = getSOImmValOneRotate(Arg);
 
     // If this cannot be handled with a single shifter_op, bail out.
     if (rotr32(~255U, RotAmt) & Arg)
diff --git a/lib/Target/ARM/ARMInstrFormats.td b/lib/Target/ARM/ARMInstrFormats.td
index 258a96b..52553f5 100644
--- a/lib/Target/ARM/ARMInstrFormats.td
+++ b/lib/Target/ARM/ARMInstrFormats.td
@@ -1464,6 +1464,29 @@
 // ARM NEON Instruction templates.
 //
 
+// NSFormat specifies further details of a NEON instruction.  This is used by
+// the disassembler to classify NEONFrm instructions for disassembly purpose.
+class NSFormat<bits<5> val> {
+  bits<5> Value = val;
+}
+def NSFormatNone               : NSFormat<0>;
+def VLDSTLaneFrm               : NSFormat<1>;
+def VLDSTLaneDblFrm            : NSFormat<2>;
+def VLDSTRQFrm                 : NSFormat<3>;
+def NVdImmFrm                  : NSFormat<4>;
+def NVdVmImmFrm                : NSFormat<5>;
+def NVdVmImmVCVTFrm            : NSFormat<6>;
+def NVdVmImmVDupLaneFrm        : NSFormat<7>;
+def NVdVmImmVSHLLFrm           : NSFormat<8>;
+def NVectorShuffleFrm          : NSFormat<9>;
+def NVectorShiftFrm            : NSFormat<10>;
+def NVectorShift2Frm           : NSFormat<11>;
+def NVdVnVmImmFrm              : NSFormat<12>;
+def NVdVnVmImmVectorShiftFrm   : NSFormat<13>;
+def NVdVnVmImmVectorExtractFrm : NSFormat<14>;
+def NVdVnVmImmMulScalarFrm     : NSFormat<15>;
+def VTBLFrm                    : NSFormat<16>;
+
 class NeonI<dag oops, dag iops, AddrMode am, IndexMode im, InstrItinClass itin,
             string opc, string dt, string asm, string cstr, list<dag> pattern>
   : InstARM<am, Size4Bytes, im, NEONFrm, NeonDomain, cstr, itin> {
@@ -1474,6 +1497,8 @@
                      !strconcat("\t", asm));
   let Pattern = pattern;
   list<Predicate> Predicates = [HasNEON];
+  NSFormat NSF = NSFormatNone;         // For disassembly.
+  bits<5> NSForm = NSFormatNone.Value; // For disassembly.
 }
 
 // Same as NeonI except it does not have a "data type" specifier.
@@ -1485,6 +1510,8 @@
   let AsmString = !strconcat(!strconcat(opc, "${p}"), !strconcat("\t", asm));
   let Pattern = pattern;
   list<Predicate> Predicates = [HasNEON];
+  NSFormat NSF = NSFormatNone;         // For disassembly.
+  bits<5> NSForm = NSFormatNone.Value; // For disassembly.
 }
 
 class NI<dag oops, dag iops, InstrItinClass itin, string opc, string asm,
@@ -1497,6 +1524,8 @@
           string asm, list<dag> pattern>
   : NeonXI<oops, iops, AddrMode4, IndexModeNone, itin, opc, asm, "",
           pattern> {
+  let NSF    = VLDSTRQFrm;       // For disassembly.
+  let NSForm = VLDSTRQFrm.Value; // For disassembly.
 }
 
 class NLdSt<bit op23, bits<2> op21_20, bits<4> op11_8, bits<4> op7_4,
@@ -1509,6 +1538,8 @@
   let Inst{21-20} = op21_20;
   let Inst{11-8} = op11_8;
   let Inst{7-4} = op7_4;
+  let NSF    = VLDSTLaneFrm;       // For disassembly.
+  let NSForm = VLDSTLaneFrm.Value; // For disassembly.
 }
 
 class NDataI<dag oops, dag iops, InstrItinClass itin,
@@ -1538,6 +1569,8 @@
   let Inst{6} = op6;
   let Inst{5} = op5;
   let Inst{4} = op4;
+  let NSF    = NVdImmFrm;       // For disassembly.
+  let NSForm = NVdImmFrm.Value; // For disassembly.
 }
 
 // NEON 2 vector register format.
@@ -1553,6 +1586,8 @@
   let Inst{11-7} = op11_7;
   let Inst{6} = op6;
   let Inst{4} = op4;
+  let NSF    = NVdVmImmFrm;       // For disassembly.
+  let NSForm = NVdVmImmFrm.Value; // For disassembly.
 }
 
 // Same as N2V except it doesn't have a datatype suffix.
@@ -1568,6 +1603,8 @@
   let Inst{11-7} = op11_7;
   let Inst{6} = op6;
   let Inst{4} = op4;
+  let NSF    = NVdVmImmFrm;       // For disassembly.
+  let NSForm = NVdVmImmFrm.Value; // For disassembly.
 }
 
 // NEON 2 vector register with immediate.
@@ -1581,6 +1618,8 @@
   let Inst{7} = op7;
   let Inst{6} = op6;
   let Inst{4} = op4;
+  let NSF    = NVdVmImmFrm;       // For disassembly.
+  let NSForm = NVdVmImmFrm.Value; // For disassembly.
 }
 
 // NEON 3 vector register format.
@@ -1594,6 +1633,8 @@
   let Inst{11-8} = op11_8;
   let Inst{6} = op6;
   let Inst{4} = op4;
+  let NSF    = NVdVnVmImmFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmFrm.Value; // For disassembly.
 }
 
 // Same as N3VX except it doesn't have a data type suffix.
@@ -1607,6 +1648,8 @@
   let Inst{11-8} = op11_8;
   let Inst{6} = op6;
   let Inst{4} = op4;
+  let NSF    = NVdVnVmImmFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmFrm.Value; // For disassembly.
 }
 
 // NEON VMOVs between scalar and core registers.
diff --git a/lib/Target/ARM/ARMInstrNEON.td b/lib/Target/ARM/ARMInstrNEON.td
index 8fee6fa..da2babf 100644
--- a/lib/Target/ARM/ARMInstrNEON.td
+++ b/lib/Target/ARM/ARMInstrNEON.td
@@ -213,7 +213,10 @@
 class VLD2Ddbl<bits<4> op7_4, string OpcodeStr, string Dt>
   : NLdSt<0,0b10,0b1001,op7_4, (outs DPR:$dst1, DPR:$dst2),
           (ins addrmode6:$addr), IIC_VLD2,
-          OpcodeStr, Dt, "\\{$dst1, $dst2\\}, $addr", "", []>;
+          OpcodeStr, Dt, "\\{$dst1, $dst2\\}, $addr", "", []> {
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 def  VLD2d8D  : VLD2Ddbl<0b0000, "vld2", "8">;
 def  VLD2d16D : VLD2Ddbl<0b0100, "vld2", "16">;
@@ -228,7 +231,10 @@
   : NLdSt<0,0b10,0b0101,op7_4, (outs DPR:$dst1, DPR:$dst2, DPR:$dst3, GPR:$wb),
           (ins addrmode6:$addr), IIC_VLD3,
           OpcodeStr, Dt, "\\{$dst1, $dst2, $dst3\\}, $addr",
-          "$addr.addr = $wb", []>;
+          "$addr.addr = $wb", []> {
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 def  VLD3d8   : VLD3D<0b0000, "vld3", "8">;
 def  VLD3d16  : VLD3D<0b0100, "vld3", "16">;
@@ -260,7 +266,10 @@
           (outs DPR:$dst1, DPR:$dst2, DPR:$dst3, DPR:$dst4, GPR:$wb),
           (ins addrmode6:$addr), IIC_VLD4,
           OpcodeStr, Dt, "\\{$dst1, $dst2, $dst3, $dst4\\}, $addr",
-          "$addr.addr = $wb", []>;
+          "$addr.addr = $wb", []> {
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 def  VLD4d8   : VLD4D<0b0000, "vld4", "8">;
 def  VLD4d16  : VLD4D<0b0100, "vld4", "16">;
@@ -297,12 +306,28 @@
 def VLD2LNd32 : VLD2LN<0b1001, "vld2", "32"> { let Inst{6} = 0; }
 
 // vld2 to double-spaced even registers.
-def VLD2LNq16a: VLD2LN<0b0101, "vld2", "16"> { let Inst{5} = 1; }
-def VLD2LNq32a: VLD2LN<0b1001, "vld2", "32"> { let Inst{6} = 1; }
+def VLD2LNq16a: VLD2LN<0b0101, "vld2", "16"> {
+  let Inst{5} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VLD2LNq32a: VLD2LN<0b1001, "vld2", "32"> {
+  let Inst{6} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 // vld2 to double-spaced odd registers.
-def VLD2LNq16b: VLD2LN<0b0101, "vld2", "16"> { let Inst{5} = 1; }
-def VLD2LNq32b: VLD2LN<0b1001, "vld2", "32"> { let Inst{6} = 1; }
+def VLD2LNq16b: VLD2LN<0b0101, "vld2", "16"> {
+  let Inst{5} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VLD2LNq32b: VLD2LN<0b1001, "vld2", "32"> {
+  let Inst{6} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 //   VLD3LN   : Vector Load (single 3-element structure to one lane)
 class VLD3LN<bits<4> op11_8, string OpcodeStr, string Dt>
@@ -318,7 +343,11 @@
 def VLD3LNd32 : VLD3LN<0b1010, "vld3", "32"> { let Inst{6-4} = 0b000; }
 
 // vld3 to double-spaced even registers.
-def VLD3LNq16a: VLD3LN<0b0110, "vld3", "16"> { let Inst{5-4} = 0b10; }
+def VLD3LNq16a: VLD3LN<0b0110, "vld3", "16"> {
+  let Inst{5-4} = 0b10;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 def VLD3LNq32a: VLD3LN<0b1010, "vld3", "32"> { let Inst{6-4} = 0b100; }
 
 // vld3 to double-spaced odd registers.
@@ -340,12 +369,28 @@
 def VLD4LNd32 : VLD4LN<0b1011, "vld4", "32"> { let Inst{6} = 0; }
 
 // vld4 to double-spaced even registers.
-def VLD4LNq16a: VLD4LN<0b0111, "vld4", "16"> { let Inst{5} = 1; }
-def VLD4LNq32a: VLD4LN<0b1011, "vld4", "32"> { let Inst{6} = 1; }
+def VLD4LNq16a: VLD4LN<0b0111, "vld4", "16"> {
+  let Inst{5} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VLD4LNq32a: VLD4LN<0b1011, "vld4", "32"> {
+  let Inst{6} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 // vld4 to double-spaced odd registers.
-def VLD4LNq16b: VLD4LN<0b0111, "vld4", "16"> { let Inst{5} = 1; }
-def VLD4LNq32b: VLD4LN<0b1011, "vld4", "32"> { let Inst{6} = 1; }
+def VLD4LNq16b: VLD4LN<0b0111, "vld4", "16"> {
+  let Inst{5} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VLD4LNq32b: VLD4LN<0b1011, "vld4", "32"> {
+  let Inst{6} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 //   VLD1DUP  : Vector Load (single element to all lanes)
 //   VLD2DUP  : Vector Load (single 2-element structure to all lanes)
@@ -433,7 +478,10 @@
 class VST2Ddbl<bits<4> op7_4, string OpcodeStr, string Dt>
   : NLdSt<0, 0b00, 0b1001, op7_4, (outs),
           (ins addrmode6:$addr, DPR:$src1, DPR:$src2), IIC_VST,
-          OpcodeStr, Dt, "\\{$src1, $src2\\}, $addr", "", []>;
+          OpcodeStr, Dt, "\\{$src1, $src2\\}, $addr", "", []> {
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 def  VST2d8D  : VST2Ddbl<0b0000, "vst2", "8">;
 def  VST2d16D : VST2Ddbl<0b0100, "vst2", "16">;
@@ -448,7 +496,10 @@
   : NLdSt<0,0b00,0b0101,op7_4, (outs GPR:$wb),
           (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3), IIC_VST,
           OpcodeStr, Dt, "\\{$src1, $src2, $src3\\}, $addr",
-          "$addr.addr = $wb", []>;
+          "$addr.addr = $wb", []> {
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 def  VST3d8   : VST3D<0b0000, "vst3", "8">;
 def  VST3d16  : VST3D<0b0100, "vst3", "16">;
@@ -478,7 +529,10 @@
   : NLdSt<0,0b00,0b0001,op7_4, (outs GPR:$wb),
           (ins addrmode6:$addr, DPR:$src1, DPR:$src2, DPR:$src3, DPR:$src4),
           IIC_VST, OpcodeStr, Dt, "\\{$src1, $src2, $src3, $src4\\}, $addr",
-          "$addr.addr = $wb", []>;
+          "$addr.addr = $wb", []> {
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 def  VST4d8   : VST4D<0b0000, "vst4", "8">;
 def  VST4d16  : VST4D<0b0100, "vst4", "16">;
@@ -515,12 +569,28 @@
 def VST2LNd32 : VST2LN<0b1001, "vst2", "32"> { let Inst{6} = 0; }
 
 // vst2 to double-spaced even registers.
-def VST2LNq16a: VST2LN<0b0101, "vst2", "16"> { let Inst{5} = 1; }
-def VST2LNq32a: VST2LN<0b1001, "vst2", "32"> { let Inst{6} = 1; }
+def VST2LNq16a: VST2LN<0b0101, "vst2", "16"> {
+  let Inst{5} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VST2LNq32a: VST2LN<0b1001, "vst2", "32"> {
+  let Inst{6} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 // vst2 to double-spaced odd registers.
-def VST2LNq16b: VST2LN<0b0101, "vst2", "16"> { let Inst{5} = 1; }
-def VST2LNq32b: VST2LN<0b1001, "vst2", "32"> { let Inst{6} = 1; }
+def VST2LNq16b: VST2LN<0b0101, "vst2", "16"> {
+  let Inst{5} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VST2LNq32b: VST2LN<0b1001, "vst2", "32"> {
+  let Inst{6} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 //   VST3LN   : Vector Store (single 3-element structure from one lane)
 class VST3LN<bits<4> op11_8, string OpcodeStr, string Dt>
@@ -535,12 +605,28 @@
 def VST3LNd32 : VST3LN<0b1010, "vst3", "32"> { let Inst{6-4} = 0b000; }
 
 // vst3 to double-spaced even registers.
-def VST3LNq16a: VST3LN<0b0110, "vst3", "16"> { let Inst{5-4} = 0b10; }
-def VST3LNq32a: VST3LN<0b1010, "vst3", "32"> { let Inst{6-4} = 0b100; }
+def VST3LNq16a: VST3LN<0b0110, "vst3", "16"> {
+  let Inst{5-4} = 0b10;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VST3LNq32a: VST3LN<0b1010, "vst3", "32"> {
+  let Inst{6-4} = 0b100;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 // vst3 to double-spaced odd registers.
-def VST3LNq16b: VST3LN<0b0110, "vst3", "16"> { let Inst{5-4} = 0b10; }
-def VST3LNq32b: VST3LN<0b1010, "vst3", "32"> { let Inst{6-4} = 0b100; }
+def VST3LNq16b: VST3LN<0b0110, "vst3", "16"> {
+  let Inst{5-4} = 0b10;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VST3LNq32b: VST3LN<0b1010, "vst3", "32"> {
+  let Inst{6-4} = 0b100;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 //   VST4LN   : Vector Store (single 4-element structure from one lane)
 class VST4LN<bits<4> op11_8, string OpcodeStr, string Dt>
@@ -556,12 +642,28 @@
 def VST4LNd32 : VST4LN<0b1011, "vst4", "32"> { let Inst{6} = 0; }
 
 // vst4 to double-spaced even registers.
-def VST4LNq16a: VST4LN<0b0111, "vst4", "16"> { let Inst{5} = 1; }
-def VST4LNq32a: VST4LN<0b1011, "vst4", "32"> { let Inst{6} = 1; }
+def VST4LNq16a: VST4LN<0b0111, "vst4", "16"> {
+  let Inst{5} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VST4LNq32a: VST4LN<0b1011, "vst4", "32"> {
+  let Inst{6} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 // vst4 to double-spaced odd registers.
-def VST4LNq16b: VST4LN<0b0111, "vst4", "16"> { let Inst{5} = 1; }
-def VST4LNq32b: VST4LN<0b1011, "vst4", "32"> { let Inst{6} = 1; }
+def VST4LNq16b: VST4LN<0b0111, "vst4", "16"> {
+  let Inst{5} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
+def VST4LNq32b: VST4LN<0b1011, "vst4", "32"> {
+  let Inst{6} = 1;
+  let NSF    = VLDSTLaneDblFrm;       // For disassembly.
+  let NSForm = VLDSTLaneDblFrm.Value; // For disassembly.
+}
 
 } // mayStore = 1, hasExtraSrcRegAllocReq = 1
 
@@ -668,12 +770,18 @@
   : N2V<0b11, 0b11, op19_18, 0b10, op11_7, 0, 0, (outs DPR:$dst1, DPR:$dst2),
         (ins DPR:$src1, DPR:$src2), IIC_VPERMD, 
         OpcodeStr, Dt, "$dst1, $dst2",
-        "$src1 = $dst1, $src2 = $dst2", []>;
+        "$src1 = $dst1, $src2 = $dst2", []> {
+  let NSF    = NVectorShuffleFrm;       // For disassembly.
+  let NSForm = NVectorShuffleFrm.Value; // For disassembly.
+}
 class N2VQShuffle<bits<2> op19_18, bits<5> op11_7,
                   InstrItinClass itin, string OpcodeStr, string Dt>
   : N2V<0b11, 0b11, op19_18, 0b10, op11_7, 1, 0, (outs QPR:$dst1, QPR:$dst2),
         (ins QPR:$src1, QPR:$src2), itin, OpcodeStr, Dt, "$dst1, $dst2",
-        "$src1 = $dst1, $src2 = $dst2", []>;
+        "$src1 = $dst1, $src2 = $dst2", []> {
+  let NSF    = NVectorShuffleFrm;       // For disassembly.
+  let NSForm = NVectorShuffleFrm.Value; // For disassembly.
+}
 
 // Basic 3-register operations: single-, double- and quad-register.
 class N3VS<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
@@ -715,6 +823,8 @@
               (Ty (ShOp (Ty DPR:$src1),
                         (Ty (NEONvduplane (Ty DPR_VFP2:$src2), imm:$lane)))))]>{
   let isCommutable = 0;
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
 }
 class N3VDSL16<bits<2> op21_20, bits<4> op11_8, 
                string OpcodeStr, string Dt, ValueType Ty, SDNode ShOp>
@@ -725,6 +835,8 @@
               (Ty (ShOp (Ty DPR:$src1),
                         (Ty (NEONvduplane (Ty DPR_8:$src2), imm:$lane)))))]> {
   let isCommutable = 0;
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
 }
 
 class N3VQ<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
@@ -756,6 +868,8 @@
                            (ResTy (NEONvduplane (OpTy DPR_VFP2:$src2),
                                                 imm:$lane)))))]> {
   let isCommutable = 0;
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
 }
 class N3VQSL16<bits<2> op21_20, bits<4> op11_8, string OpcodeStr, string Dt,
                ValueType ResTy, ValueType OpTy, SDNode ShOp>
@@ -767,6 +881,8 @@
                            (ResTy (NEONvduplane (OpTy DPR_8:$src2),
                                                 imm:$lane)))))]> {
   let isCommutable = 0;
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
 }
 
 // Basic 3-register intrinsics, both double- and quad-register.
@@ -789,6 +905,8 @@
                          (Ty (NEONvduplane (Ty DPR_VFP2:$src2),
                                            imm:$lane)))))]> {
   let isCommutable = 0;
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
 }
 class N3VDIntSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                   string OpcodeStr, string Dt, ValueType Ty, Intrinsic IntOp>
@@ -800,6 +918,8 @@
                          (Ty (NEONvduplane (Ty DPR_8:$src2),
                                            imm:$lane)))))]> {
   let isCommutable = 0;
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
 }
 
 class N3VQInt<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
@@ -822,6 +942,8 @@
                             (ResTy (NEONvduplane (OpTy DPR_VFP2:$src2),
                                                  imm:$lane)))))]> {
   let isCommutable = 0;
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
 }
 class N3VQIntSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                   string OpcodeStr, string Dt,
@@ -834,6 +956,8 @@
                             (ResTy (NEONvduplane (OpTy DPR_8:$src2),
                                                  imm:$lane)))))]> {
   let isCommutable = 0;
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
 }
 
 // Multiply-Add/Sub operations: single-, double- and quad-register.
@@ -864,7 +988,10 @@
               (Ty (ShOp (Ty DPR:$src1),
                         (Ty (MulOp DPR:$src2,
                                    (Ty (NEONvduplane (Ty DPR_VFP2:$src3),
-                                                     imm:$lane)))))))]>;
+                                                     imm:$lane)))))))]> {
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
+}
 class N3VDMulOpSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                     string OpcodeStr, string Dt,
                     ValueType Ty, SDNode MulOp, SDNode ShOp>
@@ -876,7 +1003,10 @@
               (Ty (ShOp (Ty DPR:$src1),
                         (Ty (MulOp DPR:$src2,
                                    (Ty (NEONvduplane (Ty DPR_8:$src3),
-                                                     imm:$lane)))))))]>;
+                                                     imm:$lane)))))))]> {
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
+}
 
 class N3VQMulOp<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
                 InstrItinClass itin, string OpcodeStr, string Dt, ValueType Ty,
@@ -897,7 +1027,10 @@
               (ResTy (ShOp (ResTy QPR:$src1),
                            (ResTy (MulOp QPR:$src2,
                                    (ResTy (NEONvduplane (OpTy DPR_VFP2:$src3),
-                                                        imm:$lane)))))))]>;
+                                                        imm:$lane)))))))]> {
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
+}
 class N3VQMulOpSL16<bits<2> op21_20, bits<4> op11_8, InstrItinClass itin,
                     string OpcodeStr, string Dt,
                     ValueType ResTy, ValueType OpTy,
@@ -910,7 +1043,10 @@
               (ResTy (ShOp (ResTy QPR:$src1),
                            (ResTy (MulOp QPR:$src2,
                                    (ResTy (NEONvduplane (OpTy DPR_8:$src3),
-                                                        imm:$lane)))))))]>;
+                                                        imm:$lane)))))))]> {
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
+}
 
 // Neon 3-argument intrinsics, both double- and quad-register.
 // The destination register is also used as the first source operand register.
@@ -996,7 +1132,10 @@
         [(set (ResTy QPR:$dst),
               (ResTy (IntOp (OpTy DPR:$src1),
                             (OpTy (NEONvduplane (OpTy DPR_VFP2:$src2),
-                                                imm:$lane)))))]>;
+                                                imm:$lane)))))]> {
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
+}
 class N3VLIntSL16<bit op24, bits<2> op21_20, bits<4> op11_8,
                   InstrItinClass itin, string OpcodeStr, string Dt,
                   ValueType ResTy, ValueType OpTy, Intrinsic IntOp>
@@ -1006,7 +1145,10 @@
         [(set (ResTy QPR:$dst),
               (ResTy (IntOp (OpTy DPR:$src1),
                             (OpTy (NEONvduplane (OpTy DPR_8:$src2),
-                                                imm:$lane)))))]>;
+                                                imm:$lane)))))]> {
+  let NSF    = NVdVnVmImmMulScalarFrm;       // For disassembly.
+  let NSForm = NVdVnVmImmMulScalarFrm.Value; // For disassembly.
+}
 
 // Wide 3-register intrinsics.
 class N3VWInt<bit op24, bit op23, bits<2> op21_20, bits<4> op11_8, bit op4,
@@ -1055,6 +1197,10 @@
         OpcodeStr, Dt, "$dst, $src2", "$src1 = $dst",
         [(set QPR:$dst, (ResTy (IntOp (ResTy QPR:$src1), (OpTy QPR:$src2))))]>;
 
+// This is a big let * in block to mark these instructions NVectorShiftFrm to
+// help the disassembler.
+let NSF = NVectorShiftFrm, NSForm = NVectorShiftFrm.Value in {
+
 // Shift by immediate,
 // both double- and quad-register.
 class N2VDSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4,
@@ -1072,16 +1218,6 @@
            OpcodeStr, Dt, "$dst, $src, $SIMM", "",
            [(set QPR:$dst, (Ty (OpNode (Ty QPR:$src), (i32 imm:$SIMM))))]>;
 
-// Long shift by immediate.
-class N2VLSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6, bit op4,
-             string OpcodeStr, string Dt,
-             ValueType ResTy, ValueType OpTy, SDNode OpNode>
-  : N2VImm<op24, op23, op11_8, op7, op6, op4,
-           (outs QPR:$dst), (ins DPR:$src, i32imm:$SIMM), IIC_VSHLiD,
-           OpcodeStr, Dt, "$dst, $src, $SIMM", "",
-           [(set QPR:$dst, (ResTy (OpNode (OpTy DPR:$src),
-                                          (i32 imm:$SIMM))))]>;
-
 // Narrow shift by immediate.
 class N2VNSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6, bit op4,
              InstrItinClass itin, string OpcodeStr, string Dt,
@@ -1124,8 +1260,26 @@
            OpcodeStr, Dt, "$dst, $src2, $SIMM", "$src1 = $dst",
            [(set QPR:$dst, (Ty (ShOp QPR:$src1, QPR:$src2, (i32 imm:$SIMM))))]>;
 
+} // End of "let NSF = NVectorShiftFrm, ..."
+
+// Long shift by immediate.
+class N2VLSh<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6, bit op4,
+             string OpcodeStr, string Dt,
+             ValueType ResTy, ValueType OpTy, SDNode OpNode>
+  : N2VImm<op24, op23, op11_8, op7, op6, op4,
+           (outs QPR:$dst), (ins DPR:$src, i32imm:$SIMM), IIC_VSHLiD,
+           OpcodeStr, Dt, "$dst, $src, $SIMM", "",
+           [(set QPR:$dst, (ResTy (OpNode (OpTy DPR:$src),
+                                          (i32 imm:$SIMM))))]> {
+  // This has a different interpretation of the shift amount encoding than
+  // NVectorShiftFrm.
+  let NSF    = NVectorShift2Frm;       // For disassembly.
+  let NSForm = NVectorShift2Frm.Value; // For disassembly.
+}
+
 // Convert, with fractional bits immediate,
 // both double- and quad-register.
+let NSF = NVdVmImmVCVTFrm, NSForm = NVdVmImmVCVTFrm.Value in {
 class N2VCvtD<bit op24, bit op23, bits<4> op11_8, bit op7, bit op4,
               string OpcodeStr, string Dt, ValueType ResTy, ValueType OpTy,
               Intrinsic IntOp>
@@ -1140,6 +1294,7 @@
            (outs QPR:$dst), (ins QPR:$src, i32imm:$SIMM), IIC_VUNAQ, 
            OpcodeStr, Dt, "$dst, $src, $SIMM", "",
            [(set QPR:$dst, (ResTy (IntOp (OpTy QPR:$src), (i32 imm:$SIMM))))]>;
+}
 
 //===----------------------------------------------------------------------===//
 // Multiclasses
@@ -1350,6 +1505,60 @@
                       v2i64, v2i64, IntOp, Commutable>;
 }
 
+// Same as N3VInt_QHSD, except they're for Vector Shift (Register) Instructions.
+// D:Vd M:Vm N:Vn (notice that M:Vm is the first operand)
+// This helps the disassembler.
+let NSF = NVdVnVmImmVectorShiftFrm, NSForm = NVdVnVmImmVectorShiftFrm.Value in {
+multiclass N3VInt_HS2<bit op24, bit op23, bits<4> op11_8, bit op4,
+                     InstrItinClass itinD16, InstrItinClass itinD32,
+                     InstrItinClass itinQ16, InstrItinClass itinQ32,
+                     string OpcodeStr, string Dt,
+                     Intrinsic IntOp, bit Commutable = 0> {
+  // 64-bit vector types.
+  def v4i16 : N3VDInt<op24, op23, 0b01, op11_8, op4, itinD16,
+                      OpcodeStr, !strconcat(Dt, "16"),
+                      v4i16, v4i16, IntOp, Commutable>;
+  def v2i32 : N3VDInt<op24, op23, 0b10, op11_8, op4, itinD32,
+                      OpcodeStr, !strconcat(Dt, "32"),
+                      v2i32, v2i32, IntOp, Commutable>;
+
+  // 128-bit vector types.
+  def v8i16 : N3VQInt<op24, op23, 0b01, op11_8, op4, itinQ16,
+                      OpcodeStr, !strconcat(Dt, "16"),
+                      v8i16, v8i16, IntOp, Commutable>;
+  def v4i32 : N3VQInt<op24, op23, 0b10, op11_8, op4, itinQ32,
+                      OpcodeStr, !strconcat(Dt, "32"),
+                      v4i32, v4i32, IntOp, Commutable>;
+}
+multiclass N3VInt_QHS2<bit op24, bit op23, bits<4> op11_8, bit op4,
+                      InstrItinClass itinD16, InstrItinClass itinD32,
+                      InstrItinClass itinQ16, InstrItinClass itinQ32,
+                      string OpcodeStr, string Dt,
+                      Intrinsic IntOp, bit Commutable = 0>
+  : N3VInt_HS2<op24, op23, op11_8, op4, itinD16, itinD32, itinQ16, itinQ32,
+              OpcodeStr, Dt, IntOp, Commutable> {
+  def v8i8  : N3VDInt<op24, op23, 0b00, op11_8, op4, itinD16,
+                      OpcodeStr, !strconcat(Dt, "8"),
+                      v8i8, v8i8, IntOp, Commutable>;
+  def v16i8 : N3VQInt<op24, op23, 0b00, op11_8, op4, itinQ16,
+                      OpcodeStr, !strconcat(Dt, "8"),
+                      v16i8, v16i8, IntOp, Commutable>;
+}
+multiclass N3VInt_QHSD2<bit op24, bit op23, bits<4> op11_8, bit op4,
+                       InstrItinClass itinD16, InstrItinClass itinD32,
+                       InstrItinClass itinQ16, InstrItinClass itinQ32,
+                       string OpcodeStr, string Dt,
+                       Intrinsic IntOp, bit Commutable = 0>
+  : N3VInt_QHS2<op24, op23, op11_8, op4, itinD16, itinD32, itinQ16, itinQ32,
+               OpcodeStr, Dt, IntOp, Commutable> {
+  def v1i64 : N3VDInt<op24, op23, 0b11, op11_8, op4, itinD32,
+                      OpcodeStr, !strconcat(Dt, "64"),
+                      v1i64, v1i64, IntOp, Commutable>;
+  def v2i64 : N3VQInt<op24, op23, 0b11, op11_8, op4, itinQ32,
+                      OpcodeStr, !strconcat(Dt, "64"),
+                      v2i64, v2i64, IntOp, Commutable>;
+}
+}
 
 // Neon Narrowing 3-register vector intrinsics,
 //   source operand element sizes of 16, 32 and 64 bits:
@@ -1619,6 +1828,47 @@
                              // imm6 = xxxxxx
 }
 
+// Same as N2VSh_QHSD, except the instructions have a differnt interpretation of
+// the shift amount.  This helps the disassembler.
+let NSF = NVectorShift2Frm, NSForm = NVectorShift2Frm.Value in {
+multiclass N2VSh_QHSD2<bit op24, bit op23, bits<4> op11_8, bit op4,
+                      InstrItinClass itin, string OpcodeStr, string Dt,
+                      SDNode OpNode> {
+  // 64-bit vector types.
+  def v8i8  : N2VDSh<op24, op23, op11_8, 0, op4, itin,
+                     OpcodeStr, !strconcat(Dt, "8"), v8i8, OpNode> {
+    let Inst{21-19} = 0b001; // imm6 = 001xxx
+  }
+  def v4i16 : N2VDSh<op24, op23, op11_8, 0, op4, itin,
+                     OpcodeStr, !strconcat(Dt, "16"), v4i16, OpNode> {
+    let Inst{21-20} = 0b01;  // imm6 = 01xxxx
+  }
+  def v2i32 : N2VDSh<op24, op23, op11_8, 0, op4, itin,
+                     OpcodeStr, !strconcat(Dt, "32"), v2i32, OpNode> {
+    let Inst{21} = 0b1;      // imm6 = 1xxxxx
+  }
+  def v1i64 : N2VDSh<op24, op23, op11_8, 1, op4, itin,
+                     OpcodeStr, !strconcat(Dt, "64"), v1i64, OpNode>;
+                             // imm6 = xxxxxx
+
+  // 128-bit vector types.
+  def v16i8 : N2VQSh<op24, op23, op11_8, 0, op4, itin,
+                     OpcodeStr, !strconcat(Dt, "8"), v16i8, OpNode> {
+    let Inst{21-19} = 0b001; // imm6 = 001xxx
+  }
+  def v8i16 : N2VQSh<op24, op23, op11_8, 0, op4, itin,
+                     OpcodeStr, !strconcat(Dt, "16"), v8i16, OpNode> {
+    let Inst{21-20} = 0b01;  // imm6 = 01xxxx
+  }
+  def v4i32 : N2VQSh<op24, op23, op11_8, 0, op4, itin,
+                     OpcodeStr, !strconcat(Dt, "32"), v4i32, OpNode> {
+    let Inst{21} = 0b1;      // imm6 = 1xxxxx
+  }
+  def v2i64 : N2VQSh<op24, op23, op11_8, 1, op4, itin,
+                     OpcodeStr, !strconcat(Dt, "64"), v2i64, OpNode>;
+                             // imm6 = xxxxxx
+}
+}
 
 // Neon Shift-Accumulate vector operations,
 //   element sizes of 8, 16, 32 and 64 bits:
@@ -1699,6 +1949,47 @@
                              // imm6 = xxxxxx
 }
 
+// Same as N2VShIns_QHSD, except the instructions have a differnt interpretation
+// of the shift amount.  This helps the disassembler.
+let NSF = NVectorShift2Frm, NSForm = NVectorShift2Frm.Value in {
+multiclass N2VShIns_QHSD2<bit op24, bit op23, bits<4> op11_8, bit op4,
+                         string OpcodeStr, SDNode ShOp> {
+  // 64-bit vector types.
+  def v8i8  : N2VDShIns<op24, op23, op11_8, 0, op4,
+                        OpcodeStr, "8", v8i8, ShOp> {
+    let Inst{21-19} = 0b001; // imm6 = 001xxx
+  }
+  def v4i16 : N2VDShIns<op24, op23, op11_8, 0, op4,
+                        OpcodeStr, "16", v4i16, ShOp> {
+    let Inst{21-20} = 0b01;  // imm6 = 01xxxx
+  }
+  def v2i32 : N2VDShIns<op24, op23, op11_8, 0, op4,
+                        OpcodeStr, "32", v2i32, ShOp> {
+    let Inst{21} = 0b1;      // imm6 = 1xxxxx
+  }
+  def v1i64 : N2VDShIns<op24, op23, op11_8, 1, op4,
+                        OpcodeStr, "64", v1i64, ShOp>;
+                             // imm6 = xxxxxx
+
+  // 128-bit vector types.
+  def v16i8 : N2VQShIns<op24, op23, op11_8, 0, op4,
+                        OpcodeStr, "8", v16i8, ShOp> {
+    let Inst{21-19} = 0b001; // imm6 = 001xxx
+  }
+  def v8i16 : N2VQShIns<op24, op23, op11_8, 0, op4,
+                        OpcodeStr, "16", v8i16, ShOp> {
+    let Inst{21-20} = 0b01;  // imm6 = 01xxxx
+  }
+  def v4i32 : N2VQShIns<op24, op23, op11_8, 0, op4,
+                        OpcodeStr, "32", v4i32, ShOp> {
+    let Inst{21} = 0b1;      // imm6 = 1xxxxx
+  }
+  def v2i64 : N2VQShIns<op24, op23, op11_8, 1, op4,
+                        OpcodeStr, "64", v2i64, ShOp>;
+                             // imm6 = xxxxxx
+}
+}
+
 // Neon Shift Long operations,
 //   element sizes of 8, 16, 32 bits:
 multiclass N2VLSh_QHS<bit op24, bit op23, bits<4> op11_8, bit op7, bit op6,
@@ -2329,18 +2620,21 @@
 // Vector Shifts.
 
 //   VSHL     : Vector Shift
-defm VSHLs    : N3VInt_QHSD<0, 0, 0b0100, 0, IIC_VSHLiD, IIC_VSHLiD, IIC_VSHLiQ,
-                            IIC_VSHLiQ, "vshl", "s", int_arm_neon_vshifts, 0>;
-defm VSHLu    : N3VInt_QHSD<1, 0, 0b0100, 0, IIC_VSHLiD, IIC_VSHLiD, IIC_VSHLiQ,
-                            IIC_VSHLiQ, "vshl", "u", int_arm_neon_vshiftu, 0>;
+defm VSHLs    : N3VInt_QHSD2<0,0, 0b0100, 0, IIC_VSHLiD, IIC_VSHLiD, IIC_VSHLiQ,
+                             IIC_VSHLiQ, "vshl", "s", int_arm_neon_vshifts, 0>;
+defm VSHLu    : N3VInt_QHSD2<1,0, 0b0100, 0, IIC_VSHLiD, IIC_VSHLiD, IIC_VSHLiQ,
+                             IIC_VSHLiQ, "vshl", "u", int_arm_neon_vshiftu, 0>;
 //   VSHL     : Vector Shift Left (Immediate)
-defm VSHLi    : N2VSh_QHSD<0, 1, 0b0101, 1, IIC_VSHLiD, "vshl", "i", NEONvshl>;
+//   (disassembly note: this has a different interpretation of the shift amont)
+defm VSHLi    : N2VSh_QHSD2<0, 1, 0b0101, 1, IIC_VSHLiD, "vshl", "i", NEONvshl>;
 //   VSHR     : Vector Shift Right (Immediate)
 defm VSHRs    : N2VSh_QHSD<0, 1, 0b0000, 1, IIC_VSHLiD, "vshr", "s", NEONvshrs>;
 defm VSHRu    : N2VSh_QHSD<1, 1, 0b0000, 1, IIC_VSHLiD, "vshr", "u", NEONvshru>;
 
 //   VSHLL    : Vector Shift Left Long
+//   (disassembly note: this has a different interpretation of the shift amont)
 defm VSHLLs   : N2VLSh_QHS<0, 1, 0b1010, 0, 0, 1, "vshll", "s", NEONvshlls>;
+//   (disassembly note: this has a different interpretation of the shift amont)
 defm VSHLLu   : N2VLSh_QHS<1, 1, 0b1010, 0, 0, 1, "vshll", "u", NEONvshllu>;
 
 //   VSHLL    : Vector Shift Left Long (with maximum shift count)
@@ -2350,6 +2644,8 @@
   : N2VLSh<op24, op23, op11_8, op7, op6, op4, OpcodeStr, Dt,
            ResTy, OpTy, OpNode> {
   let Inst{21-16} = op21_16;
+  let NSF    = NVdVmImmVSHLLFrm;       // For disassembly.
+  let NSForm = NVdVmImmVSHLLFrm.Value; // For disassembly.
 }
 def  VSHLLi8  : N2VLShMax<1, 1, 0b110010, 0b0011, 0, 0, 0, "vshll", "i8",
                           v8i16, v8i8, NEONvshlli>;
@@ -2363,10 +2659,10 @@
                            NEONvshrn>;
 
 //   VRSHL    : Vector Rounding Shift
-defm VRSHLs   : N3VInt_QHSD<0,0,0b0101,0, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
-                            IIC_VSHLi4Q, "vrshl", "s", int_arm_neon_vrshifts,0>;
-defm VRSHLu   : N3VInt_QHSD<1,0,0b0101,0, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
-                            IIC_VSHLi4Q, "vrshl", "u", int_arm_neon_vrshiftu,0>;
+defm VRSHLs   : N3VInt_QHSD2<0,0,0b0101,0,IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
+                             IIC_VSHLi4Q,"vrshl", "s", int_arm_neon_vrshifts,0>;
+defm VRSHLu   : N3VInt_QHSD2<1,0,0b0101,0,IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
+                             IIC_VSHLi4Q,"vrshl", "u", int_arm_neon_vrshiftu,0>;
 //   VRSHR    : Vector Rounding Shift Right
 defm VRSHRs   : N2VSh_QHSD<0,1,0b0010,1, IIC_VSHLi4D, "vrshr", "s", NEONvrshrs>;
 defm VRSHRu   : N2VSh_QHSD<1,1,0b0010,1, IIC_VSHLi4D, "vrshr", "u", NEONvrshru>;
@@ -2376,15 +2672,18 @@
                            NEONvrshrn>;
 
 //   VQSHL    : Vector Saturating Shift
-defm VQSHLs   : N3VInt_QHSD<0,0,0b0100,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
-                            IIC_VSHLi4Q, "vqshl", "s", int_arm_neon_vqshifts,0>;
-defm VQSHLu   : N3VInt_QHSD<1,0,0b0100,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
-                            IIC_VSHLi4Q, "vqshl", "u", int_arm_neon_vqshiftu,0>;
+defm VQSHLs   : N3VInt_QHSD2<0,0,0b0100,1,IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
+                             IIC_VSHLi4Q, "vqshl", "s", int_arm_neon_vqshifts,0>;
+defm VQSHLu   : N3VInt_QHSD2<1,0,0b0100,1,IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
+                             IIC_VSHLi4Q, "vqshl", "u", int_arm_neon_vqshiftu,0>;
 //   VQSHL    : Vector Saturating Shift Left (Immediate)
-defm VQSHLsi  : N2VSh_QHSD<0,1,0b0111,1, IIC_VSHLi4D, "vqshl", "s", NEONvqshls>;
-defm VQSHLui  : N2VSh_QHSD<1,1,0b0111,1, IIC_VSHLi4D, "vqshl", "u", NEONvqshlu>;
+//   (disassembly note: this has a different interpretation of the shift amont)
+defm VQSHLsi  : N2VSh_QHSD2<0,1,0b0111,1, IIC_VSHLi4D, "vqshl", "s", NEONvqshls>;
+//   (disassembly note: this has a different interpretation of the shift amont)
+defm VQSHLui  : N2VSh_QHSD2<1,1,0b0111,1, IIC_VSHLi4D, "vqshl", "u", NEONvqshlu>;
 //   VQSHLU   : Vector Saturating Shift Left (Immediate, Unsigned)
-defm VQSHLsu  : N2VSh_QHSD<1,1,0b0110,1, IIC_VSHLi4D, "vqshlu","s",NEONvqshlsu>;
+//   (disassembly note: this has a different interpretation of the shift amont)
+defm VQSHLsu  : N2VSh_QHSD2<1,1,0b0110,1, IIC_VSHLi4D, "vqshlu","s",NEONvqshlsu>;
 
 //   VQSHRN   : Vector Saturating Shift Right and Narrow
 defm VQSHRNs  : N2VNSh_HSD<0, 1, 0b1001, 0, 0, 1, IIC_VSHLi4D, "vqshrn", "s",
@@ -2397,12 +2696,12 @@
                            NEONvqshrnsu>;
 
 //   VQRSHL   : Vector Saturating Rounding Shift
-defm VQRSHLs  : N3VInt_QHSD<0,0,0b0101,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
-                            IIC_VSHLi4Q, "vqrshl", "s",
-                            int_arm_neon_vqrshifts, 0>;
-defm VQRSHLu  : N3VInt_QHSD<1,0,0b0101,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
-                            IIC_VSHLi4Q, "vqrshl", "u",
-                            int_arm_neon_vqrshiftu, 0>;
+defm VQRSHLs  : N3VInt_QHSD2<0,0,0b0101,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
+                             IIC_VSHLi4Q, "vqrshl", "s",
+                             int_arm_neon_vqrshifts, 0>;
+defm VQRSHLu  : N3VInt_QHSD2<1,0,0b0101,1, IIC_VSHLi4D, IIC_VSHLi4D, IIC_VSHLi4Q,
+                             IIC_VSHLi4Q, "vqrshl", "u",
+                             int_arm_neon_vqrshiftu, 0>;
 
 //   VQRSHRN  : Vector Saturating Rounding Shift Right and Narrow
 defm VQRSHRNs : N2VNSh_HSD<0, 1, 0b1001, 0, 1, 1, IIC_VSHLi4D, "vqrshrn", "s",
@@ -2422,7 +2721,8 @@
 defm VRSRAu   : N2VShAdd_QHSD<1, 1, 0b0011, 1, "vrsra", "u", NEONvrshru>;
 
 //   VSLI     : Vector Shift Left and Insert
-defm VSLI     : N2VShIns_QHSD<1, 1, 0b0101, 1, "vsli", NEONvsli>;
+//   (disassembly note: this has a different interpretation of the shift amont)
+defm VSLI     : N2VShIns_QHSD2<1, 1, 0b0101, 1, "vsli", NEONvsli>;
 //   VSRI     : Vector Shift Right and Insert
 defm VSRI     : N2VShIns_QHSD<1, 1, 0b0100, 1, "vsri", NEONvsri>;
 
@@ -2518,10 +2818,13 @@
 
 //   VMOV     : Vector Move (Register)
 
+// Mark these instructions as 2-register instructions to help the disassembler.
+let NSF = NVdVmImmFrm, NSForm = NVdVmImmFrm.Value in {
 def  VMOVDneon: N3VX<0, 0, 0b10, 0b0001, 0, 1, (outs DPR:$dst), (ins DPR:$src),
                     IIC_VMOVD, "vmov", "$dst, $src", "", []>;
 def  VMOVQ    : N3VX<0, 0, 0b10, 0b0001, 1, 1, (outs QPR:$dst), (ins QPR:$src),
                     IIC_VMOVD, "vmov", "$dst, $src", "", []>;
+}
 
 //   VMOV     : Vector Move (Immediate)
 
@@ -2762,6 +3065,7 @@
 
 //   VDUP     : Vector Duplicate Lane (from scalar to all elements)
 
+let NSF = NVdVmImmVDupLaneFrm, NSForm = NVdVmImmVDupLaneFrm.Value in {
 class VDUPLND<bits<2> op19_18, bits<2> op17_16,
               string OpcodeStr, string Dt, ValueType Ty>
   : N2V<0b11, 0b11, op19_18, op17_16, 0b11000, 0, 0,
@@ -2775,6 +3079,7 @@
         (outs QPR:$dst), (ins DPR:$src, nohash_imm:$lane), IIC_VMOVD,
         OpcodeStr, Dt, "$dst, $src[$lane]", "",
         [(set QPR:$dst, (ResTy (NEONvduplane (OpTy DPR:$src), imm:$lane)))]>;
+}
 
 // Inst{19-16} is partially specified depending on the element size.
 
@@ -2843,24 +3148,37 @@
 
 // Vector Conversions.
 
-//   VCVT     : Vector Convert Between Floating-Point and Integers
-def  VCVTf2sd : N2VD<0b11, 0b11, 0b10, 0b11, 0b01110, 0, "vcvt", "s32.f32",
-                     v2i32, v2f32, fp_to_sint>;
-def  VCVTf2ud : N2VD<0b11, 0b11, 0b10, 0b11, 0b01111, 0, "vcvt", "u32.f32",
-                     v2i32, v2f32, fp_to_uint>;
-def  VCVTs2fd : N2VD<0b11, 0b11, 0b10, 0b11, 0b01100, 0, "vcvt", "f32.s32",
-                     v2f32, v2i32, sint_to_fp>;
-def  VCVTu2fd : N2VD<0b11, 0b11, 0b10, 0b11, 0b01101, 0, "vcvt", "f32.u32",
-                     v2f32, v2i32, uint_to_fp>;
+let NSF = NVdVmImmVCVTFrm, NSForm = NVdVmImmVCVTFrm.Value in {
+class N2VDX<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
+           bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr,
+           string Dt, ValueType ResTy, ValueType OpTy, SDNode OpNode>
+  : N2VD<op24_23, op21_20, op19_18, op17_16, op11_7, op4, OpcodeStr, Dt,
+         ResTy, OpTy, OpNode>;
+class N2VQX<bits<2> op24_23, bits<2> op21_20, bits<2> op19_18,
+           bits<2> op17_16, bits<5> op11_7, bit op4, string OpcodeStr,
+           string Dt, ValueType ResTy, ValueType OpTy, SDNode OpNode>
+  : N2VQ<op24_23, op21_20, op19_18, op17_16, op11_7, op4, OpcodeStr, Dt,
+         ResTy, OpTy, OpNode>;
+}
 
-def  VCVTf2sq : N2VQ<0b11, 0b11, 0b10, 0b11, 0b01110, 0, "vcvt", "s32.f32",
-                     v4i32, v4f32, fp_to_sint>;
-def  VCVTf2uq : N2VQ<0b11, 0b11, 0b10, 0b11, 0b01111, 0, "vcvt", "u32.f32",
-                     v4i32, v4f32, fp_to_uint>;
-def  VCVTs2fq : N2VQ<0b11, 0b11, 0b10, 0b11, 0b01100, 0, "vcvt", "f32.s32",
-                     v4f32, v4i32, sint_to_fp>;
-def  VCVTu2fq : N2VQ<0b11, 0b11, 0b10, 0b11, 0b01101, 0, "vcvt", "f32.u32",
-                     v4f32, v4i32, uint_to_fp>;
+//   VCVT     : Vector Convert Between Floating-Point and Integers
+def  VCVTf2sd : N2VDX<0b11, 0b11, 0b10, 0b11, 0b01110, 0, "vcvt", "s32.f32",
+                      v2i32, v2f32, fp_to_sint>;
+def  VCVTf2ud : N2VDX<0b11, 0b11, 0b10, 0b11, 0b01111, 0, "vcvt", "u32.f32",
+                      v2i32, v2f32, fp_to_uint>;
+def  VCVTs2fd : N2VDX<0b11, 0b11, 0b10, 0b11, 0b01100, 0, "vcvt", "f32.s32",
+                      v2f32, v2i32, sint_to_fp>;
+def  VCVTu2fd : N2VDX<0b11, 0b11, 0b10, 0b11, 0b01101, 0, "vcvt", "f32.u32",
+                      v2f32, v2i32, uint_to_fp>;
+
+def  VCVTf2sq : N2VQX<0b11, 0b11, 0b10, 0b11, 0b01110, 0, "vcvt", "s32.f32",
+                      v4i32, v4f32, fp_to_sint>;
+def  VCVTf2uq : N2VQX<0b11, 0b11, 0b10, 0b11, 0b01111, 0, "vcvt", "u32.f32",
+                      v4i32, v4f32, fp_to_uint>;
+def  VCVTs2fq : N2VQX<0b11, 0b11, 0b10, 0b11, 0b01100, 0, "vcvt", "f32.s32",
+                      v4f32, v4i32, sint_to_fp>;
+def  VCVTu2fq : N2VQX<0b11, 0b11, 0b10, 0b11, 0b01101, 0, "vcvt", "f32.u32",
+                      v4f32, v4i32, uint_to_fp>;
 
 //   VCVT     : Vector Convert Between Floating-Point and Fixed-Point.
 def VCVTf2xsd : N2VCvtD<0, 1, 0b1111, 0, 1, "vcvt", "s32.f32",
@@ -2945,6 +3263,8 @@
 
 //   VEXT     : Vector Extract
 
+let NSF    = NVdVnVmImmVectorExtractFrm,
+    NSForm = NVdVnVmImmVectorExtractFrm.Value in {
 class VEXTd<string OpcodeStr, string Dt, ValueType Ty>
   : N3V<0,1,0b11,{?,?,?,?},0,0, (outs DPR:$dst),
         (ins DPR:$lhs, DPR:$rhs, i32imm:$index), IIC_VEXTD,
@@ -2958,6 +3278,7 @@
         OpcodeStr, Dt, "$dst, $lhs, $rhs, $index", "",
         [(set QPR:$dst, (Ty (NEONvext (Ty QPR:$lhs),
                                       (Ty QPR:$rhs), imm:$index)))]>;
+}
 
 def VEXTd8  : VEXTd<"vext", "8",  v8i8>;
 def VEXTd16 : VEXTd<"vext", "16", v4i16>;
@@ -3001,6 +3322,8 @@
 
 // Vector Table Lookup and Table Extension.
 
+let NSF = VTBLFrm, NSForm = VTBLFrm.Value in {
+
 //   VTBL     : Vector Table Lookup
 def  VTBL1
   : N3V<1,1,0b11,0b1000,0,0, (outs DPR:$dst),
@@ -3057,6 +3380,8 @@
                                DPR:$tbl2, DPR:$tbl3, DPR:$tbl4, DPR:$src)))]>;
 } // hasExtraSrcRegAllocReq = 1
 
+} // End of "let NSF = VTBLFrm, ..."
+
 //===----------------------------------------------------------------------===//
 // NEON instructions for single-precision FP math
 //===----------------------------------------------------------------------===//
diff --git a/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp b/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp
index 447afdd..bf7e919 100644
--- a/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp
+++ b/lib/Target/ARM/AsmPrinter/ARMAsmPrinter.cpp
@@ -120,7 +120,7 @@
     void printT2AddrModeImm8Operand(const MachineInstr *MI, int OpNum);
     void printT2AddrModeImm8s4Operand(const MachineInstr *MI, int OpNum);
     void printT2AddrModeImm8OffsetOperand(const MachineInstr *MI, int OpNum);
-    void printT2AddrModeImm8s4OffsetOperand(const MachineInstr *MI, int OpNum) {}
+    void printT2AddrModeImm8s4OffsetOperand(const MachineInstr *MI, int OpNum);
     void printT2AddrModeSoRegOperand(const MachineInstr *MI, int OpNum);
 
     void printCPSOptionOperand(const MachineInstr *MI, int OpNum) {}
@@ -431,16 +431,16 @@
   O << "[" << getRegisterName(MO1.getReg());
 
   if (!MO2.getReg()) {
-    if (ARM_AM::getAM2Offset(MO3.getImm()))  // Don't print +0.
+    if (ARM_AM::getAM2Offset(MO3.getImm())) // Don't print +0.
       O << ", #"
-        << (char)ARM_AM::getAM2Op(MO3.getImm())
+        << ARM_AM::getAddrOpcStr(ARM_AM::getAM2Op(MO3.getImm()))
         << ARM_AM::getAM2Offset(MO3.getImm());
     O << "]";
     return;
   }
 
   O << ", "
-    << (char)ARM_AM::getAM2Op(MO3.getImm())
+    << ARM_AM::getAddrOpcStr(ARM_AM::getAM2Op(MO3.getImm()))
     << getRegisterName(MO2.getReg());
 
   if (unsigned ShImm = ARM_AM::getAM2Offset(MO3.getImm()))
@@ -458,12 +458,12 @@
     unsigned ImmOffs = ARM_AM::getAM2Offset(MO2.getImm());
     assert(ImmOffs && "Malformed indexed load / store!");
     O << "#"
-      << (char)ARM_AM::getAM2Op(MO2.getImm())
+      << ARM_AM::getAddrOpcStr(ARM_AM::getAM2Op(MO2.getImm()))
       << ImmOffs;
     return;
   }
 
-  O << (char)ARM_AM::getAM2Op(MO2.getImm())
+  O << ARM_AM::getAddrOpcStr(ARM_AM::getAM2Op(MO2.getImm()))
     << getRegisterName(MO1.getReg());
 
   if (unsigned ShImm = ARM_AM::getAM2Offset(MO2.getImm()))
@@ -490,7 +490,7 @@
 
   if (unsigned ImmOffs = ARM_AM::getAM3Offset(MO3.getImm()))
     O << ", #"
-      << (char)ARM_AM::getAM3Op(MO3.getImm())
+      << ARM_AM::getAddrOpcStr(ARM_AM::getAM3Op(MO3.getImm()))
       << ImmOffs;
   O << "]";
 }
@@ -508,7 +508,7 @@
   unsigned ImmOffs = ARM_AM::getAM3Offset(MO2.getImm());
   assert(ImmOffs && "Malformed indexed load / store!");
   O << "#"
-    << (char)ARM_AM::getAM3Op(MO2.getImm())
+    << ARM_AM::getAddrOpcStr(ARM_AM::getAM3Op(MO2.getImm()))
     << ImmOffs;
 }
 
@@ -553,7 +553,7 @@
 
   if (unsigned ImmOffs = ARM_AM::getAM5Offset(MO2.getImm())) {
     O << ", #"
-      << (char)ARM_AM::getAM5Op(MO2.getImm())
+      << ARM_AM::getAddrOpcStr(ARM_AM::getAM5Op(MO2.getImm()))
       << ImmOffs*4;
   }
   O << "]";
@@ -589,7 +589,7 @@
 
   const MachineOperand &MO1 = MI->getOperand(Op);
   assert(TargetRegisterInfo::isPhysicalRegister(MO1.getReg()));
-  O << "[pc, +" << getRegisterName(MO1.getReg()) << "]";
+  O << "[pc, " << getRegisterName(MO1.getReg()) << "]";
 }
 
 void
@@ -612,10 +612,11 @@
 ARMAsmPrinter::printThumbITMask(const MachineInstr *MI, int Op) {
   // (3 - the number of trailing zeros) is the number of then / else.
   unsigned Mask = MI->getOperand(Op).getImm();
+  unsigned CondBit0 = Mask >> 4 & 1;
   unsigned NumTZ = CountTrailingZeros_32(Mask);
   assert(NumTZ <= 3 && "Invalid IT mask!");
   for (unsigned Pos = 3, e = NumTZ; Pos > e; --Pos) {
-    bool T = (Mask & (1 << Pos)) == 0;
+    bool T = ((Mask >> Pos) & 1) == CondBit0;
     if (T)
       O << 't';
     else
@@ -647,7 +648,7 @@
   if (MO3.getReg())
     O << ", " << getRegisterName(MO3.getReg());
   else if (unsigned ImmOffs = MO2.getImm())
-    O << ", #+" << ImmOffs * Scale;
+    O << ", #" << ImmOffs * Scale;
   O << "]";
 }
 
@@ -669,7 +670,7 @@
   const MachineOperand &MO2 = MI->getOperand(Op+1);
   O << "[" << getRegisterName(MO1.getReg());
   if (unsigned ImmOffs = MO2.getImm())
-    O << ", #+" << ImmOffs*4;
+    O << ", #" << ImmOffs*4;
   O << "]";
 }
 
@@ -705,7 +706,7 @@
 
   unsigned OffImm = MO2.getImm();
   if (OffImm)  // Don't print +0.
-    O << ", #+" << OffImm;
+    O << ", #" << OffImm;
   O << "]";
 }
 
@@ -721,7 +722,7 @@
   if (OffImm < 0)
     O << ", #-" << -OffImm;
   else if (OffImm > 0)
-    O << ", #+" << OffImm;
+    O << ", #" << OffImm;
   O << "]";
 }
 
@@ -737,7 +738,7 @@
   if (OffImm < 0)
     O << ", #-" << -OffImm * 4;
   else if (OffImm > 0)
-    O << ", #+" << OffImm * 4;
+    O << ", #" << OffImm * 4;
   O << "]";
 }
 
@@ -749,7 +750,18 @@
   if (OffImm < 0)
     O << "#-" << -OffImm;
   else if (OffImm > 0)
-    O << "#+" << OffImm;
+    O << "#" << OffImm;
+}
+
+void ARMAsmPrinter::printT2AddrModeImm8s4OffsetOperand(const MachineInstr *MI,
+                                                     int OpNum) {
+  const MachineOperand &MO1 = MI->getOperand(OpNum);
+  int32_t OffImm = (int32_t)MO1.getImm() / 4;
+  // Don't print +0.
+  if (OffImm < 0)
+    O << "#-" << -OffImm * 4;
+  else if (OffImm > 0)
+    O << "#" << OffImm * 4;
 }
 
 void ARMAsmPrinter::printT2AddrModeSoRegOperand(const MachineInstr *MI,
diff --git a/lib/Target/ARM/AsmPrinter/ARMInstPrinter.cpp b/lib/Target/ARM/AsmPrinter/ARMInstPrinter.cpp
index c1d68fb..fba380d 100644
--- a/lib/Target/ARM/AsmPrinter/ARMInstPrinter.cpp
+++ b/lib/Target/ARM/AsmPrinter/ARMInstPrinter.cpp
@@ -28,7 +28,159 @@
 #undef MachineInstr
 #undef ARMAsmPrinter
 
-void ARMInstPrinter::printInst(const MCInst *MI) { printInstruction(MI); }
+static unsigned NextReg(unsigned Reg) {
+  switch (Reg) {
+  case ARM::D0:
+    return ARM::D1;
+  case ARM::D1:
+    return ARM::D2;
+  case ARM::D2:
+    return ARM::D3;
+  case ARM::D3:
+    return ARM::D4;
+  case ARM::D4:
+    return ARM::D5;
+  case ARM::D5:
+    return ARM::D6;
+  case ARM::D6:
+    return ARM::D7;
+  case ARM::D7:
+    return ARM::D8;
+  case ARM::D8:
+    return ARM::D9;
+  case ARM::D9:
+    return ARM::D10;
+  case ARM::D10:
+    return ARM::D11;
+  case ARM::D11:
+    return ARM::D12;
+  case ARM::D12:
+    return ARM::D13;
+  case ARM::D13:
+    return ARM::D14;
+  case ARM::D14:
+    return ARM::D15;
+  case ARM::D15:
+    return ARM::D16;
+  case ARM::D16:
+    return ARM::D17;
+  case ARM::D17:
+    return ARM::D18;
+  case ARM::D18:
+    return ARM::D19;
+  case ARM::D19:
+    return ARM::D20;
+  case ARM::D20:
+    return ARM::D21;
+  case ARM::D21:
+    return ARM::D22;
+  case ARM::D22:
+    return ARM::D23;
+  case ARM::D23:
+    return ARM::D24;
+  case ARM::D24:
+    return ARM::D25;
+  case ARM::D25:
+    return ARM::D26;
+  case ARM::D26:
+    return ARM::D27;
+  case ARM::D27:
+    return ARM::D28;
+  case ARM::D28:
+    return ARM::D29;
+  case ARM::D29:
+    return ARM::D30;
+  case ARM::D30:
+    return ARM::D31;
+
+  default:
+    assert(0 && "Unexpected register enum");
+  }
+}
+
+void ARMInstPrinter::printInst(const MCInst *MI) {
+  // Check for MOVs and print canonical forms, instead.
+  if (MI->getOpcode() == ARM::MOVs) {
+    const MCOperand &Dst = MI->getOperand(0);
+    const MCOperand &MO1 = MI->getOperand(1);
+    const MCOperand &MO2 = MI->getOperand(2);
+    const MCOperand &MO3 = MI->getOperand(3);
+
+    O << '\t' << ARM_AM::getShiftOpcStr(ARM_AM::getSORegShOp(MO3.getImm()));
+    printSBitModifierOperand(MI, 6);
+    printPredicateOperand(MI, 4);
+
+    O << '\t' << getRegisterName(Dst.getReg())
+      << ", " << getRegisterName(MO1.getReg());
+
+    if (ARM_AM::getSORegShOp(MO3.getImm()) == ARM_AM::rrx)
+      return;
+
+    O << ", ";
+
+    if (MO2.getReg()) {
+      O << getRegisterName(MO2.getReg());
+      assert(ARM_AM::getSORegOffset(MO3.getImm()) == 0);
+    } else {
+      O << "#" << ARM_AM::getSORegOffset(MO3.getImm());
+    }
+    return;
+  }
+
+  // A8.6.123 PUSH
+  if ((MI->getOpcode() == ARM::STM_UPD || MI->getOpcode() == ARM::t2STM_UPD) &&
+      MI->getOperand(0).getReg() == ARM::SP) {
+    const MCOperand &MO1 = MI->getOperand(2);
+    if (ARM_AM::getAM4SubMode(MO1.getImm()) == ARM_AM::db) {
+      O << '\t' << "push";
+      printPredicateOperand(MI, 3);
+      O << '\t';
+      printRegisterList(MI, 5);
+      return;
+    }
+  }
+
+  // A8.6.122 POP
+  if ((MI->getOpcode() == ARM::LDM_UPD || MI->getOpcode() == ARM::t2LDM_UPD) &&
+      MI->getOperand(0).getReg() == ARM::SP) {
+    const MCOperand &MO1 = MI->getOperand(2);
+    if (ARM_AM::getAM4SubMode(MO1.getImm()) == ARM_AM::ia) {
+      O << '\t' << "pop";
+      printPredicateOperand(MI, 3);
+      O << '\t';
+      printRegisterList(MI, 5);
+      return;
+    }
+  }
+
+  // A8.6.355 VPUSH
+  if ((MI->getOpcode() == ARM::VSTMS_UPD || MI->getOpcode() ==ARM::VSTMD_UPD) &&
+      MI->getOperand(0).getReg() == ARM::SP) {
+    const MCOperand &MO1 = MI->getOperand(2);
+    if (ARM_AM::getAM5SubMode(MO1.getImm()) == ARM_AM::db) {
+      O << '\t' << "vpush";
+      printPredicateOperand(MI, 3);
+      O << '\t';
+      printRegisterList(MI, 5);
+      return;
+    }
+  }
+
+  // A8.6.354 VPOP
+  if ((MI->getOpcode() == ARM::VLDMS_UPD || MI->getOpcode() ==ARM::VLDMD_UPD) &&
+      MI->getOperand(0).getReg() == ARM::SP) {
+    const MCOperand &MO1 = MI->getOperand(2);
+    if (ARM_AM::getAM5SubMode(MO1.getImm()) == ARM_AM::ia) {
+      O << '\t' << "vpop";
+      printPredicateOperand(MI, 3);
+      O << '\t';
+      printRegisterList(MI, 5);
+      return;
+    }
+  }
+
+  printInstruction(MI);
+ }
 
 void ARMInstPrinter::printOperand(const MCInst *MI, unsigned OpNo,
                                   const char *Modifier) {
@@ -36,6 +188,9 @@
   if (Op.isReg()) {
     unsigned Reg = Op.getReg();
     if (Modifier && strcmp(Modifier, "dregpair") == 0) {
+      O << '{' << getRegisterName(Reg) << ", "
+               << getRegisterName(NextReg(Reg)) << '}';
+#if 0
       // FIXME: Breaks e.g. ARM/vmul.ll.
       assert(0);
       /*
@@ -44,6 +199,7 @@
       O << '{'
       << getRegisterName(DRegLo) << ',' << getRegisterName(DRegHi)
       << '}';*/
+#endif
     } else if (Modifier && strcmp(Modifier, "lane") == 0) {
       assert(0);
       /*
@@ -56,7 +212,9 @@
       O << getRegisterName(Reg);
     }
   } else if (Op.isImm()) {
-    assert((Modifier == 0 || Modifier[0] == 0) && "No modifiers supported");
+    bool isCallOp = Modifier && !strcmp(Modifier, "call");
+    assert(isCallOp ||
+           ((Modifier == 0 || Modifier[0] == 0) && "No modifiers supported"));
     O << '#' << Op.getImm();
   } else {
     assert((Modifier == 0 || Modifier[0] == 0) && "No modifiers supported");
@@ -142,17 +300,17 @@
   O << "[" << getRegisterName(MO1.getReg());
   
   if (!MO2.getReg()) {
-    if (ARM_AM::getAM2Offset(MO3.getImm()))  // Don't print +0.
+    if (ARM_AM::getAM2Offset(MO3.getImm())) // Don't print +0.
       O << ", #"
-      << (char)ARM_AM::getAM2Op(MO3.getImm())
-      << ARM_AM::getAM2Offset(MO3.getImm());
+        << ARM_AM::getAddrOpcStr(ARM_AM::getAM2Op(MO3.getImm()))
+        << ARM_AM::getAM2Offset(MO3.getImm());
     O << "]";
     return;
   }
   
   O << ", "
-  << (char)ARM_AM::getAM2Op(MO3.getImm())
-  << getRegisterName(MO2.getReg());
+    << ARM_AM::getAddrOpcStr(ARM_AM::getAM2Op(MO3.getImm()))
+    << getRegisterName(MO2.getReg());
   
   if (unsigned ShImm = ARM_AM::getAM2Offset(MO3.getImm()))
     O << ", "
@@ -169,11 +327,14 @@
   if (!MO1.getReg()) {
     unsigned ImmOffs = ARM_AM::getAM2Offset(MO2.getImm());
     assert(ImmOffs && "Malformed indexed load / store!");
-    O << '#' << (char)ARM_AM::getAM2Op(MO2.getImm()) << ImmOffs;
+    O << '#'
+      << ARM_AM::getAddrOpcStr(ARM_AM::getAM2Op(MO2.getImm()))
+      << ImmOffs;
     return;
   }
   
-  O << (char)ARM_AM::getAM2Op(MO2.getImm()) << getRegisterName(MO1.getReg());
+  O << ARM_AM::getAddrOpcStr(ARM_AM::getAM2Op(MO2.getImm()))
+    << getRegisterName(MO1.getReg());
   
   if (unsigned ShImm = ARM_AM::getAM2Offset(MO2.getImm()))
     O << ", "
@@ -196,8 +357,8 @@
   
   if (unsigned ImmOffs = ARM_AM::getAM3Offset(MO3.getImm()))
     O << ", #"
-    << (char)ARM_AM::getAM3Op(MO3.getImm())
-    << ImmOffs;
+      << ARM_AM::getAddrOpcStr(ARM_AM::getAM3Op(MO3.getImm()))
+      << ImmOffs;
   O << ']';
 }
 
@@ -214,9 +375,9 @@
   
   unsigned ImmOffs = ARM_AM::getAM3Offset(MO2.getImm());
   assert(ImmOffs && "Malformed indexed load / store!");
-  O << "#"
-  << (char)ARM_AM::getAM3Op(MO2.getImm())
-  << ImmOffs;
+  O << '#'
+    << ARM_AM::getAddrOpcStr(ARM_AM::getAM3Op(MO2.getImm()))
+    << ImmOffs;
 }
 
 
@@ -259,7 +420,7 @@
   
   if (unsigned ImmOffs = ARM_AM::getAM5Offset(MO2.getImm())) {
     O << ", #"
-      << (char)ARM_AM::getAM5Op(MO2.getImm())
+      << ARM_AM::getAddrOpcStr(ARM_AM::getAM5Op(MO2.getImm()))
       << ImmOffs*4;
   }
   O << "]";
@@ -298,14 +459,56 @@
 
 void ARMInstPrinter::printRegisterList(const MCInst *MI, unsigned OpNum) {
   O << "{";
-  // Always skip the first operand, it's the optional (and implicit writeback).
-  for (unsigned i = OpNum+1, e = MI->getNumOperands(); i != e; ++i) {
-    if (i != OpNum+1) O << ", ";
+  for (unsigned i = OpNum, e = MI->getNumOperands(); i != e; ++i) {
+    if (i != OpNum) O << ", ";
     O << getRegisterName(MI->getOperand(i).getReg());
   }
   O << "}";
 }
 
+void ARMInstPrinter::printCPSOptionOperand(const MCInst *MI, unsigned OpNum) {
+  const MCOperand &Op = MI->getOperand(OpNum);
+  unsigned option = Op.getImm();
+  unsigned mode = option & 31;
+  bool changemode = option >> 5 & 1;
+  unsigned AIF = option >> 6 & 7;
+  unsigned imod = option >> 9 & 3;
+  if (imod == 2)
+    O << "ie";
+  else if (imod == 3)
+    O << "id";
+  O << '\t';
+  if (imod > 1) {
+    if (AIF & 4) O << 'a';
+    if (AIF & 2) O << 'i';
+    if (AIF & 1) O << 'f';
+    if (AIF > 0 && changemode) O << ", ";
+  }
+  if (changemode)
+    O << '#' << mode;
+}
+
+void ARMInstPrinter::printMSRMaskOperand(const MCInst *MI, unsigned OpNum) {
+  const MCOperand &Op = MI->getOperand(OpNum);
+  unsigned Mask = Op.getImm();
+  if (Mask) {
+    O << '_';
+    if (Mask & 8) O << 'f';
+    if (Mask & 4) O << 's';
+    if (Mask & 2) O << 'x';
+    if (Mask & 1) O << 'c';
+  }
+}
+
+void ARMInstPrinter::printNegZeroOperand(const MCInst *MI, unsigned OpNum){
+  const MCOperand &Op = MI->getOperand(OpNum);
+  O << '#';
+  if (Op.getImm() < 0)
+    O << '-' << (-Op.getImm() - 1);
+  else
+    O << Op.getImm();
+}
+
 void ARMInstPrinter::printPredicateOperand(const MCInst *MI, unsigned OpNum) {
   ARMCC::CondCodes CC = (ARMCC::CondCodes)MI->getOperand(OpNum).getImm();
   if (CC != ARMCC::AL)
@@ -347,3 +550,191 @@
 void ARMInstPrinter::printThumbS4ImmOperand(const MCInst *MI, unsigned OpNum) {
   O << "#" <<  MI->getOperand(OpNum).getImm() * 4;
 }
+
+void ARMInstPrinter::printThumbITMask(const MCInst *MI, unsigned OpNum) {
+  // (3 - the number of trailing zeros) is the number of then / else.
+  unsigned Mask = MI->getOperand(OpNum).getImm();
+  unsigned CondBit0 = Mask >> 4 & 1;
+  unsigned NumTZ = CountTrailingZeros_32(Mask);
+  assert(NumTZ <= 3 && "Invalid IT mask!");
+  for (unsigned Pos = 3, e = NumTZ; Pos > e; --Pos) {
+    bool T = ((Mask >> Pos) & 1) == CondBit0;
+    if (T)
+      O << 't';
+    else
+      O << 'e';
+  }
+}
+
+void ARMInstPrinter::printThumbAddrModeRROperand(const MCInst *MI, unsigned Op)
+{
+  const MCOperand &MO1 = MI->getOperand(Op);
+  const MCOperand &MO2 = MI->getOperand(Op+1);
+  O << "[" << getRegisterName(MO1.getReg());
+  O << ", " << getRegisterName(MO2.getReg()) << "]";
+}
+
+void ARMInstPrinter::printThumbAddrModeRI5Operand(const MCInst *MI, unsigned Op,
+                                                  unsigned Scale) {
+  const MCOperand &MO1 = MI->getOperand(Op);
+  const MCOperand &MO2 = MI->getOperand(Op+1);
+  const MCOperand &MO3 = MI->getOperand(Op+2);
+
+  if (!MO1.isReg()) {   // FIXME: This is for CP entries, but isn't right.
+    printOperand(MI, Op);
+    return;
+  }
+
+  O << "[" << getRegisterName(MO1.getReg());
+  if (MO3.getReg())
+    O << ", " << getRegisterName(MO3.getReg());
+  else if (unsigned ImmOffs = MO2.getImm())
+    O << ", #" << ImmOffs * Scale;
+  O << "]";
+}
+
+void ARMInstPrinter::printThumbAddrModeS1Operand(const MCInst *MI, unsigned Op)
+{
+  printThumbAddrModeRI5Operand(MI, Op, 1);
+}
+
+void ARMInstPrinter::printThumbAddrModeS2Operand(const MCInst *MI, unsigned Op)
+{
+  printThumbAddrModeRI5Operand(MI, Op, 2);
+}
+
+void ARMInstPrinter::printThumbAddrModeS4Operand(const MCInst *MI, unsigned Op)
+{
+  printThumbAddrModeRI5Operand(MI, Op, 4);
+}
+
+void ARMInstPrinter::printThumbAddrModeSPOperand(const MCInst *MI,unsigned Op) {
+  const MCOperand &MO1 = MI->getOperand(Op);
+  const MCOperand &MO2 = MI->getOperand(Op+1);
+  O << "[" << getRegisterName(MO1.getReg());
+  if (unsigned ImmOffs = MO2.getImm())
+    O << ", #" << ImmOffs*4;
+  O << "]";
+}
+
+void ARMInstPrinter::printTBAddrMode(const MCInst *MI, unsigned OpNum) {
+  O << "[pc, " << getRegisterName(MI->getOperand(OpNum).getReg());
+  if (MI->getOpcode() == ARM::t2TBH)
+    O << ", lsl #1";
+  O << ']';
+}
+
+// Constant shifts t2_so_reg is a 2-operand unit corresponding to the Thumb2
+// register with shift forms.
+// REG 0   0           - e.g. R5
+// REG IMM, SH_OPC     - e.g. R5, LSL #3
+void ARMInstPrinter::printT2SOOperand(const MCInst *MI, unsigned OpNum) {
+  const MCOperand &MO1 = MI->getOperand(OpNum);
+  const MCOperand &MO2 = MI->getOperand(OpNum+1);
+
+  unsigned Reg = MO1.getReg();
+  O << getRegisterName(Reg);
+
+  // Print the shift opc.
+  O << ", "
+    << ARM_AM::getShiftOpcStr(ARM_AM::getSORegShOp(MO2.getImm()))
+    << " ";
+
+  assert(MO2.isImm() && "Not a valid t2_so_reg value!");
+  O << "#" << ARM_AM::getSORegOffset(MO2.getImm());
+}
+
+void ARMInstPrinter::printT2AddrModeImm12Operand(const MCInst *MI,
+                                                 unsigned OpNum) {
+  const MCOperand &MO1 = MI->getOperand(OpNum);
+  const MCOperand &MO2 = MI->getOperand(OpNum+1);
+
+  O << "[" << getRegisterName(MO1.getReg());
+
+  unsigned OffImm = MO2.getImm();
+  if (OffImm)  // Don't print +0.
+    O << ", #" << OffImm;
+  O << "]";
+}
+
+void ARMInstPrinter::printT2AddrModeImm8Operand(const MCInst *MI,
+                                                unsigned OpNum) {
+  const MCOperand &MO1 = MI->getOperand(OpNum);
+  const MCOperand &MO2 = MI->getOperand(OpNum+1);
+
+  O << "[" << getRegisterName(MO1.getReg());
+
+  int32_t OffImm = (int32_t)MO2.getImm();
+  // Don't print +0.
+  if (OffImm < 0)
+    O << ", #-" << -OffImm;
+  else if (OffImm > 0)
+    O << ", #" << OffImm;
+  O << "]";
+}
+
+void ARMInstPrinter::printT2AddrModeImm8s4Operand(const MCInst *MI,
+                                                 unsigned OpNum) {
+  const MCOperand &MO1 = MI->getOperand(OpNum);
+  const MCOperand &MO2 = MI->getOperand(OpNum+1);
+
+  O << "[" << getRegisterName(MO1.getReg());
+
+  int32_t OffImm = (int32_t)MO2.getImm() / 4;
+  // Don't print +0.
+  if (OffImm < 0)
+    O << ", #-" << -OffImm * 4;
+  else if (OffImm > 0)
+    O << ", #" << OffImm * 4;
+  O << "]";
+}
+
+void ARMInstPrinter::printT2AddrModeImm8OffsetOperand(const MCInst *MI,
+                                                     unsigned OpNum) {
+  const MCOperand &MO1 = MI->getOperand(OpNum);
+  int32_t OffImm = (int32_t)MO1.getImm();
+  // Don't print +0.
+  if (OffImm < 0)
+    O << "#-" << -OffImm;
+  else if (OffImm > 0)
+    O << "#" << OffImm;
+}
+
+void ARMInstPrinter::printT2AddrModeImm8s4OffsetOperand(const MCInst *MI,
+                                                        unsigned OpNum) {
+  const MCOperand &MO1 = MI->getOperand(OpNum);
+  int32_t OffImm = (int32_t)MO1.getImm() / 4;
+  // Don't print +0.
+  if (OffImm < 0)
+    O << "#-" << -OffImm * 4;
+  else if (OffImm > 0)
+    O << "#" << OffImm * 4;
+}
+
+void ARMInstPrinter::printT2AddrModeSoRegOperand(const MCInst *MI,
+                                                 unsigned OpNum) {
+  const MCOperand &MO1 = MI->getOperand(OpNum);
+  const MCOperand &MO2 = MI->getOperand(OpNum+1);
+  const MCOperand &MO3 = MI->getOperand(OpNum+2);
+
+  O << "[" << getRegisterName(MO1.getReg());
+
+  assert(MO2.getReg() && "Invalid so_reg load / store address!");
+  O << ", " << getRegisterName(MO2.getReg());
+
+  unsigned ShAmt = MO3.getImm();
+  if (ShAmt) {
+    assert(ShAmt <= 3 && "Not a valid Thumb2 addressing mode!");
+    O << ", lsl #" << ShAmt;
+  }
+  O << "]";
+}
+
+void ARMInstPrinter::printVFPf32ImmOperand(const MCInst *MI, unsigned OpNum) {
+  O << '#' << MI->getOperand(OpNum).getImm();
+}
+
+void ARMInstPrinter::printVFPf64ImmOperand(const MCInst *MI, unsigned OpNum) {
+  O << '#' << MI->getOperand(OpNum).getImm();
+}
+
diff --git a/lib/Target/ARM/AsmPrinter/ARMInstPrinter.h b/lib/Target/ARM/AsmPrinter/ARMInstPrinter.h
index 9a3cbc3..ceb641d 100644
--- a/lib/Target/ARM/AsmPrinter/ARMInstPrinter.h
+++ b/lib/Target/ARM/AsmPrinter/ARMInstPrinter.h
@@ -54,26 +54,26 @@
   void printBitfieldInvMaskImmOperand(const MCInst *MI, unsigned OpNum);
 
   void printThumbS4ImmOperand(const MCInst *MI, unsigned OpNum);
-  void printThumbITMask(const MCInst *MI, unsigned OpNum) {}
-  void printThumbAddrModeRROperand(const MCInst *MI, unsigned OpNum) {}
+  void printThumbITMask(const MCInst *MI, unsigned OpNum);
+  void printThumbAddrModeRROperand(const MCInst *MI, unsigned OpNum);
   void printThumbAddrModeRI5Operand(const MCInst *MI, unsigned OpNum,
-                                    unsigned Scale) {}
-  void printThumbAddrModeS1Operand(const MCInst *MI, unsigned OpNum) {}
-  void printThumbAddrModeS2Operand(const MCInst *MI, unsigned OpNum) {}
-  void printThumbAddrModeS4Operand(const MCInst *MI, unsigned OpNum) {}
-  void printThumbAddrModeSPOperand(const MCInst *MI, unsigned OpNum) {}
+                                    unsigned Scale);
+  void printThumbAddrModeS1Operand(const MCInst *MI, unsigned OpNum);
+  void printThumbAddrModeS2Operand(const MCInst *MI, unsigned OpNum);
+  void printThumbAddrModeS4Operand(const MCInst *MI, unsigned OpNum);
+  void printThumbAddrModeSPOperand(const MCInst *MI, unsigned OpNum);
   
-  void printT2SOOperand(const MCInst *MI, unsigned OpNum) {}
-  void printT2AddrModeImm12Operand(const MCInst *MI, unsigned OpNum) {}
-  void printT2AddrModeImm8Operand(const MCInst *MI, unsigned OpNum) {}
-  void printT2AddrModeImm8s4Operand(const MCInst *MI, unsigned OpNum) {}
-  void printT2AddrModeImm8OffsetOperand(const MCInst *MI, unsigned OpNum) {}
-  void printT2AddrModeImm8s4OffsetOperand(const MCInst *MI, unsigned OpNum) {}
-  void printT2AddrModeSoRegOperand(const MCInst *MI, unsigned OpNum) {}
+  void printT2SOOperand(const MCInst *MI, unsigned OpNum);
+  void printT2AddrModeImm12Operand(const MCInst *MI, unsigned OpNum);
+  void printT2AddrModeImm8Operand(const MCInst *MI, unsigned OpNum);
+  void printT2AddrModeImm8s4Operand(const MCInst *MI, unsigned OpNum);
+  void printT2AddrModeImm8OffsetOperand(const MCInst *MI, unsigned OpNum);
+  void printT2AddrModeImm8s4OffsetOperand(const MCInst *MI, unsigned OpNum);
+  void printT2AddrModeSoRegOperand(const MCInst *MI, unsigned OpNum);
   
-  void printCPSOptionOperand(const MCInst *MI, unsigned OpNum) {}
-  void printMSRMaskOperand(const MCInst *MI, unsigned OpNum) {}
-  void printNegZeroOperand(const MCInst *MI, unsigned OpNum) {}
+  void printCPSOptionOperand(const MCInst *MI, unsigned OpNum);
+  void printMSRMaskOperand(const MCInst *MI, unsigned OpNum);
+  void printNegZeroOperand(const MCInst *MI, unsigned OpNum);
   void printPredicateOperand(const MCInst *MI, unsigned OpNum);
   void printMandatoryPredicateOperand(const MCInst *MI, unsigned OpNum);
   void printSBitModifierOperand(const MCInst *MI, unsigned OpNum);
@@ -82,10 +82,10 @@
                           const char *Modifier);
   void printJTBlockOperand(const MCInst *MI, unsigned OpNum) {}
   void printJT2BlockOperand(const MCInst *MI, unsigned OpNum) {}
-  void printTBAddrMode(const MCInst *MI, unsigned OpNum) {}
+  void printTBAddrMode(const MCInst *MI, unsigned OpNum);
   void printNoHashImmediate(const MCInst *MI, unsigned OpNum);
-  void printVFPf32ImmOperand(const MCInst *MI, int OpNum) {}
-  void printVFPf64ImmOperand(const MCInst *MI, int OpNum) {}
+  void printVFPf32ImmOperand(const MCInst *MI, unsigned OpNum);
+  void printVFPf64ImmOperand(const MCInst *MI, unsigned OpNum);
   void printHex8ImmOperand(const MCInst *MI, int OpNum) {}
   void printHex16ImmOperand(const MCInst *MI, int OpNum) {}
   void printHex32ImmOperand(const MCInst *MI, int OpNum) {}
diff --git a/test/CodeGen/ARM/2009-10-27-double-align.ll b/test/CodeGen/ARM/2009-10-27-double-align.ll
index a4e7685..f17d059 100644
--- a/test/CodeGen/ARM/2009-10-27-double-align.ll
+++ b/test/CodeGen/ARM/2009-10-27-double-align.ll
@@ -4,8 +4,8 @@
 
 define arm_aapcscc void @g() {
 entry:
-;CHECK: [sp, #+8]
-;CHECK: [sp, #+12]
+;CHECK: [sp, #8]
+;CHECK: [sp, #12]
 ;CHECK: [sp]
         tail call arm_aapcscc  void (i8*, ...)* @f(i8* getelementptr ([1 x i8]* @.str, i32 0, i32 0), i32 1, double 2.000000e+00, i32 3, double 4.000000e+00)
         ret void
diff --git a/test/CodeGen/ARM/2009-10-30.ll b/test/CodeGen/ARM/2009-10-30.ll
index 90a5bd2..87d1a8b 100644
--- a/test/CodeGen/ARM/2009-10-30.ll
+++ b/test/CodeGen/ARM/2009-10-30.ll
@@ -6,7 +6,7 @@
 entry:
 ;CHECK: sub	sp, sp, #4
 ;CHECK: add	r{{[0-9]+}}, sp, #8
-;CHECK: str	r{{[0-9]+}}, [sp], #+4
+;CHECK: str	r{{[0-9]+}}, [sp], #4
 ;CHECK: bx	lr
 	%ap = alloca i8*, align 4
 	%ap1 = bitcast i8** %ap to i8*
diff --git a/test/CodeGen/ARM/arm-negative-stride.ll b/test/CodeGen/ARM/arm-negative-stride.ll
index 52ab871..fb0f8ff 100644
--- a/test/CodeGen/ARM/arm-negative-stride.ll
+++ b/test/CodeGen/ARM/arm-negative-stride.ll
@@ -5,7 +5,7 @@
 
 define void @test(i32* %P, i32 %A, i32 %i) nounwind {
 entry:
-; CHECK: str r1, [{{r.*}}, +{{r.*}}, lsl #2]
+; CHECK: str r1, [{{r.*}}, {{r.*}}, lsl #2]
         icmp eq i32 %i, 0               ; <i1>:0 [#uses=1]
         br i1 %0, label %return, label %bb
 
diff --git a/test/CodeGen/ARM/globals.ll b/test/CodeGen/ARM/globals.ll
index 886c0d5..adb4497 100644
--- a/test/CodeGen/ARM/globals.ll
+++ b/test/CodeGen/ARM/globals.ll
@@ -41,7 +41,7 @@
 ; DarwinPIC: _test1:
 ; DarwinPIC: 	ldr r0, LCPI1_0
 ; DarwinPIC: LPC1_0:
-; DarwinPIC:    ldr r0, [pc, +r0]
+; DarwinPIC:    ldr r0, [pc, r0]
 ; DarwinPIC:    ldr r0, [r0]
 ; DarwinPIC:    bx lr
 
@@ -63,7 +63,7 @@
 	
 ; LinuxPIC: .LPC1_0:
 ; LinuxPIC: 	add r0, pc, r0
-; LinuxPIC: 	ldr r0, [r1, +r0]
+; LinuxPIC: 	ldr r0, [r1, r0]
 ; LinuxPIC: 	ldr r0, [r0]
 ; LinuxPIC: 	bx lr
 
diff --git a/test/CodeGen/ARM/ldrd.ll b/test/CodeGen/ARM/ldrd.ll
index c366e2d..895562a 100644
--- a/test/CodeGen/ARM/ldrd.ll
+++ b/test/CodeGen/ARM/ldrd.ll
@@ -10,10 +10,10 @@
 ;V6:   ldrd r2, [r2]
 
 ;V5:   ldr r3, [r2]
-;V5:   ldr r2, [r2, #+4]
+;V5:   ldr r2, [r2, #4]
 
 ;EABI: ldr r3, [r2]
-;EABI: ldr r2, [r2, #+4]
+;EABI: ldr r2, [r2, #4]
 
 	%0 = load i64** @b, align 4
 	%1 = load i64* %0, align 4
diff --git a/test/CodeGen/ARM/str_pre-2.ll b/test/CodeGen/ARM/str_pre-2.ll
index f8d3df2..553cd64 100644
--- a/test/CodeGen/ARM/str_pre-2.ll
+++ b/test/CodeGen/ARM/str_pre-2.ll
@@ -1,5 +1,5 @@
 ; RUN: llc < %s -mtriple=arm-linux-gnu | grep {str.*\\!}
-; RUN: llc < %s -mtriple=arm-linux-gnu | grep {ldr.*\\\[.*\], #+4}
+; RUN: llc < %s -mtriple=arm-linux-gnu | grep {ldr.*\\\[.*\], #4}
 
 @b = external global i64*
 
diff --git a/test/CodeGen/ARM/tls2.ll b/test/CodeGen/ARM/tls2.ll
index d932f90..57370c4 100644
--- a/test/CodeGen/ARM/tls2.ll
+++ b/test/CodeGen/ARM/tls2.ll
@@ -7,7 +7,7 @@
 
 define i32 @f() {
 ; CHECK-NONPIC: f:
-; CHECK-NONPIC: ldr {{r.}}, [pc, +{{r.}}]
+; CHECK-NONPIC: ldr {{r.}}, [pc, {{r.}}]
 ; CHECK-NONPIC: i(gottpoff)
 ; CHECK-PIC: f:
 ; CHECK-PIC: __tls_get_addr
@@ -18,7 +18,7 @@
 
 define i32* @g() {
 ; CHECK-NONPIC: g:
-; CHECK-NONPIC: ldr {{r.}}, [pc, +{{r.}}]
+; CHECK-NONPIC: ldr {{r.}}, [pc, {{r.}}]
 ; CHECK-NONPIC: i(gottpoff)
 ; CHECK-PIC: g:
 ; CHECK-PIC: __tls_get_addr
diff --git a/test/CodeGen/Thumb2/ldr-str-imm12.ll b/test/CodeGen/Thumb2/ldr-str-imm12.ll
index f007b5c..55cdac9 100644
--- a/test/CodeGen/Thumb2/ldr-str-imm12.ll
+++ b/test/CodeGen/Thumb2/ldr-str-imm12.ll
@@ -22,7 +22,7 @@
 
 define arm_apcscc %union.rec* @Manifest(%union.rec* %x, %union.rec* %env, %struct.STYLE* %style, %union.rec** %bthr, %union.rec** %fthr, %union.rec** %target, %union.rec** %crs, i32 %ok, i32 %need_expand, %union.rec** %enclose, i32 %fcr) nounwind {
 entry:
-; CHECK:       ldr.w	r9, [r7, #+28]
+; CHECK:       ldr.w	r9, [r7, #28]
   %xgaps.i = alloca [32 x %union.rec*], align 4   ; <[32 x %union.rec*]*> [#uses=0]
   %ycomp.i = alloca [32 x %union.rec*], align 4   ; <[32 x %union.rec*]*> [#uses=0]
   br i1 false, label %bb, label %bb20
@@ -50,9 +50,9 @@
 bb420:                                            ; preds = %bb20, %bb20
 ; CHECK: bb420
 ; CHECK: str r{{[0-7]}}, [sp]
-; CHECK: str r{{[0-7]}}, [sp, #+4]
-; CHECK: str r{{[0-7]}}, [sp, #+8]
-; CHECK: str{{(.w)?}} r{{[0-9]+}}, [sp, #+24]
+; CHECK: str r{{[0-7]}}, [sp, #4]
+; CHECK: str r{{[0-7]}}, [sp, #8]
+; CHECK: str{{(.w)?}} r{{[0-9]+}}, [sp, #24]
   store %union.rec* null, %union.rec** @zz_hold, align 4
   store %union.rec* null, %union.rec** @zz_res, align 4
   store %union.rec* %x, %union.rec** @zz_hold, align 4
diff --git a/test/CodeGen/Thumb2/thumb2-ldr.ll b/test/CodeGen/Thumb2/thumb2-ldr.ll
index 94888fd..88434f1 100644
--- a/test/CodeGen/Thumb2/thumb2-ldr.ll
+++ b/test/CodeGen/Thumb2/thumb2-ldr.ll
@@ -11,7 +11,7 @@
 define i32 @f2(i32* %v) {
 entry:
 ; CHECK: f2:
-; CHECK: ldr.w r0, [r0, #+4092]
+; CHECK: ldr.w r0, [r0, #4092]
         %tmp2 = getelementptr i32* %v, i32 1023
         %tmp = load i32* %tmp2
         ret i32 %tmp
diff --git a/test/CodeGen/Thumb2/thumb2-ldrh.ll b/test/CodeGen/Thumb2/thumb2-ldrh.ll
index f1fb79c..fee97bf 100644
--- a/test/CodeGen/Thumb2/thumb2-ldrh.ll
+++ b/test/CodeGen/Thumb2/thumb2-ldrh.ll
@@ -11,7 +11,7 @@
 define i16 @f2(i16* %v) {
 entry:
 ; CHECK: f2:
-; CHECK: ldrh.w r0, [r0, #+2046]
+; CHECK: ldrh.w r0, [r0, #2046]
         %tmp2 = getelementptr i16* %v, i16 1023
         %tmp = load i16* %tmp2
         ret i16 %tmp
diff --git a/test/CodeGen/Thumb2/thumb2-str.ll b/test/CodeGen/Thumb2/thumb2-str.ll
index 3eeec8c..11bb936 100644
--- a/test/CodeGen/Thumb2/thumb2-str.ll
+++ b/test/CodeGen/Thumb2/thumb2-str.ll
@@ -9,7 +9,7 @@
 
 define i32 @f2(i32 %a, i32* %v) {
 ; CHECK: f2:
-; CHECK: str.w r0, [r1, #+4092]
+; CHECK: str.w r0, [r1, #4092]
         %tmp2 = getelementptr i32* %v, i32 1023
         store i32 %a, i32* %tmp2
         ret i32 %a
diff --git a/test/CodeGen/Thumb2/thumb2-str_pre.ll b/test/CodeGen/Thumb2/thumb2-str_pre.ll
index 9af960b..1e6616a 100644
--- a/test/CodeGen/Thumb2/thumb2-str_pre.ll
+++ b/test/CodeGen/Thumb2/thumb2-str_pre.ll
@@ -2,7 +2,7 @@
 
 define void @test1(i32* %X, i32* %A, i32** %dest) {
 ; CHECK: test1
-; CHECK: str  r1, [r0, #+16]!
+; CHECK: str  r1, [r0, #16]!
         %B = load i32* %A               ; <i32> [#uses=1]
         %Y = getelementptr i32* %X, i32 4               ; <i32*> [#uses=2]
         store i32 %B, i32* %Y
@@ -12,7 +12,7 @@
 
 define i16* @test2(i16* %X, i32* %A) {
 ; CHECK: test2
-; CHECK: strh r1, [r0, #+8]!
+; CHECK: strh r1, [r0, #8]!
         %B = load i32* %A               ; <i32> [#uses=1]
         %Y = getelementptr i16* %X, i32 4               ; <i16*> [#uses=2]
         %tmp = trunc i32 %B to i16              ; <i16> [#uses=1]
diff --git a/test/CodeGen/Thumb2/thumb2-strb.ll b/test/CodeGen/Thumb2/thumb2-strb.ll
index 1ebb938..7978e7f 100644
--- a/test/CodeGen/Thumb2/thumb2-strb.ll
+++ b/test/CodeGen/Thumb2/thumb2-strb.ll
@@ -9,7 +9,7 @@
 
 define i8 @f2(i8 %a, i8* %v) {
 ; CHECK: f2:
-; CHECK: strb.w r0, [r1, #+4092]
+; CHECK: strb.w r0, [r1, #4092]
         %tmp2 = getelementptr i8* %v, i32 4092
         store i8 %a, i8* %tmp2
         ret i8 %a
diff --git a/test/CodeGen/Thumb2/thumb2-strh.ll b/test/CodeGen/Thumb2/thumb2-strh.ll
index b0eb8c1..97110a72 100644
--- a/test/CodeGen/Thumb2/thumb2-strh.ll
+++ b/test/CodeGen/Thumb2/thumb2-strh.ll
@@ -9,7 +9,7 @@
 
 define i16 @f2(i16 %a, i16* %v) {
 ; CHECK: f2:
-; CHECK: strh.w r0, [r1, #+4092]
+; CHECK: strh.w r0, [r1, #4092]
         %tmp2 = getelementptr i16* %v, i32 2046
         store i16 %a, i16* %tmp2
         ret i16 %a