X-Git-Url: http://plrg.eecs.uci.edu/git/?p=oota-llvm.git;a=blobdiff_plain;f=lib%2FTarget%2FARM%2FARMInstrThumb.td;h=1c94fe4ba214e167539b4a9a123d043f4e3e04d0;hp=63d8b48734e6d504e688c5df89e4352b9858e14d;hb=716c5d8a308a2257298f1f227edf7f7ae102cf4f;hpb=05b01567349dc6c98f9e68c1d4a639aca7ad5ac4

diff --git a/lib/Target/ARM/ARMInstrThumb.td b/lib/Target/ARM/ARMInstrThumb.td
index 63d8b48734e..1c94fe4ba21 100644
--- a/lib/Target/ARM/ARMInstrThumb.td
+++ b/lib/Target/ARM/ARMInstrThumb.td
@@ -1,4 +1,4 @@
-//===- ARMInstrThumb.td - Thumb support for ARM ------------*- tablegen -*-===//
+//===-- ARMInstrThumb.td - Thumb support for ARM -----------*- tablegen -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
@@ -21,7 +21,7 @@ def ARMtcall : SDNode<"ARMISD::tCALL", SDT_ARMcall,
 
 def imm_sr_XFORM: SDNodeXForm<imm, [{
   unsigned Imm = N->getZExtValue();
-  return CurDAG->getTargetConstant((Imm == 32 ? 0 : Imm), MVT::i32);
+  return CurDAG->getTargetConstant((Imm == 32 ? 0 : Imm), SDLoc(N), MVT::i32);
 }]>;
 def ThumbSRImmAsmOperand: AsmOperandClass { let Name = "ImmThumbSR"; }
 def imm_sr : Operand<i32>, PatLeaf<(imm), [{
@@ -32,11 +32,9 @@ def imm_sr : Operand<i32>, PatLeaf<(imm), [{
   let ParserMatchClass = ThumbSRImmAsmOperand;
 }
 
-def imm_neg_XFORM : SDNodeXForm<imm, [{
-  return CurDAG->getTargetConstant(-(int)N->getZExtValue(), MVT::i32);
-}]>;
 def imm_comp_XFORM : SDNodeXForm<imm, [{
-  return CurDAG->getTargetConstant(~((uint32_t)N->getZExtValue()), MVT::i32);
+  return CurDAG->getTargetConstant(~((uint32_t)N->getZExtValue()), SDLoc(N),
+                                   MVT::i32);
 }]>;
 
 def imm0_7_neg : PatLeaf<(i32 imm), [{
@@ -64,33 +62,65 @@ def thumb_immshifted : PatLeaf<(imm), [{
 
 def thumb_immshifted_val : SDNodeXForm<imm, [{
   unsigned V = ARM_AM::getThumbImmNonShiftedVal((unsigned)N->getZExtValue());
-  return CurDAG->getTargetConstant(V, MVT::i32);
+  return CurDAG->getTargetConstant(V, SDLoc(N), MVT::i32);
 }]>;
 
 def thumb_immshifted_shamt : SDNodeXForm<imm, [{
   unsigned V = ARM_AM::getThumbImmValShift((unsigned)N->getZExtValue());
-  return CurDAG->getTargetConstant(V, MVT::i32);
+  return CurDAG->getTargetConstant(V, SDLoc(N), MVT::i32);
 }]>;
 
-// ADR instruction labels.
-def t_adrlabel : Operand<i32> {
-  let EncoderMethod = "getThumbAdrLabelOpValue";
+// Scaled 4 immediate.
+def t_imm0_1020s4_asmoperand: AsmOperandClass { let Name = "Imm0_1020s4"; }
+def t_imm0_1020s4 : Operand<i32> {
+  let PrintMethod = "printThumbS4ImmOperand";
+  let ParserMatchClass = t_imm0_1020s4_asmoperand;
+  let OperandType = "OPERAND_IMMEDIATE";
 }
 
-// Scaled 4 immediate.
-def t_imm_s4 : Operand<i32> {
+def t_imm0_508s4_asmoperand: AsmOperandClass { let Name = "Imm0_508s4"; }
+def t_imm0_508s4 : Operand<i32> {
   let PrintMethod = "printThumbS4ImmOperand";
+  let ParserMatchClass = t_imm0_508s4_asmoperand;
+  let OperandType = "OPERAND_IMMEDIATE";
+}
+// Alias use only, so no printer is necessary.
+def t_imm0_508s4_neg_asmoperand: AsmOperandClass { let Name = "Imm0_508s4Neg"; }
+def t_imm0_508s4_neg : Operand<i32> {
+  let ParserMatchClass = t_imm0_508s4_neg_asmoperand;
   let OperandType = "OPERAND_IMMEDIATE";
 }
 
 // Define Thumb specific addressing modes.
 
+// unsigned 8-bit, 2-scaled memory offset
+class OperandUnsignedOffset_b8s2 : AsmOperandClass {
+  let Name = "UnsignedOffset_b8s2";
+  let PredicateMethod = "isUnsignedOffset<8, 2>";
+}
+
+def UnsignedOffset_b8s2 : OperandUnsignedOffset_b8s2;
+
+// thumb style PC relative operand. signed, 8 bits magnitude,
+// two bits shift. can be represented as either [pc, #imm], #imm,
+// or relocatable expression...
+def ThumbMemPC : AsmOperandClass {
+  let Name = "ThumbMemPC";
+}
+
 let OperandType = "OPERAND_PCREL" in {
 def t_brtarget : Operand<OtherVT> {
   let EncoderMethod = "getThumbBRTargetOpValue";
   let DecoderMethod = "DecodeThumbBROperand";
 }
 
+// ADR instruction labels.
+def t_adrlabel : Operand<i32> {
+  let EncoderMethod = "getThumbAdrLabelOpValue";
+  let PrintMethod = "printAdrLabelOperand<2>";
+  let ParserMatchClass = UnsignedOffset_b8s2;
+}
+
 def t_bcctarget : Operand<i32> {
   let EncoderMethod = "getThumbBCCTargetOpValue";
   let DecoderMethod = "DecodeThumbBCCTargetOperand";
@@ -110,12 +140,21 @@ def t_blxtarget : Operand<i32> {
   let EncoderMethod = "getThumbBLXTargetOpValue";
   let DecoderMethod = "DecodeThumbBLXOffset";
 }
+
+// t_addrmode_pc := <label> => pc + imm8 * 4
+//
+def t_addrmode_pc : MemOperand {
+  let EncoderMethod = "getAddrModePCOpValue";
+  let DecoderMethod = "DecodeThumbAddrModePC";
+  let PrintMethod = "printThumbLdrLabelOperand";
+  let ParserMatchClass = ThumbMemPC;
+}
 }
 
 // t_addrmode_rr := reg + reg
 //
 def t_addrmode_rr_asm_operand : AsmOperandClass { let Name = "MemThumbRR"; }
-def t_addrmode_rr : Operand<i32>,
+def t_addrmode_rr : MemOperand,
                     ComplexPattern<i32, 2, "SelectThumbAddrModeRR", []> {
   let EncoderMethod = "getThumbAddrModeRegRegOpValue";
   let PrintMethod = "printThumbAddrModeRROperand";
@@ -131,7 +170,7 @@ def t_addrmode_rr : Operand<i32>,
 // the reg+imm forms will match instead. This is a horrible way to do that,
 // as it forces tight coupling between the methods, but it's how selectiondag
 // currently works.
-def t_addrmode_rrs1 : Operand<i32>,
+def t_addrmode_rrs1 : MemOperand,
                       ComplexPattern<i32, 2, "SelectThumbAddrModeRI5S1", []> {
   let EncoderMethod = "getThumbAddrModeRegRegOpValue";
   let PrintMethod = "printThumbAddrModeRROperand";
@@ -139,7 +178,7 @@ def t_addrmode_rrs1 : Operand<i32>,
   let ParserMatchClass = t_addrmode_rr_asm_operand;
   let MIOperandInfo = (ops tGPR:$base, tGPR:$offsreg);
 }
-def t_addrmode_rrs2 : Operand<i32>,
+def t_addrmode_rrs2 : MemOperand,
                       ComplexPattern<i32, 2, "SelectThumbAddrModeRI5S2", []> {
   let EncoderMethod = "getThumbAddrModeRegRegOpValue";
   let DecoderMethod = "DecodeThumbAddrModeRR";
@@ -147,7 +186,7 @@ def t_addrmode_rrs2 : Operand<i32>,
   let ParserMatchClass = t_addrmode_rr_asm_operand;
   let MIOperandInfo = (ops tGPR:$base, tGPR:$offsreg);
 }
-def t_addrmode_rrs4 : Operand<i32>,
+def t_addrmode_rrs4 : MemOperand,
                       ComplexPattern<i32, 2, "SelectThumbAddrModeRI5S4", []> {
   let EncoderMethod = "getThumbAddrModeRegRegOpValue";
   let DecoderMethod = "DecodeThumbAddrModeRR";
@@ -159,7 +198,7 @@ def t_addrmode_rrs4 : Operand<i32>,
 // t_addrmode_is4 := reg + imm5 * 4
 //
 def t_addrmode_is4_asm_operand : AsmOperandClass { let Name = "MemThumbRIs4"; }
-def t_addrmode_is4 : Operand<i32>,
+def t_addrmode_is4 : MemOperand,
                      ComplexPattern<i32, 2, "SelectThumbAddrModeImm5S4", []> {
   let EncoderMethod = "getAddrModeISOpValue";
   let DecoderMethod = "DecodeThumbAddrModeIS";
@@ -171,7 +210,7 @@ def t_addrmode_is4 : Operand<i32>,
 // t_addrmode_is2 := reg + imm5 * 2
 //
 def t_addrmode_is2_asm_operand : AsmOperandClass { let Name = "MemThumbRIs2"; }
-def t_addrmode_is2 : Operand<i32>,
+def t_addrmode_is2 : MemOperand,
                      ComplexPattern<i32, 2, "SelectThumbAddrModeImm5S2", []> {
   let EncoderMethod = "getAddrModeISOpValue";
   let DecoderMethod = "DecodeThumbAddrModeIS";
@@ -183,7 +222,7 @@ def t_addrmode_is2 : Operand<i32>,
 // t_addrmode_is1 := reg + imm5
 //
 def t_addrmode_is1_asm_operand : AsmOperandClass { let Name = "MemThumbRIs1"; }
-def t_addrmode_is1 : Operand<i32>,
+def t_addrmode_is1 : MemOperand,
                      ComplexPattern<i32, 2, "SelectThumbAddrModeImm5S1", []> {
   let EncoderMethod = "getAddrModeISOpValue";
   let DecoderMethod = "DecodeThumbAddrModeIS";
@@ -194,8 +233,10 @@ def t_addrmode_is1 : Operand<i32>,
 
 // t_addrmode_sp := sp + imm8 * 4
 //
+// FIXME: This really shouldn't have an explicit SP operand at all. It should
+// be implicit, just like in the instruction encoding itself.
 def t_addrmode_sp_asm_operand : AsmOperandClass { let Name = "MemThumbSPI"; }
-def t_addrmode_sp : Operand<i32>,
+def t_addrmode_sp : MemOperand,
                     ComplexPattern<i32, 2, "SelectThumbAddrModeSP", []> {
   let EncoderMethod = "getAddrModeThumbSPOpValue";
   let DecoderMethod = "DecodeThumbAddrModeSP";
@@ -204,13 +245,6 @@ def t_addrmode_sp : Operand<i32>,
   let MIOperandInfo = (ops GPR:$base, i32imm:$offsimm);
 }
 
-// t_addrmode_pc := <label> => pc + imm8 * 4
-//
-def t_addrmode_pc : Operand<i32> {
-  let EncoderMethod = "getAddrModePCOpValue";
-  let DecoderMethod = "DecodeThumbAddrModePC";
-}
-
 //===----------------------------------------------------------------------===//
 //  Miscellaneous Instructions.
 //
@@ -236,20 +270,26 @@ class T1SystemEncoding<bits<8> opc>
   let Inst{7-0} = opc;
 }
 
-def tNOP : T1pI<(outs), (ins), NoItinerary, "nop", "", []>,
-           T1SystemEncoding<0x00>; // A8.6.110
-
-def tYIELD : T1pI<(outs), (ins), NoItinerary, "yield", "", []>,
-           T1SystemEncoding<0x10>; // A8.6.410
-
-def tWFE : T1pI<(outs), (ins), NoItinerary, "wfe", "", []>,
-           T1SystemEncoding<0x20>; // A8.6.408
+def tHINT : T1pI<(outs), (ins imm0_15:$imm), NoItinerary, "hint", "\t$imm",
+                 [(int_arm_hint imm0_15:$imm)]>,
+            T1SystemEncoding<0x00>,
+            Requires<[IsThumb, HasV6M]> {
+  bits<4> imm;
+  let Inst{7-4} = imm;
+}
 
-def tWFI : T1pI<(outs), (ins), NoItinerary, "wfi", "", []>,
-           T1SystemEncoding<0x30>; // A8.6.409
+class tHintAlias<string Asm, dag Result> : tInstAlias<Asm, Result> {
+  let Predicates = [IsThumb, HasV6M];
+}
 
-def tSEV : T1pI<(outs), (ins), NoItinerary, "sev", "", []>,
-           T1SystemEncoding<0x40>; // A8.6.157
+def : tHintAlias<"nop$p", (tHINT 0, pred:$p)>; // A8.6.110
+def : tHintAlias<"yield$p", (tHINT 1, pred:$p)>; // A8.6.410
+def : tHintAlias<"wfe$p", (tHINT 2, pred:$p)>; // A8.6.408
+def : tHintAlias<"wfi$p", (tHINT 3, pred:$p)>; // A8.6.409
+def : tHintAlias<"sev$p", (tHINT 4, pred:$p)>; // A8.6.157
+def : tInstAlias<"sevl$p", (tHINT 5, pred:$p)> {
+  let Predicates = [IsThumb2, HasV8];
+}
 
 // The imm operand $val can be used by a debugger to store more information
 // about the breakpoint.
@@ -261,9 +301,18 @@ def tBKPT : T1I<(outs), (ins imm0_255:$val), NoItinerary, "bkpt\t$val",
   bits<8> val;
   let Inst{7-0} = val;
 }
+// default immediate for breakpoint mnemonic
+def : InstAlias<"bkpt", (tBKPT 0)>, Requires<[IsThumb]>;
+
+def tHLT : T1I<(outs), (ins imm0_63:$val), NoItinerary, "hlt\t$val",
+                []>, T1Encoding<0b101110>, Requires<[IsThumb, HasV8]> {
+  let Inst{9-6} = 0b1010;
+  bits<6> val;
+  let Inst{5-0} = val;
+}
 
 def tSETEND : T1I<(outs), (ins setend_op:$end), NoItinerary, "setend\t$end",
-                  []>, T1Encoding<0b101101> {
+                  []>, T1Encoding<0b101101>, Requires<[IsNotMClass]>, Deprecated<HasV8Ops> {
   bits<1> end;
   // A8.6.156
   let Inst{9-5} = 0b10010;
@@ -274,8 +323,7 @@ def tSETEND : T1I<(outs), (ins setend_op:$end), NoItinerary, "setend\t$end",
 
 // Change Processor State is a system instruction -- for disassembly only.
 def tCPS : T1I<(outs), (ins imod_op:$imod, iflags_op:$iflags),
-                NoItinerary, "cps$imod $iflags",
-                [/* For disassembly only; pattern left blank */]>,
+                NoItinerary, "cps$imod $iflags", []>,
            T1Misc<0b0110011> {
   // A8.6.38 & B6.1.1
   bit imod;
@@ -291,7 +339,7 @@ def tCPS : T1I<(outs), (ins imod_op:$imod, iflags_op:$iflags),
 let isNotDuplicable = 1, isCodeGenOnly = 1 in
 def tPICADD : TIt<(outs GPR:$dst), (ins GPR:$lhs, pclabel:$cp), IIC_iALUr, "",
                   [(set GPR:$dst, (ARMpic_add GPR:$lhs, imm:$cp))]>,
-              T1Special<{0,0,?,?}> {
+              T1Special<{0,0,?,?}>, Sched<[WriteALU]> {
   // A8.6.6
   bits<3> dst;
   let Inst{6-3} = 0b1111; // Rm = pc
@@ -299,61 +347,80 @@ def tPICADD : TIt<(outs GPR:$dst), (ins GPR:$lhs, pclabel:$cp), IIC_iALUr, "",
 }
 
 // ADD <Rd>, sp, #<imm8>
-// This is rematerializable, which is particularly useful for taking the
-// address of locals.
-let isReMaterializable = 1 in
-def tADDrSPi : T1I<(outs tGPR:$dst), (ins GPR:$sp, t_imm_s4:$rhs), IIC_iALUi,
-                   "add\t$dst, $sp, $rhs", []>,
-               T1Encoding<{1,0,1,0,1,?}> {
+// FIXME: This should not be marked as having side effects, and it should be
+// rematerializable. Clearing the side effect bit causes miscompilations,
+// probably because the instruction can be moved around.
+def tADDrSPi : T1pI<(outs tGPR:$dst), (ins GPRsp:$sp, t_imm0_1020s4:$imm),
+                    IIC_iALUi, "add", "\t$dst, $sp, $imm", []>,
+               T1Encoding<{1,0,1,0,1,?}>, Sched<[WriteALU]> {
   // A6.2 & A8.6.8
   bits<3> dst;
-  bits<8> rhs;
+  bits<8> imm;
   let Inst{10-8} = dst;
-  let Inst{7-0}  = rhs;
+  let Inst{7-0}  = imm;
   let DecoderMethod = "DecodeThumbAddSpecialReg";
 }
 
+// Thumb1 frame lowering is rather fragile, we hope to be able to use
+// tADDrSPi, but we may need to insert a sequence that clobbers CPSR.
+def tADDframe : PseudoInst<(outs tGPR:$dst), (ins i32imm:$base, i32imm:$offset),
+                           NoItinerary, []>,
+                Requires<[IsThumb, IsThumb1Only]> {
+  let Defs = [CPSR];
+}
+
 // ADD sp, sp, #<imm7>
-def tADDspi : TIt<(outs GPR:$dst), (ins GPR:$lhs, t_imm_s4:$rhs), IIC_iALUi,
-                  "add\t$dst, $rhs", []>,
-              T1Misc<{0,0,0,0,0,?,?}> {
+def tADDspi : T1pIt<(outs GPRsp:$Rdn), (ins GPRsp:$Rn, t_imm0_508s4:$imm),
+                     IIC_iALUi, "add", "\t$Rdn, $imm", []>,
+              T1Misc<{0,0,0,0,0,?,?}>, Sched<[WriteALU]> {
   // A6.2.5 & A8.6.8
-  bits<7> rhs;
-  let Inst{6-0} = rhs;
+  bits<7> imm;
+  let Inst{6-0} = imm;
   let DecoderMethod = "DecodeThumbAddSPImm";
 }
 
 // SUB sp, sp, #<imm7>
 // FIXME: The encoding and the ASM string don't match up.
-def tSUBspi : TIt<(outs GPR:$dst), (ins GPR:$lhs, t_imm_s4:$rhs), IIC_iALUi,
-                  "sub\t$dst, $rhs", []>,
-              T1Misc<{0,0,0,0,1,?,?}> {
+def tSUBspi : T1pIt<(outs GPRsp:$Rdn), (ins GPRsp:$Rn, t_imm0_508s4:$imm),
+                    IIC_iALUi, "sub", "\t$Rdn, $imm", []>,
+              T1Misc<{0,0,0,0,1,?,?}>, Sched<[WriteALU]> {
   // A6.2.5 & A8.6.214
-  bits<7> rhs;
-  let Inst{6-0} = rhs;
+  bits<7> imm;
+  let Inst{6-0} = imm;
   let DecoderMethod = "DecodeThumbAddSPImm";
 }
 
+def : tInstAlias<"add${p} sp, $imm",
+                 (tSUBspi SP, t_imm0_508s4_neg:$imm, pred:$p)>;
+def : tInstAlias<"add${p} sp, sp, $imm",
+                 (tSUBspi SP, t_imm0_508s4_neg:$imm, pred:$p)>;
+
+// Can optionally specify SP as a three operand instruction.
+def : tInstAlias<"add${p} sp, sp, $imm",
+                 (tADDspi SP, t_imm0_508s4:$imm, pred:$p)>;
+def : tInstAlias<"sub${p} sp, sp, $imm",
+                 (tSUBspi SP, t_imm0_508s4:$imm, pred:$p)>;
+
 // ADD <Rm>, sp
-def tADDrSP : TIt<(outs GPR:$dst), (ins GPR:$lhs, GPR:$rhs), IIC_iALUr,
-                  "add\t$dst, $rhs", []>,
-              T1Special<{0,0,?,?}> {
+def tADDrSP : T1pI<(outs GPR:$Rdn), (ins GPRsp:$sp, GPR:$Rn), IIC_iALUr,
+                   "add", "\t$Rdn, $sp, $Rn", []>,
+              T1Special<{0,0,?,?}>, Sched<[WriteALU]> {
   // A8.6.9 Encoding T1
-  bits<4> dst;
-  let Inst{7}   = dst{3};
+  bits<4> Rdn;
+  let Inst{7}   = Rdn{3};
   let Inst{6-3} = 0b1101;
-  let Inst{2-0} = dst{2-0};
+  let Inst{2-0} = Rdn{2-0};
   let DecoderMethod = "DecodeThumbAddSPReg";
 }
 
 // ADD sp, <Rm>
-def tADDspr : TIt<(outs GPR:$dst), (ins GPR:$lhs, GPR:$rhs), IIC_iALUr,
-                  "add\t$dst, $rhs", []>,
-              T1Special<{0,0,?,?}> {
+def tADDspr : T1pIt<(outs GPRsp:$Rdn), (ins GPRsp:$Rn, GPR:$Rm), IIC_iALUr,
+                  "add", "\t$Rdn, $Rm", []>,
+              T1Special<{0,0,?,?}>, Sched<[WriteALU]> {
   // A8.6.9 Encoding T2
-  bits<4> dst;
+  bits<4> Rm;
   let Inst{7} = 1;
-  let Inst{6-3} = dst;
+  let Inst{6-3} = Rm;
   let Inst{2-0} = 0b101;
   let DecoderMethod = "DecodeThumbAddSPReg";
 }
@@ -365,128 +432,100 @@ def tADDspr : TIt<(outs GPR:$dst), (ins GPR:$lhs, GPR:$rhs), IIC_iALUr,
 // Indirect branches
 let isBranch = 1, isTerminator = 1, isBarrier = 1, isIndirectBranch = 1 in {
   def tBX : TI<(outs), (ins GPR:$Rm, pred:$p), IIC_Br, "bx${p}\t$Rm", []>,
-            T1Special<{1,1,0,?}> {
+            T1Special<{1,1,0,?}>, Sched<[WriteBr]> {
     // A6.2.3 & A8.6.25
     bits<4> Rm;
     let Inst{6-3} = Rm;
     let Inst{2-0} = 0b000;
+    let Unpredictable{2-0} = 0b111;
   }
 }
 
 let isReturn = 1, isTerminator = 1, isBarrier = 1 in {
   def tBX_RET : tPseudoExpand<(outs), (ins pred:$p), 2, IIC_Br,
-                   [(ARMretflag)], (tBX LR, pred:$p)>;
+                   [(ARMretflag)], (tBX LR, pred:$p)>, Sched<[WriteBr]>;
 
   // Alternative return instruction used by vararg functions.
   def tBX_RET_vararg : tPseudoExpand<(outs), (ins tGPR:$Rm, pred:$p),
                    2, IIC_Br, [],
-                   (tBX GPR:$Rm, pred:$p)>;
+                   (tBX GPR:$Rm, pred:$p)>, Sched<[WriteBr]>;
 }
 
 // All calls clobber the non-callee saved registers. SP is marked as a use to
 // prevent stack-pointer assignments that appear immediately before calls from
 // potentially appearing dead.
 let isCall = 1,
-  // On non-Darwin platforms R9 is callee-saved.
-  Defs = [R0,  R1,  R2,  R3,  R12, LR, QQQQ0, QQQQ2, QQQQ3, CPSR, FPSCR],
-  Uses = [SP] in {
+  Defs = [LR], Uses = [SP] in {
   // Also used for Thumb2
   def tBL  : TIx2<0b11110, 0b11, 1,
-                  (outs), (ins pred:$p, t_bltarget:$func, variable_ops), IIC_Br,
+                  (outs), (ins pred:$p, t_bltarget:$func), IIC_Br,
                   "bl${p}\t$func",
                   [(ARMtcall tglobaladdr:$func)]>,
-             Requires<[IsThumb, IsNotDarwin]> {
-    bits<22> func;
-    let Inst{26} = func{21};
+             Requires<[IsThumb]>, Sched<[WriteBrL]> {
+    bits<24> func;
+    let Inst{26} = func{23};
     let Inst{25-16} = func{20-11};
-    let Inst{13} = 1;
-    let Inst{11} = 1;
+    let Inst{13} = func{22};
+    let Inst{11} = func{21};
     let Inst{10-0} = func{10-0};
   }
 
   // ARMv5T and above, also used for Thumb2
   def tBLXi : TIx2<0b11110, 0b11, 0,
-                 (outs), (ins pred:$p, t_blxtarget:$func, variable_ops), IIC_Br,
+                 (outs), (ins pred:$p, t_blxtarget:$func), IIC_Br,
                    "blx${p}\t$func",
                    [(ARMcall tglobaladdr:$func)]>,
-              Requires<[IsThumb, HasV5T, IsNotDarwin]> {
-    bits<21> func;
+              Requires<[IsThumb, HasV5T, IsNotMClass]>, Sched<[WriteBrL]> {
+    bits<24> func;
+    let Inst{26} = func{23};
     let Inst{25-16} = func{20-11};
-    let Inst{13} = 1;
-    let Inst{11} = 1;
+    let Inst{13} = func{22};
+    let Inst{11} = func{21};
     let Inst{10-1} = func{10-1};
     let Inst{0} = 0; // func{0} is assumed zero
   }
 
   // Also used for Thumb2
-  def tBLXr : TI<(outs), (ins pred:$p, GPR:$func, variable_ops), IIC_Br,
+  def tBLXr : TI<(outs), (ins pred:$p, GPR:$func), IIC_Br,
                   "blx${p}\t$func",
                   [(ARMtcall GPR:$func)]>,
-              Requires<[IsThumb, HasV5T, IsNotDarwin]>,
-              T1Special<{1,1,1,?}> { // A6.2.3 & A8.6.24;
+              Requires<[IsThumb, HasV5T]>,
+              T1Special<{1,1,1,?}>, Sched<[WriteBrL]> { // A6.2.3 & A8.6.24;
     bits<4> func;
     let Inst{6-3} = func;
     let Inst{2-0} = 0b000;
   }
 
   // ARMv4T
-  def tBX_CALL : tPseudoInst<(outs), (ins tGPR:$func, variable_ops),
+  def tBX_CALL : tPseudoInst<(outs), (ins tGPR:$func),
                   4, IIC_Br,
                   [(ARMcall_nolink tGPR:$func)]>,
-            Requires<[IsThumb, IsThumb1Only, IsNotDarwin]>;
-}
-
-let isCall = 1,
-  // On Darwin R9 is call-clobbered.
-  // R7 is marked as a use to prevent frame-pointer assignments from being
-  // moved above / below calls.
-  Defs = [R0,  R1,  R2,  R3,  R9,  R12, LR, QQQQ0, QQQQ2, QQQQ3, CPSR, FPSCR],
-  Uses = [R7, SP] in {
-  // Also used for Thumb2
-  def tBLr9 : tPseudoExpand<(outs), (ins pred:$p, t_bltarget:$func, variable_ops),
-                          4, IIC_Br, [(ARMtcall tglobaladdr:$func)],
-                          (tBL pred:$p, t_bltarget:$func)>,
-              Requires<[IsThumb, IsDarwin]>;
-
-  // ARMv5T and above, also used for Thumb2
-  def tBLXi_r9 : tPseudoExpand<(outs), (ins pred:$p, t_blxtarget:$func, variable_ops),
-                      4, IIC_Br, [(ARMcall tglobaladdr:$func)],
-                      (tBLXi pred:$p, t_blxtarget:$func)>,
-                 Requires<[IsThumb, HasV5T, IsDarwin]>;
-
-  // Also used for Thumb2
-  def tBLXr_r9 : tPseudoExpand<(outs), (ins pred:$p, GPR:$func, variable_ops),
-                    2, IIC_Br, [(ARMtcall GPR:$func)],
-                    (tBLXr pred:$p, GPR:$func)>,
-                 Requires<[IsThumb, HasV5T, IsDarwin]>;
-
-  // ARMv4T
-  def tBXr9_CALL : tPseudoInst<(outs), (ins tGPR:$func, variable_ops),
-                   4, IIC_Br,
-                   [(ARMcall_nolink tGPR:$func)]>,
-              Requires<[IsThumb, IsThumb1Only, IsDarwin]>;
+            Requires<[IsThumb, IsThumb1Only]>, Sched<[WriteBr]>;
 }
 
 let isBranch = 1, isTerminator = 1, isBarrier = 1 in {
   let isPredicable = 1 in
-  def tB   : T1I<(outs), (ins t_brtarget:$target), IIC_Br,
-                 "b\t$target", [(br bb:$target)]>,
-             T1Encoding<{1,1,1,0,0,?}> {
+  def tB   : T1pI<(outs), (ins t_brtarget:$target), IIC_Br,
+                 "b", "\t$target", [(br bb:$target)]>,
+             T1Encoding<{1,1,1,0,0,?}>, Sched<[WriteBr]> {
     bits<11> target;
     let Inst{10-0} = target;
-  }
+    let AsmMatchConverter = "cvtThumbBranches";
+ }
 
   // Far jump
   // Just a pseudo for a tBL instruction. Needed to let regalloc know about
   // the clobber of LR.
   let Defs = [LR] in
   def tBfar : tPseudoExpand<(outs), (ins t_bltarget:$target, pred:$p),
-                          4, IIC_Br, [], (tBL pred:$p, t_bltarget:$target)>;
+                          4, IIC_Br, [], (tBL pred:$p, t_bltarget:$target)>,
+                          Sched<[WriteBrTbl]>;
 
   def tBR_JTr : tPseudoInst<(outs),
                       (ins tGPR:$target, i32imm:$jt, i32imm:$id),
                       0, IIC_Br,
-                      [(ARMbrjt tGPR:$target, tjumptable:$jt, imm:$id)]> {
+                      [(ARMbrjt tGPR:$target, tjumptable:$jt, imm:$id)]>,
+                      Sched<[WriteBrTbl]> {
     list<Predicate> Predicates = [IsThumb, IsThumb1Only];
   }
 }
@@ -497,46 +536,43 @@ let isBranch = 1, isTerminator = 1 in
   def tBcc : T1I<(outs), (ins t_bcctarget:$target, pred:$p), IIC_Br,
                  "b${p}\t$target",
                  [/*(ARMbrcond bb:$target, imm:$cc)*/]>,
-             T1BranchCond<{1,1,0,1}> {
+             T1BranchCond<{1,1,0,1}>, Sched<[WriteBr]> {
   bits<4> p;
   bits<8> target;
   let Inst{11-8} = p;
   let Inst{7-0} = target;
+  let AsmMatchConverter = "cvtThumbBranches";
 }
 
+
 // Tail calls
 let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1 in {
-  // Darwin versions.
-  let Defs = [R0, R1, R2, R3, R9, R12, QQQQ0, QQQQ2, QQQQ3, PC],
-      Uses = [SP] in {
-    // tTAILJMPd: Darwin version uses a Thumb2 branch (no Thumb1 tail calls
-    // on Darwin), so it's in ARMInstrThumb2.td.
-    def tTAILJMPr : tPseudoExpand<(outs), (ins tcGPR:$dst, variable_ops),
+  // IOS versions.
+  let Uses = [SP] in {
+    def tTAILJMPr : tPseudoExpand<(outs), (ins tcGPR:$dst),
                      4, IIC_Br, [],
                      (tBX GPR:$dst, (ops 14, zero_reg))>,
-                     Requires<[IsThumb, IsDarwin]>;
+                     Requires<[IsThumb]>, Sched<[WriteBr]>;
   }
-  // Non-Darwin versions (the difference is R9).
-  let Defs = [R0, R1, R2, R3, R12, QQQQ0, QQQQ2, QQQQ3, PC],
-      Uses = [SP] in {
-    def tTAILJMPdND : tPseudoExpand<(outs), (ins t_brtarget:$dst, variable_ops),
+  // tTAILJMPd: MachO version uses a Thumb2 branch (no Thumb1 tail calls
+  // on MachO), so it's in ARMInstrThumb2.td.
+  // Non-MachO version:
+  let Uses = [SP] in {
+    def tTAILJMPdND : tPseudoExpand<(outs),
+                   (ins t_brtarget:$dst, pred:$p),
                    4, IIC_Br, [],
-                   (tB t_brtarget:$dst)>,
-                 Requires<[IsThumb, IsNotDarwin]>;
-    def tTAILJMPrND : tPseudoExpand<(outs), (ins tcGPR:$dst, variable_ops),
-                     4, IIC_Br, [],
-                     (tBX GPR:$dst, (ops 14, zero_reg))>,
-                     Requires<[IsThumb, IsNotDarwin]>;
+                   (tB t_brtarget:$dst, pred:$p)>,
+                 Requires<[IsThumb, IsNotMachO]>, Sched<[WriteBr]>;
   }
 }
 
 
-// A8.6.218 Supervisor Call (Software Interrupt) -- for disassembly only
+// A8.6.218 Supervisor Call (Software Interrupt)
 // A8.6.16 B: Encoding T1
 // If Inst{11-8} == 0b1111 then SEE SVC
 let isCall = 1, Uses = [SP] in
 def tSVC : T1pI<(outs), (ins imm0_255:$imm), IIC_Br,
-                "svc", "\t$imm", []>, Encoding16 {
+                "svc", "\t$imm", []>, Encoding16, Sched<[WriteBr]> {
   bits<8> imm;
   let Inst{15-12} = 0b1101;
   let Inst{11-8}  = 0b1111;
@@ -546,7 +582,7 @@ def tSVC : T1pI<(outs), (ins imm0_255:$imm), IIC_Br,
 // The assembler uses 0xDEFE for a trap instruction.
 let isBarrier = 1, isTerminator = 1 in
 def tTRAP : TI<(outs), (ins), IIC_Br,
-               "trap", [(trap)]>, Encoding16 {
+               "trap", [(trap)]>, Encoding16, Sched<[WriteBr]> {
   let Inst = 0xdefe;
 }
 
@@ -633,11 +669,9 @@ def tLDRspi : T1pIs<(outs tGPR:$Rt), (ins t_addrmode_sp:$addr), IIC_iLoad_i,
   let Inst{7-0} = addr;
 }
 
-// Load tconstpool
-// FIXME: Use ldr.n to work around a Darwin assembler bug.
-let canFoldAsLoad = 1, isReMaterializable = 1, isCodeGenOnly = 1 in
+let canFoldAsLoad = 1, isReMaterializable = 1 in
 def tLDRpci : T1pIs<(outs tGPR:$Rt), (ins t_addrmode_pc:$addr), IIC_iLoad_i,
-                  "ldr", ".n\t$Rt, $addr",
+                  "ldr", "\t$Rt, $addr",
                   [(set tGPR:$Rt, (load (ARMWrapper tconstpool:$addr)))]>,
               T1Encoding<{0,1,0,0,1,?}> {
   // A6.2 & A8.6.59
@@ -647,19 +681,6 @@ def tLDRpci : T1pIs<(outs tGPR:$Rt), (ins t_addrmode_pc:$addr), IIC_iLoad_i,
   let Inst{7-0}  = addr;
 }
 
-// FIXME: Remove this entry when the above ldr.n workaround is fixed.
-// For disassembly use only.
-def tLDRpciDIS : T1pIs<(outs tGPR:$Rt), (ins t_addrmode_pc:$addr), IIC_iLoad_i,
-                       "ldr", "\t$Rt, $addr",
-                       [/* disassembly only */]>,
-                 T1Encoding<{0,1,0,0,1,?}> {
-  // A6.2 & A8.6.59
-  bits<3> Rt;
-  bits<8> addr;
-  let Inst{10-8} = Rt;
-  let Inst{7-0}  = addr;
-}
-
 // A8.6.194 & A8.6.192
 defm tSTR  : thumb_st_rr_ri_enc<0b000, 0b0110, t_addrmode_rrs4,
                                 t_addrmode_is4, AddrModeT1_4,
@@ -693,52 +714,53 @@ def tSTRspi : T1pIs<(outs), (ins tGPR:$Rt, t_addrmode_sp:$addr), IIC_iStore_i,
 //  Load / store multiple Instructions.
 //
 
-multiclass thumb_ldst_mult<string asm, InstrItinClass itin,
-                           InstrItinClass itin_upd, bits<6> T1Enc,
-                           bit L_bit, string baseOpc> {
-  def IA :
-    T1I<(outs), (ins tGPR:$Rn, pred:$p, reglist:$regs, variable_ops),
-        itin, !strconcat(asm, "${p}\t$Rn, $regs"), []>,
-       T1Encoding<T1Enc> {
-    bits<3> Rn;
-    bits<8> regs;
-    let Inst{10-8} = Rn;
-    let Inst{7-0}  = regs;
-  }
-
-  def IA_UPD :
-    InstTemplate<AddrModeNone, 0, IndexModeNone, Pseudo, GenericDomain, 
-                 "$Rn = $wb", itin_upd>,
-    PseudoInstExpansion<(!cast<Instruction>(!strconcat(baseOpc, "IA"))
-                       tGPR:$Rn, pred:$p, reglist:$regs)> {
-    let Size = 2;
-    let OutOperandList = (outs GPR:$wb);
-    let InOperandList = (ins GPR:$Rn, pred:$p, reglist:$regs, variable_ops);
-    let Pattern = [];
-    let isCodeGenOnly = 1;
-    let isPseudo = 1;
-    list<Predicate> Predicates = [IsThumb];
-  }
-}
-
 // These require base address to be written back or one of the loaded regs.
-let neverHasSideEffects = 1 in {
+let hasSideEffects = 0 in {
 
 let mayLoad = 1, hasExtraDefRegAllocReq = 1 in
-defm tLDM : thumb_ldst_mult<"ldm", IIC_iLoad_m, IIC_iLoad_mu,
-                            {1,1,0,0,1,?}, 1, "tLDM">;
-
+def tLDMIA : T1I<(outs), (ins tGPR:$Rn, pred:$p, reglist:$regs, variable_ops),
+        IIC_iLoad_m, "ldm${p}\t$Rn, $regs", []>, T1Encoding<{1,1,0,0,1,?}> {
+  bits<3> Rn;
+  bits<8> regs;
+  let Inst{10-8} = Rn;
+  let Inst{7-0}  = regs;
+}
+
+// Writeback version is just a pseudo, as there's no encoding difference.
+// Writeback happens iff the base register is not in the destination register
+// list.
+def tLDMIA_UPD :
+    InstTemplate<AddrModeNone, 0, IndexModeNone, Pseudo, GenericDomain,
+                 "$Rn = $wb", IIC_iLoad_mu>,
+    PseudoInstExpansion<(tLDMIA tGPR:$Rn, pred:$p, reglist:$regs)> {
+  let Size = 2;
+  let OutOperandList = (outs GPR:$wb);
+  let InOperandList = (ins GPR:$Rn, pred:$p, reglist:$regs, variable_ops);
+  let Pattern = [];
+  let isCodeGenOnly = 1;
+  let isPseudo = 1;
+  list<Predicate> Predicates = [IsThumb];
+}
+
+// There is no non-writeback version of STM for Thumb.
 let mayStore = 1, hasExtraSrcRegAllocReq = 1 in
-defm tSTM : thumb_ldst_mult<"stm", IIC_iStore_m, IIC_iStore_mu,
-                            {1,1,0,0,0,?}, 0, "tSTM">;
+def tSTMIA_UPD : Thumb1I<(outs GPR:$wb),
+                         (ins tGPR:$Rn, pred:$p, reglist:$regs, variable_ops),
+                         AddrModeNone, 2, IIC_iStore_mu,
+                         "stm${p}\t$Rn!, $regs", "$Rn = $wb", []>,
+                     T1Encoding<{1,1,0,0,0,?}> {
+  bits<3> Rn;
+  bits<8> regs;
+  let Inst{10-8} = Rn;
+  let Inst{7-0}  = regs;
+}
 
-} // neverHasSideEffects
+} // hasSideEffects
 
 def : InstAlias<"ldm${p} $Rn!, $regs",
                 (tLDMIA tGPR:$Rn, pred:$p, reglist:$regs)>,
         Requires<[IsThumb, IsThumb1Only]>;
 
-
 let mayLoad = 1, Uses = [SP], Defs = [SP], hasExtraDefRegAllocReq = 1 in
 def tPOP : T1I<(outs), (ins pred:$p, reglist:$regs, variable_ops),
                IIC_iPop,
@@ -839,14 +861,15 @@ let isCommutable = 1, Uses = [CPSR] in
 def tADC :                      // A8.6.2
   T1sItDPEncode<0b0101, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm), IIC_iALUr,
                 "adc", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (adde tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rdn, (adde tGPR:$Rn, tGPR:$Rm))]>, Sched<[WriteALU]>;
 
 // Add immediate
 def tADDi3 :                    // A8.6.4 T1
   T1sIGenEncodeImm<0b01110, (outs tGPR:$Rd), (ins tGPR:$Rm, imm0_7:$imm3),
                    IIC_iALUi,
                    "add", "\t$Rd, $Rm, $imm3",
-                   [(set tGPR:$Rd, (add tGPR:$Rm, imm0_7:$imm3))]> {
+                   [(set tGPR:$Rd, (add tGPR:$Rm, imm0_7:$imm3))]>,
+                   Sched<[WriteALU]> {
   bits<3> imm3;
   let Inst{8-6} = imm3;
 }
@@ -855,7 +878,8 @@ def tADDi8 :                    // A8.6.4 T2
   T1sItGenEncodeImm<{1,1,0,?,?}, (outs tGPR:$Rdn),
                     (ins tGPR:$Rn, imm0_255:$imm8), IIC_iALUi,
                     "add", "\t$Rdn, $imm8",
-                    [(set tGPR:$Rdn, (add tGPR:$Rn, imm8_255:$imm8))]>;
+                    [(set tGPR:$Rdn, (add tGPR:$Rn, imm8_255:$imm8))]>,
+                    Sched<[WriteALU]>;
 
 // Add register
 let isCommutable = 1 in
@@ -863,12 +887,12 @@ def tADDrr :                    // A8.6.6 T1
   T1sIGenEncode<0b01100, (outs tGPR:$Rd), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iALUr,
                 "add", "\t$Rd, $Rn, $Rm",
-                [(set tGPR:$Rd, (add tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rd, (add tGPR:$Rn, tGPR:$Rm))]>, Sched<[WriteALU]>;
 
-let neverHasSideEffects = 1 in
+let hasSideEffects = 0 in
 def tADDhirr : T1pIt<(outs GPR:$Rdn), (ins GPR:$Rn, GPR:$Rm), IIC_iALUr,
                      "add", "\t$Rdn, $Rm", []>,
-               T1Special<{0,0,?,?}> {
+               T1Special<{0,0,?,?}>, Sched<[WriteALU]> {
   // A8.6.6 T2
   bits<4> Rdn;
   bits<4> Rm;
@@ -883,14 +907,15 @@ def tAND :                      // A8.6.12
   T1sItDPEncode<0b0000, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iBITr,
                 "and", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (and tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rdn, (and tGPR:$Rn, tGPR:$Rm))]>, Sched<[WriteALU]>;
 
 // ASR immediate
 def tASRri :                    // A8.6.14
   T1sIGenEncodeImm<{0,1,0,?,?}, (outs tGPR:$Rd), (ins tGPR:$Rm, imm_sr:$imm5),
                    IIC_iMOVsi,
                    "asr", "\t$Rd, $Rm, $imm5",
-                   [(set tGPR:$Rd, (sra tGPR:$Rm, (i32 imm_sr:$imm5)))]> {
+                   [(set tGPR:$Rd, (sra tGPR:$Rm, (i32 imm_sr:$imm5)))]>,
+                   Sched<[WriteALU]> {
   bits<5> imm5;
   let Inst{10-6} = imm5;
 }
@@ -900,14 +925,15 @@ def tASRrr :                    // A8.6.15
   T1sItDPEncode<0b0100, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iMOVsr,
                 "asr", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (sra tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rdn, (sra tGPR:$Rn, tGPR:$Rm))]>, Sched<[WriteALU]>;
 
 // BIC register
 def tBIC :                      // A8.6.20
   T1sItDPEncode<0b1110, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iBITr,
                 "bic", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (and tGPR:$Rn, (not tGPR:$Rm)))]>;
+                [(set tGPR:$Rdn, (and tGPR:$Rn, (not tGPR:$Rm)))]>,
+                Sched<[WriteALU]>;
 
 // CMN register
 let isCompare = 1, Defs = [CPSR] in {
@@ -923,7 +949,7 @@ def tCMNz :                     // A8.6.33
   T1pIDPEncode<0b1011, (outs), (ins tGPR:$Rn, tGPR:$Rm),
                IIC_iCMPr,
                "cmn", "\t$Rn, $Rm",
-               [(ARMcmpZ tGPR:$Rn, (ineg tGPR:$Rm))]>;
+               [(ARMcmpZ tGPR:$Rn, (ineg tGPR:$Rm))]>, Sched<[WriteCMP]>;
 
 } // isCompare = 1, Defs = [CPSR]
 
@@ -932,7 +958,7 @@ let isCompare = 1, Defs = [CPSR] in {
 def tCMPi8 : T1pI<(outs), (ins tGPR:$Rn, imm0_255:$imm8), IIC_iCMPi,
                   "cmp", "\t$Rn, $imm8",
                   [(ARMcmp tGPR:$Rn, imm0_255:$imm8)]>,
-             T1General<{1,0,1,?,?}> {
+             T1General<{1,0,1,?,?}>, Sched<[WriteCMP]> {
   // A8.6.35
   bits<3> Rn;
   bits<8> imm8;
@@ -945,11 +971,11 @@ def tCMPr :                     // A8.6.36 T1
   T1pIDPEncode<0b1010, (outs), (ins tGPR:$Rn, tGPR:$Rm),
                IIC_iCMPr,
                "cmp", "\t$Rn, $Rm",
-               [(ARMcmp tGPR:$Rn, tGPR:$Rm)]>;
+               [(ARMcmp tGPR:$Rn, tGPR:$Rm)]>, Sched<[WriteCMP]>;
 
 def tCMPhir : T1pI<(outs), (ins GPR:$Rn, GPR:$Rm), IIC_iCMPr,
                    "cmp", "\t$Rn, $Rm", []>,
-              T1Special<{0,1,?,?}> {
+              T1Special<{0,1,?,?}>, Sched<[WriteCMP]> {
   // A8.6.36 T2
   bits<4> Rm;
   bits<4> Rn;
@@ -966,14 +992,15 @@ def tEOR :                      // A8.6.45
   T1sItDPEncode<0b0001, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iBITr,
                 "eor", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (xor tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rdn, (xor tGPR:$Rn, tGPR:$Rm))]>, Sched<[WriteALU]>;
 
 // LSL immediate
 def tLSLri :                    // A8.6.88
-  T1sIGenEncodeImm<{0,0,0,?,?}, (outs tGPR:$Rd), (ins tGPR:$Rm, i32imm:$imm5),
+  T1sIGenEncodeImm<{0,0,0,?,?}, (outs tGPR:$Rd), (ins tGPR:$Rm, imm0_31:$imm5),
                    IIC_iMOVsi,
                    "lsl", "\t$Rd, $Rm, $imm5",
-                   [(set tGPR:$Rd, (shl tGPR:$Rm, (i32 imm:$imm5)))]> {
+                   [(set tGPR:$Rd, (shl tGPR:$Rm, (i32 imm:$imm5)))]>,
+                   Sched<[WriteALU]> {
   bits<5> imm5;
   let Inst{10-6} = imm5;
 }
@@ -983,14 +1010,15 @@ def tLSLrr :                    // A8.6.89
   T1sItDPEncode<0b0010, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iMOVsr,
                 "lsl", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (shl tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rdn, (shl tGPR:$Rn, tGPR:$Rm))]>, Sched<[WriteALU]>;
 
 // LSR immediate
 def tLSRri :                    // A8.6.90
   T1sIGenEncodeImm<{0,0,1,?,?}, (outs tGPR:$Rd), (ins tGPR:$Rm, imm_sr:$imm5),
                    IIC_iMOVsi,
                    "lsr", "\t$Rd, $Rm, $imm5",
-                   [(set tGPR:$Rd, (srl tGPR:$Rm, (i32 imm_sr:$imm5)))]> {
+                   [(set tGPR:$Rd, (srl tGPR:$Rm, (i32 imm_sr:$imm5)))]>,
+                   Sched<[WriteALU]> {
   bits<5> imm5;
   let Inst{10-6} = imm5;
 }
@@ -1000,28 +1028,32 @@ def tLSRrr :                    // A8.6.91
   T1sItDPEncode<0b0011, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iMOVsr,
                 "lsr", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (srl tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rdn, (srl tGPR:$Rn, tGPR:$Rm))]>, Sched<[WriteALU]>;
 
 // Move register
 let isMoveImm = 1 in
 def tMOVi8 : T1sI<(outs tGPR:$Rd), (ins imm0_255:$imm8), IIC_iMOVi,
                   "mov", "\t$Rd, $imm8",
                   [(set tGPR:$Rd, imm0_255:$imm8)]>,
-             T1General<{1,0,0,?,?}> {
+             T1General<{1,0,0,?,?}>, Sched<[WriteALU]> {
   // A8.6.96
   bits<3> Rd;
   bits<8> imm8;
   let Inst{10-8} = Rd;
   let Inst{7-0}  = imm8;
 }
+// Because we have an explicit tMOVSr below, we need an alias to handle
+// the immediate "movs" form here. Blech.
+def : tInstAlias <"movs $Rdn, $imm",
+                 (tMOVi8 tGPR:$Rdn, CPSR, imm0_255:$imm, 14, 0)>;
 
 // A7-73: MOV(2) - mov setting flag.
 
-let neverHasSideEffects = 1 in {
+let hasSideEffects = 0 in {
 def tMOVr : Thumb1pI<(outs GPR:$Rd), (ins GPR:$Rm), AddrModeNone,
                       2, IIC_iMOVr,
                       "mov", "\t$Rd, $Rm", "", []>,
-                  T1Special<{1,0,?,?}> {
+                  T1Special<{1,0,?,?}>, Sched<[WriteALU]> {
   // A8.6.97
   bits<4> Rd;
   bits<4> Rm;
@@ -1031,7 +1063,7 @@ def tMOVr : Thumb1pI<(outs GPR:$Rd), (ins GPR:$Rm), AddrModeNone,
 }
 let Defs = [CPSR] in
 def tMOVSr      : T1I<(outs tGPR:$Rd), (ins tGPR:$Rm), IIC_iMOVr,
-                      "movs\t$Rd, $Rm", []>, Encoding16 {
+                      "movs\t$Rd, $Rm", []>, Encoding16, Sched<[WriteALU]> {
   // A8.6.97
   bits<3> Rd;
   bits<3> Rm;
@@ -1039,21 +1071,30 @@ def tMOVSr      : T1I<(outs tGPR:$Rd), (ins tGPR:$Rm), IIC_iMOVr,
   let Inst{5-3}  = Rm;
   let Inst{2-0}  = Rd;
 }
-} // neverHasSideEffects
+} // hasSideEffects
 
 // Multiply register
 let isCommutable = 1 in
 def tMUL :                      // A8.6.105 T1
-  T1sItDPEncode<0b1101, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
-                IIC_iMUL32,
-                "mul", "\t$Rdn, $Rm, $Rdn",
-                [(set tGPR:$Rdn, (mul tGPR:$Rn, tGPR:$Rm))]>;
+  Thumb1sI<(outs tGPR:$Rd), (ins tGPR:$Rn, tGPR:$Rm), AddrModeNone, 2,
+           IIC_iMUL32, "mul", "\t$Rd, $Rn, $Rm", "$Rm = $Rd",
+           [(set tGPR:$Rd, (mul tGPR:$Rn, tGPR:$Rm))]>,
+      T1DataProcessing<0b1101> {
+  bits<3> Rd;
+  bits<3> Rn;
+  let Inst{5-3} = Rn;
+  let Inst{2-0} = Rd;
+  let AsmMatchConverter = "cvtThumbMultiply";
+}
+
+def :tInstAlias<"mul${s}${p} $Rdm, $Rn", (tMUL tGPR:$Rdm, s_cc_out:$s, tGPR:$Rn,
+                                               pred:$p)>;
 
 // Move inverse register
 def tMVN :                      // A8.6.107
   T1sIDPEncode<0b1111, (outs tGPR:$Rd), (ins tGPR:$Rn), IIC_iMVNr,
                "mvn", "\t$Rd, $Rn",
-               [(set tGPR:$Rd, (not tGPR:$Rn))]>;
+               [(set tGPR:$Rd, (not tGPR:$Rn))]>, Sched<[WriteALU]>;
 
 // Bitwise or register
 let isCommutable = 1 in
@@ -1061,7 +1102,7 @@ def tORR :                      // A8.6.114
   T1sItDPEncode<0b1100, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iBITr,
                 "orr", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (or tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rdn, (or tGPR:$Rn, tGPR:$Rm))]>, Sched<[WriteALU]>;
 
 // Swaps
 def tREV :                      // A8.6.134
@@ -1069,35 +1110,36 @@ def tREV :                      // A8.6.134
                  IIC_iUNAr,
                  "rev", "\t$Rd, $Rm",
                  [(set tGPR:$Rd, (bswap tGPR:$Rm))]>,
-                 Requires<[IsThumb, IsThumb1Only, HasV6]>;
+                 Requires<[IsThumb, IsThumb1Only, HasV6]>, Sched<[WriteALU]>;
 
 def tREV16 :                    // A8.6.135
   T1pIMiscEncode<{1,0,1,0,0,1,?}, (outs tGPR:$Rd), (ins tGPR:$Rm),
                  IIC_iUNAr,
                  "rev16", "\t$Rd, $Rm",
              [(set tGPR:$Rd, (rotr (bswap tGPR:$Rm), (i32 16)))]>,
-                Requires<[IsThumb, IsThumb1Only, HasV6]>;
+                Requires<[IsThumb, IsThumb1Only, HasV6]>, Sched<[WriteALU]>;
 
 def tREVSH :                    // A8.6.136
   T1pIMiscEncode<{1,0,1,0,1,1,?}, (outs tGPR:$Rd), (ins tGPR:$Rm),
                  IIC_iUNAr,
                  "revsh", "\t$Rd, $Rm",
                  [(set tGPR:$Rd, (sra (bswap tGPR:$Rm), (i32 16)))]>,
-                 Requires<[IsThumb, IsThumb1Only, HasV6]>;
+                 Requires<[IsThumb, IsThumb1Only, HasV6]>, Sched<[WriteALU]>;
 
 // Rotate right register
 def tROR :                      // A8.6.139
   T1sItDPEncode<0b0111, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iMOVsr,
                 "ror", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (rotr tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rdn, (rotr tGPR:$Rn, tGPR:$Rm))]>,
+                Sched<[WriteALU]>;
 
 // Negate register
 def tRSB :                      // A8.6.141
   T1sIDPEncode<0b1001, (outs tGPR:$Rd), (ins tGPR:$Rn),
                IIC_iALUi,
                "rsb", "\t$Rd, $Rn, #0",
-               [(set tGPR:$Rd, (ineg tGPR:$Rn))]>;
+               [(set tGPR:$Rd, (ineg tGPR:$Rn))]>, Sched<[WriteALU]>;
 
 // Subtract with carry register
 let Uses = [CPSR] in
@@ -1105,32 +1147,34 @@ def tSBC :                      // A8.6.151
   T1sItDPEncode<0b0110, (outs tGPR:$Rdn), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iALUr,
                 "sbc", "\t$Rdn, $Rm",
-                [(set tGPR:$Rdn, (sube tGPR:$Rn, tGPR:$Rm))]>;
+                [(set tGPR:$Rdn, (sube tGPR:$Rn, tGPR:$Rm))]>,
+                Sched<[WriteALU]>;
 
 // Subtract immediate
 def tSUBi3 :                    // A8.6.210 T1
-  T1sIGenEncodeImm<0b01111, (outs tGPR:$Rd), (ins tGPR:$Rm, i32imm:$imm3),
+  T1sIGenEncodeImm<0b01111, (outs tGPR:$Rd), (ins tGPR:$Rm, imm0_7:$imm3),
                    IIC_iALUi,
                    "sub", "\t$Rd, $Rm, $imm3",
-                   [(set tGPR:$Rd, (add tGPR:$Rm, imm0_7_neg:$imm3))]> {
+                   [(set tGPR:$Rd, (add tGPR:$Rm, imm0_7_neg:$imm3))]>,
+                   Sched<[WriteALU]> {
   bits<3> imm3;
   let Inst{8-6} = imm3;
 }
 
 def tSUBi8 :                    // A8.6.210 T2
-  T1sItGenEncodeImm<{1,1,1,?,?}, (outs tGPR:$Rdn), (ins tGPR:$Rn, i32imm:$imm8),
-                    IIC_iALUi,
+  T1sItGenEncodeImm<{1,1,1,?,?}, (outs tGPR:$Rdn),
+                    (ins tGPR:$Rn, imm0_255:$imm8), IIC_iALUi,
                     "sub", "\t$Rdn, $imm8",
-                    [(set tGPR:$Rdn, (add tGPR:$Rn, imm8_255_neg:$imm8))]>;
+                    [(set tGPR:$Rdn, (add tGPR:$Rn, imm8_255_neg:$imm8))]>,
+                    Sched<[WriteALU]>;
 
 // Subtract register
 def tSUBrr :                    // A8.6.212
   T1sIGenEncode<0b01101, (outs tGPR:$Rd), (ins tGPR:$Rn, tGPR:$Rm),
                 IIC_iALUr,
                 "sub", "\t$Rd, $Rn, $Rm",
-                [(set tGPR:$Rd, (sub tGPR:$Rn, tGPR:$Rm))]>;
-
-// TODO: A7-96: STMIA - store multiple.
+                [(set tGPR:$Rd, (sub tGPR:$Rn, tGPR:$Rm))]>,
+                Sched<[WriteALU]>;
 
 // Sign-extend byte
 def tSXTB :                     // A8.6.222
@@ -1138,7 +1182,8 @@ def tSXTB :                     // A8.6.222
                  IIC_iUNAr,
                  "sxtb", "\t$Rd, $Rm",
                  [(set tGPR:$Rd, (sext_inreg tGPR:$Rm, i8))]>,
-                 Requires<[IsThumb, IsThumb1Only, HasV6]>;
+                 Requires<[IsThumb, IsThumb1Only, HasV6]>,
+                 Sched<[WriteALU]>;
 
 // Sign-extend short
 def tSXTH :                     // A8.6.224
@@ -1146,14 +1191,25 @@ def tSXTH :                     // A8.6.224
                  IIC_iUNAr,
                  "sxth", "\t$Rd, $Rm",
                  [(set tGPR:$Rd, (sext_inreg tGPR:$Rm, i16))]>,
-                 Requires<[IsThumb, IsThumb1Only, HasV6]>;
+                 Requires<[IsThumb, IsThumb1Only, HasV6]>,
+                 Sched<[WriteALU]>;
 
 // Test
 let isCompare = 1, isCommutable = 1, Defs = [CPSR] in
 def tTST :                      // A8.6.230
   T1pIDPEncode<0b1000, (outs), (ins tGPR:$Rn, tGPR:$Rm), IIC_iTSTr,
                "tst", "\t$Rn, $Rm",
-               [(ARMcmpZ (and_su tGPR:$Rn, tGPR:$Rm), 0)]>;
+               [(ARMcmpZ (and_su tGPR:$Rn, tGPR:$Rm), 0)]>,
+               Sched<[WriteALU]>;
+
+// A8.8.247  UDF - Undefined (Encoding T1)
+def tUDF : TI<(outs), (ins imm0_255:$imm8), IIC_Br, "udf\t$imm8",
+              [(int_arm_undefined imm0_255:$imm8)]>, Encoding16 {
+  bits<8> imm8;
+  let Inst{15-12} = 0b1101;
+  let Inst{11-8} = 0b1110;
+  let Inst{7-0} = imm8;
+}
 
 // Zero-extend byte
 def tUXTB :                     // A8.6.262
@@ -1161,7 +1217,8 @@ def tUXTB :                     // A8.6.262
                  IIC_iUNAr,
                  "uxtb", "\t$Rd, $Rm",
                  [(set tGPR:$Rd, (and tGPR:$Rm, 0xFF))]>,
-                 Requires<[IsThumb, IsThumb1Only, HasV6]>;
+                 Requires<[IsThumb, IsThumb1Only, HasV6]>,
+                 Sched<[WriteALU]>;
 
 // Zero-extend short
 def tUXTH :                     // A8.6.264
@@ -1169,22 +1226,22 @@ def tUXTH :                     // A8.6.264
                  IIC_iUNAr,
                  "uxth", "\t$Rd, $Rm",
                  [(set tGPR:$Rd, (and tGPR:$Rm, 0xFFFF))]>,
-                 Requires<[IsThumb, IsThumb1Only, HasV6]>;
+                 Requires<[IsThumb, IsThumb1Only, HasV6]>, Sched<[WriteALU]>;
 
 // Conditional move tMOVCCr - Used to implement the Thumb SELECT_CC operation.
 // Expanded after instruction selection into a branch sequence.
 let usesCustomInserter = 1 in  // Expanded after instruction selection.
   def tMOVCCr_pseudo :
-  PseudoInst<(outs tGPR:$dst), (ins tGPR:$false, tGPR:$true, pred:$cc),
-              NoItinerary,
-             [/*(set tGPR:$dst, (ARMcmov tGPR:$false, tGPR:$true, imm:$cc))*/]>;
+  PseudoInst<(outs tGPR:$dst), (ins tGPR:$false, tGPR:$true, cmovpred:$p),
+             NoItinerary,
+             [(set tGPR:$dst, (ARMcmov tGPR:$false, tGPR:$true, cmovpred:$p))]>;
 
 // tLEApcrel - Load a pc-relative address into a register without offending the
 // assembler.
 
 def tADR : T1I<(outs tGPR:$Rd), (ins t_adrlabel:$addr, pred:$p),
                IIC_iALUi, "adr{$p}\t$Rd, $addr", []>,
-               T1Encoding<{1,0,1,0,0,?}> {
+               T1Encoding<{1,0,1,0,0,?}>, Sched<[WriteALU]> {
   bits<3> Rd;
   bits<8> addr;
   let Inst{10-8} = Rd;
@@ -1192,13 +1249,14 @@ def tADR : T1I<(outs tGPR:$Rd), (ins t_adrlabel:$addr, pred:$p),
   let DecoderMethod = "DecodeThumbAddSpecialReg";
 }
 
-let neverHasSideEffects = 1, isReMaterializable = 1 in
+let hasSideEffects = 0, isReMaterializable = 1 in
 def tLEApcrel   : tPseudoInst<(outs tGPR:$Rd), (ins i32imm:$label, pred:$p),
-                              2, IIC_iALUi, []>;
+                              2, IIC_iALUi, []>, Sched<[WriteALU]>;
 
+let hasSideEffects = 1 in
 def tLEApcrelJT : tPseudoInst<(outs tGPR:$Rd),
                               (ins i32imm:$label, nohash_imm:$id, pred:$p),
-                              2, IIC_iALUi, []>;
+                              2, IIC_iALUi, []>, Sched<[WriteALU]>;
 
 //===----------------------------------------------------------------------===//
 // TLS Instructions
@@ -1209,7 +1267,8 @@ def tLEApcrelJT : tPseudoInst<(outs tGPR:$Rd),
 // complete with fixup for the aeabi_read_tp function.
 let isCall = 1, Defs = [R0, R12, LR, CPSR], Uses = [SP] in
 def tTPsoft : tPseudoInst<(outs), (ins), 4, IIC_Br,
-                          [(set R0, ARMthread_pointer)]>;
+                          [(set R0, ARMthread_pointer)]>,
+                          Sched<[WriteBr]>;
 
 //===----------------------------------------------------------------------===//
 // SJLJ Exception handling intrinsics
@@ -1226,19 +1285,20 @@ def tTPsoft : tPseudoInst<(outs), (ins), 4, IIC_Br,
 // preserve all of the callee-saved resgisters, which is exactly what we want.
 // $val is a scratch register for our use.
 let Defs = [ R0,  R1,  R2,  R3,  R4,  R5,  R6,  R7, R12, CPSR ],
-    hasSideEffects = 1, isBarrier = 1, isCodeGenOnly = 1 in
+    hasSideEffects = 1, isBarrier = 1, isCodeGenOnly = 1,
+    usesCustomInserter = 1 in
 def tInt_eh_sjlj_setjmp : ThumbXI<(outs),(ins tGPR:$src, tGPR:$val),
                                   AddrModeNone, 0, NoItinerary, "","",
                           [(set R0, (ARMeh_sjlj_setjmp tGPR:$src, tGPR:$val))]>;
 
-// FIXME: Non-Darwin version(s)
+// FIXME: Non-IOS version(s)
 let isBarrier = 1, hasSideEffects = 1, isTerminator = 1, isCodeGenOnly = 1,
     Defs = [ R7, LR, SP ] in
 def tInt_eh_sjlj_longjmp : XI<(outs), (ins GPR:$src, GPR:$scratch),
                               AddrModeNone, 0, IndexModeNone,
                               Pseudo, NoItinerary, "", "",
                               [(ARMeh_sjlj_longjmp GPR:$src, GPR:$scratch)]>,
-                             Requires<[IsThumb, IsDarwin]>;
+                             Requires<[IsThumb]>;
 
 //===----------------------------------------------------------------------===//
 // Non-Instruction Patterns
@@ -1266,30 +1326,49 @@ def : T1Pat<(addc   tGPR:$lhs, imm8_255_neg:$rhs),
 def : T1Pat<(subc   tGPR:$lhs, tGPR:$rhs),
             (tSUBrr tGPR:$lhs, tGPR:$rhs)>;
 
-// ConstantPool, GlobalAddress
-def : T1Pat<(ARMWrapper  tglobaladdr :$dst), (tLEApcrel tglobaladdr :$dst)>;
+// Bswap 16 with load/store
+def : T1Pat<(srl (bswap (extloadi16 t_addrmode_rrs2:$addr)), (i32 16)),
+            (tREV16 (tLDRHr t_addrmode_rrs2:$addr))>;
+def : T1Pat<(srl (bswap (extloadi16 t_addrmode_is2:$addr)), (i32 16)),
+            (tREV16 (tLDRHi t_addrmode_is2:$addr))>;
+def : T1Pat<(truncstorei16 (srl (bswap tGPR:$Rn), (i32 16)),
+                           t_addrmode_rrs2:$addr),
+            (tSTRHr (tREV16 tGPR:$Rn), t_addrmode_rrs2:$addr)>;
+def : T1Pat<(truncstorei16 (srl (bswap tGPR:$Rn), (i32 16)),
+                           t_addrmode_is2:$addr),
+            (tSTRHi(tREV16 tGPR:$Rn), t_addrmode_is2:$addr)>;
+
+// ConstantPool
 def : T1Pat<(ARMWrapper  tconstpool  :$dst), (tLEApcrel tconstpool  :$dst)>;
 
+// GlobalAddress
+def tLDRLIT_ga_pcrel : PseudoInst<(outs tGPR:$dst), (ins i32imm:$addr),
+                                  IIC_iLoadiALU,
+                                  [(set tGPR:$dst,
+                                        (ARMWrapperPIC tglobaladdr:$addr))]>,
+                       Requires<[IsThumb, DontUseMovt]>;
+
+def tLDRLIT_ga_abs : PseudoInst<(outs tGPR:$dst), (ins i32imm:$src),
+                                IIC_iLoad_i,
+                                [(set tGPR:$dst,
+                                      (ARMWrapper tglobaladdr:$src))]>,
+                     Requires<[IsThumb, DontUseMovt]>;
+
+
 // JumpTable
 def : T1Pat<(ARMWrapperJT tjumptable:$dst, imm:$id),
             (tLEApcrelJT tjumptable:$dst, imm:$id)>;
 
 // Direct calls
 def : T1Pat<(ARMtcall texternalsym:$func), (tBL texternalsym:$func)>,
-      Requires<[IsThumb, IsNotDarwin]>;
-def : T1Pat<(ARMtcall texternalsym:$func), (tBLr9 texternalsym:$func)>,
-      Requires<[IsThumb, IsDarwin]>;
+      Requires<[IsThumb]>;
 
 def : Tv5Pat<(ARMcall texternalsym:$func), (tBLXi texternalsym:$func)>,
-      Requires<[IsThumb, HasV5T, IsNotDarwin]>;
-def : Tv5Pat<(ARMcall texternalsym:$func), (tBLXi_r9 texternalsym:$func)>,
-      Requires<[IsThumb, HasV5T, IsDarwin]>;
+      Requires<[IsThumb, HasV5T, IsNotMClass]>;
 
 // Indirect calls to ARM routines
 def : Tv5Pat<(ARMcall GPR:$dst), (tBLXr GPR:$dst)>,
-      Requires<[IsThumb, HasV5T, IsNotDarwin]>;
-def : Tv5Pat<(ARMcall GPR:$dst), (tBLXr_r9 GPR:$dst)>,
-      Requires<[IsThumb, HasV5T, IsDarwin]>;
+      Requires<[IsThumb, HasV5T]>;
 
 // zextload i1 -> zextload i8
 def : T1Pat<(zextloadi1 t_addrmode_rrs1:$addr),
@@ -1297,6 +1376,17 @@ def : T1Pat<(zextloadi1 t_addrmode_rrs1:$addr),
 def : T1Pat<(zextloadi1 t_addrmode_is1:$addr),
             (tLDRBi t_addrmode_is1:$addr)>;
 
+// extload from the stack -> word load from the stack, as it avoids having to
+// materialize the base in a separate register. This only works when a word
+// load puts the byte/halfword value in the same place in the register that the
+// byte/halfword load would, i.e. when little-endian.
+def : T1Pat<(extloadi1  t_addrmode_sp:$addr), (tLDRspi t_addrmode_sp:$addr)>,
+      Requires<[IsThumb, IsThumb1Only, IsLE]>;
+def : T1Pat<(extloadi8  t_addrmode_sp:$addr), (tLDRspi t_addrmode_sp:$addr)>,
+      Requires<[IsThumb, IsThumb1Only, IsLE]>;
+def : T1Pat<(extloadi16 t_addrmode_sp:$addr), (tLDRspi t_addrmode_sp:$addr)>,
+      Requires<[IsThumb, IsThumb1Only, IsLE]>;
+
 // extload -> zextload
 def : T1Pat<(extloadi1  t_addrmode_rrs1:$addr), (tLDRBr t_addrmode_rrs1:$addr)>;
 def : T1Pat<(extloadi1  t_addrmode_is1:$addr),  (tLDRBi t_addrmode_is1:$addr)>;
@@ -1329,6 +1419,31 @@ def : T1Pat<(sextloadi16 t_addrmode_rrs2:$addr),
 def : T1Pat<(sextloadi16 t_addrmode_is2:$addr),
             (tASRri (tLSLri (tLDRHi t_addrmode_is2:$addr), 16), 16)>;
 
+def : T1Pat<(atomic_load_8 t_addrmode_is1:$src),
+             (tLDRBi t_addrmode_is1:$src)>;
+def : T1Pat<(atomic_load_8 t_addrmode_rrs1:$src),
+             (tLDRBr t_addrmode_rrs1:$src)>;
+def : T1Pat<(atomic_load_16 t_addrmode_is2:$src),
+             (tLDRHi t_addrmode_is2:$src)>;
+def : T1Pat<(atomic_load_16 t_addrmode_rrs2:$src),
+             (tLDRHr t_addrmode_rrs2:$src)>;
+def : T1Pat<(atomic_load_32 t_addrmode_is4:$src),
+             (tLDRi t_addrmode_is4:$src)>;
+def : T1Pat<(atomic_load_32 t_addrmode_rrs4:$src),
+             (tLDRr t_addrmode_rrs4:$src)>;
+def : T1Pat<(atomic_store_8 t_addrmode_is1:$ptr, tGPR:$val),
+             (tSTRBi tGPR:$val, t_addrmode_is1:$ptr)>;
+def : T1Pat<(atomic_store_8 t_addrmode_rrs1:$ptr, tGPR:$val),
+             (tSTRBr tGPR:$val, t_addrmode_rrs1:$ptr)>;
+def : T1Pat<(atomic_store_16 t_addrmode_is2:$ptr, tGPR:$val),
+             (tSTRHi tGPR:$val, t_addrmode_is2:$ptr)>;
+def : T1Pat<(atomic_store_16 t_addrmode_rrs2:$ptr, tGPR:$val),
+             (tSTRHr tGPR:$val, t_addrmode_rrs2:$ptr)>;
+def : T1Pat<(atomic_store_32 t_addrmode_is4:$ptr, tGPR:$val),
+             (tSTRi tGPR:$val, t_addrmode_is4:$ptr)>;
+def : T1Pat<(atomic_store_32 t_addrmode_rrs4:$ptr, tGPR:$val),
+             (tSTRr tGPR:$val, t_addrmode_rrs4:$ptr)>;
+
 // Large immediate handling.
 
 // Two piece imms.
@@ -1355,11 +1470,37 @@ let isReturn = 1, isTerminator = 1, isBarrier = 1, mayLoad = 1,
     hasExtraDefRegAllocReq = 1 in
 def tPOP_RET : tPseudoExpand<(outs), (ins pred:$p, reglist:$regs, variable_ops),
                            2, IIC_iPop_Br, [],
-                           (tPOP pred:$p, reglist:$regs)>;
+                           (tPOP pred:$p, reglist:$regs)>, Sched<[WriteBrL]>;
 
 // Indirect branch using "mov pc, $Rm"
 let isBranch = 1, isTerminator = 1, isBarrier = 1, isIndirectBranch = 1 in {
   def tBRIND : tPseudoExpand<(outs), (ins GPR:$Rm, pred:$p),
                   2, IIC_Br, [(brind GPR:$Rm)],
-                  (tMOVr PC, GPR:$Rm, pred:$p)>;
+                  (tMOVr PC, GPR:$Rm, pred:$p)>, Sched<[WriteBr]>;
 }
+
+
+// In Thumb1, "nop" is encoded as a "mov r8, r8". Technically, the bf00
+// encoding is available on ARMv6K, but we don't differentiate that finely.
+def : InstAlias<"nop", (tMOVr R8, R8, 14, 0)>,Requires<[IsThumb, IsThumb1Only]>;
+
+
+// For round-trip assembly/disassembly, we have to handle a CPS instruction
+// without any iflags. That's not, strictly speaking, valid syntax, but it's
+// a useful extension and assembles to defined behaviour (the insn does
+// nothing).
+def : tInstAlias<"cps$imod", (tCPS imod_op:$imod, 0)>;
+def : tInstAlias<"cps$imod", (tCPS imod_op:$imod, 0)>;
+
+// "neg" is and alias for "rsb rd, rn, #0"
+def : tInstAlias<"neg${s}${p} $Rd, $Rm",
+                 (tRSB tGPR:$Rd, s_cc_out:$s, tGPR:$Rm, pred:$p)>;
+
+
+// Implied destination operand forms for shifts.
+def : tInstAlias<"lsl${s}${p} $Rdm, $imm",
+             (tLSLri tGPR:$Rdm, cc_out:$s, tGPR:$Rdm, imm0_31:$imm, pred:$p)>;
+def : tInstAlias<"lsr${s}${p} $Rdm, $imm",
+             (tLSRri tGPR:$Rdm, cc_out:$s, tGPR:$Rdm, imm_sr:$imm, pred:$p)>;
+def : tInstAlias<"asr${s}${p} $Rdm, $imm",
+             (tASRri tGPR:$Rdm, cc_out:$s, tGPR:$Rdm, imm_sr:$imm, pred:$p)>;