MFC 261991:

Upgrade our copy of llvm/clang to 3.4 release.  This version supports
all of the features in the current working draft of the upcoming C++
standard, provisionally named C++1y.

The code generator's performance is greatly increased, and the loop
auto-vectorizer is now enabled at -Os and -O2 in addition to -O3.  The
PowerPC backend has made several major improvements to code generation
quality and compile time, and the X86, SPARC, ARM32, Aarch64 and SystemZ
backends have all seen major feature work.

Release notes for llvm and clang can be found here:
<http://llvm.org/releases/3.4/docs/ReleaseNotes.html>
<http://llvm.org/releases/3.4/tools/clang/docs/ReleaseNotes.html>

MFC 262121 (by emaste):

Update lldb for clang/llvm 3.4 import

This commit largely restores the lldb source to the upstream r196259
snapshot with the addition of threaded inferior support and a few bug
fixes.

Specific upstream lldb revisions restored include:
   SVN      git
  181387  779e6ac
  181703  7bef4e2
  182099  b31044e
  182650  f2dcf35
  182683  0d91b80
  183862  15c1774
  183929  99447a6
  184177  0b2934b
  184948  4dc3761
  184954  007e7bc
  186990  eebd175

Sponsored by:	DARPA, AFRL

MFC 262186 (by emaste):

Fix mismerge in r262121

A break statement was lost in the merge.  The error had no functional
impact, but restore it to reduce the diff against upstream.

MFC 262303:

Pull in r197521 from upstream clang trunk (by rdivacky):

  Use the integrated assembler by default on FreeBSD/ppc and ppc64.

Requested by:	jhibbits

MFC 262611:

Pull in r196874 from upstream llvm trunk:

  Fix a crash that occurs when PWD is invalid.

  MCJIT needs to be able to run in hostile environments, even when PWD
  is invalid. There's no need to crash MCJIT in this case.

  The obvious fix is to simply leave MCContext's CompilationDir empty
  when PWD can't be determined. This way, MCJIT clients,
  and other clients that link with LLVM don't need a valid working directory.

  If we do want to guarantee valid CompilationDir, that should be done
  only for clients of getCompilationDir(). This is as simple as checking
  for an empty string.

  The only current use of getCompilationDir is EmitGenDwarfInfo, which
  won't conceivably run with an invalid working dir. However, in the
  purely hypothetically and untestable case that this happens, the
  AT_comp_dir will be omitted from the compilation_unit DIE.

This should help fix assertions occurring with ports-mgmt/tinderbox,
when it is using jails, and sometimes invalidates clang's current
working directory.

Reported by:	decke

MFC 262809:

Pull in r203007 from upstream clang trunk:

  Don't produce an alias between destructors with different calling conventions.

  Fixes pr19007.

(Please note that is an LLVM PR identifier, not a FreeBSD one.)

This should fix Firefox and/or libxul crashes (due to problems with
regparm/stdcall calling conventions) on i386.

Reported by:	multiple users on freebsd-current
PR:		bin/187103

MFC 263048:

Repair recognition of "CC" as an alias for the C++ compiler, since it
was silently broken by upstream for a Windows-specific use-case.

Apparently some versions of CMake still rely on this archaic feature...

Reported by:	rakuco

MFC 263049:

Garbage collect the old way of adding the libstdc++ include directories
in clang's InitHeaderSearch.cpp.  This has been superseded by David
Chisnall's commit in r255321.

Moreover, if libc++ is used, the libstdc++ include directories should
not be in the search path at all.  These directories are now only used
if you pass -stdlib=libstdc++.

1 files changed, 586 insertions, 299 deletions

diff --git a/contrib/llvm/lib/Target/ARM/ARMInstrInfo.td b/contrib/llvm/lib/Target/ARM/ARMInstrInfo.td
index 1bd174e..2042c04 100644
--- a/contrib/llvm/lib/Target/ARM/ARMInstrInfo.td
+++ b/contrib/llvm/lib/Target/ARM/ARMInstrInfo.td
@@ -71,6 +71,9 @@ def SDT_ARMTCRET : SDTypeProfile<0, 1, [SDTCisPtrTy<0>]>;
 def SDT_ARMBFI : SDTypeProfile<1, 3, [SDTCisVT<0, i32>, SDTCisVT<1, i32>,
                                       SDTCisVT<2, i32>, SDTCisVT<3, i32>]>;
 
+def SDT_ARMVMAXNM : SDTypeProfile<1, 2, [SDTCisFP<0>, SDTCisFP<1>, SDTCisFP<2>]>;
+def SDT_ARMVMINNM : SDTypeProfile<1, 2, [SDTCisFP<0>, SDTCisFP<1>, SDTCisFP<2>]>;
+
 def SDTBinaryArithWithFlags : SDTypeProfile<2, 2,
                                             [SDTCisSameAs<0, 2>,
                                              SDTCisSameAs<0, 3>,
@@ -118,7 +121,8 @@ def ARMcall_nolink   : SDNode<"ARMISD::CALL_NOLINK", SDT_ARMcall,
 
 def ARMretflag       : SDNode<"ARMISD::RET_FLAG", SDTNone,
                               [SDNPHasChain, SDNPOptInGlue, SDNPVariadic]>;
-
+def ARMintretflag    : SDNode<"ARMISD::INTRET_FLAG", SDT_ARMcall,
+                              [SDNPHasChain, SDNPOptInGlue, SDNPVariadic]>;
 def ARMcmov          : SDNode<"ARMISD::CMOV", SDT_ARMCMov,
                               [SDNPInGlue]>;
 
@@ -162,8 +166,6 @@ def ARMeh_sjlj_longjmp: SDNode<"ARMISD::EH_SJLJ_LONGJMP",
                                SDT_ARMEH_SJLJ_Longjmp,
                                [SDNPHasChain, SDNPSideEffect]>;
 
-def ARMMemBarrier     : SDNode<"ARMISD::MEMBARRIER", SDT_ARMMEMBARRIER,
-                               [SDNPHasChain, SDNPSideEffect]>;
 def ARMMemBarrierMCR  : SDNode<"ARMISD::MEMBARRIER_MCR", SDT_ARMMEMBARRIER,
                                [SDNPHasChain, SDNPSideEffect]>;
 def ARMPreload        : SDNode<"ARMISD::PRELOAD", SDT_ARMPREFETCH,
@@ -174,9 +176,11 @@ def ARMrbit          : SDNode<"ARMISD::RBIT", SDTIntUnaryOp>;
 def ARMtcret         : SDNode<"ARMISD::TC_RETURN", SDT_ARMTCRET,
                         [SDNPHasChain,  SDNPOptInGlue, SDNPVariadic]>;
 
-
 def ARMbfi           : SDNode<"ARMISD::BFI", SDT_ARMBFI>;
 
+def ARMvmaxnm        : SDNode<"ARMISD::VMAXNM", SDT_ARMVMAXNM, []>;
+def ARMvminnm        : SDNode<"ARMISD::VMINNM", SDT_ARMVMINNM, []>;
+
 //===----------------------------------------------------------------------===//
 // ARM Instruction Predicate Definitions.
 //
@@ -189,11 +193,18 @@ def HasV5TE          : Predicate<"Subtarget->hasV5TEOps()">,
 def HasV6            : Predicate<"Subtarget->hasV6Ops()">,
                                  AssemblerPredicate<"HasV6Ops", "armv6">;
 def NoV6             : Predicate<"!Subtarget->hasV6Ops()">;
+def HasV6M           : Predicate<"Subtarget->hasV6MOps()">,
+                                 AssemblerPredicate<"HasV6MOps",
+                                                    "armv6m or armv6t2">;
 def HasV6T2          : Predicate<"Subtarget->hasV6T2Ops()">,
                                  AssemblerPredicate<"HasV6T2Ops", "armv6t2">;
 def NoV6T2           : Predicate<"!Subtarget->hasV6T2Ops()">;
 def HasV7            : Predicate<"Subtarget->hasV7Ops()">,
                                  AssemblerPredicate<"HasV7Ops", "armv7">;
+def HasV8            : Predicate<"Subtarget->hasV8Ops()">,
+                                 AssemblerPredicate<"HasV8Ops", "armv8">;
+def PreV8            : Predicate<"!Subtarget->hasV8Ops()">,
+                                 AssemblerPredicate<"!HasV8Ops", "armv7 or earlier">;
 def NoVFP            : Predicate<"!Subtarget->hasVFP2()">;
 def HasVFP2          : Predicate<"Subtarget->hasVFP2()">,
                                  AssemblerPredicate<"FeatureVFP2", "VFP2">;
@@ -201,14 +212,23 @@ def HasVFP3          : Predicate<"Subtarget->hasVFP3()">,
                                  AssemblerPredicate<"FeatureVFP3", "VFP3">;
 def HasVFP4          : Predicate<"Subtarget->hasVFP4()">,
                                  AssemblerPredicate<"FeatureVFP4", "VFP4">;
+def HasDPVFP         : Predicate<"!Subtarget->isFPOnlySP()">,
+                                 AssemblerPredicate<"!FeatureVFPOnlySP",
+                                                    "double precision VFP">;
+def HasFPARMv8       : Predicate<"Subtarget->hasFPARMv8()">,
+                                 AssemblerPredicate<"FeatureFPARMv8", "FPARMv8">;
 def HasNEON          : Predicate<"Subtarget->hasNEON()">,
                                  AssemblerPredicate<"FeatureNEON", "NEON">;
+def HasCrypto        : Predicate<"Subtarget->hasCrypto()">,
+                                 AssemblerPredicate<"FeatureCrypto", "crypto">;
+def HasCRC           : Predicate<"Subtarget->hasCRC()">,
+                                 AssemblerPredicate<"FeatureCRC", "crc">;
 def HasFP16          : Predicate<"Subtarget->hasFP16()">,
                                  AssemblerPredicate<"FeatureFP16","half-float">;
 def HasDivide        : Predicate<"Subtarget->hasDivide()">,
-                                 AssemblerPredicate<"FeatureHWDiv", "divide">;
+                                 AssemblerPredicate<"FeatureHWDiv", "divide in THUMB">;
 def HasDivideInARM   : Predicate<"Subtarget->hasDivideInARMMode()">,
-                                 AssemblerPredicate<"FeatureHWDivARM">;
+                                 AssemblerPredicate<"FeatureHWDivARM", "divide in ARM">;
 def HasT2ExtractPack : Predicate<"Subtarget->hasT2ExtractPack()">,
                                  AssemblerPredicate<"FeatureT2XtPk",
                                                      "pack/extract">;
@@ -233,10 +253,10 @@ def IsThumb2         : Predicate<"Subtarget->isThumb2()">,
                                  AssemblerPredicate<"ModeThumb,FeatureThumb2",
                                                     "thumb2">;
 def IsMClass         : Predicate<"Subtarget->isMClass()">,
-                                 AssemblerPredicate<"FeatureMClass", "armv7m">;
-def IsARClass        : Predicate<"!Subtarget->isMClass()">,
+                                 AssemblerPredicate<"FeatureMClass", "armv*m">;
+def IsNotMClass      : Predicate<"!Subtarget->isMClass()">,
                                  AssemblerPredicate<"!FeatureMClass",
-                                                    "armv7a/r">;
+                                                    "!armv*m">;
 def IsARM            : Predicate<"!Subtarget->isThumb()">,
                                  AssemblerPredicate<"!ModeThumb", "arm-mode">;
 def IsIOS            : Predicate<"Subtarget->isTargetIOS()">;
@@ -258,7 +278,9 @@ def UseMulOps        : Predicate<"Subtarget->useMulOps()">;
 def UseFusedMAC      : Predicate<"(TM.Options.AllowFPOpFusion =="
                                  " FPOpFusion::Fast) && "
                                  "!Subtarget->isTargetDarwin()">;
-def DontUseFusedMAC  : Predicate<"!Subtarget->hasVFP4() || "
+def DontUseFusedMAC  : Predicate<"!(TM.Options.AllowFPOpFusion =="
+                                 " FPOpFusion::Fast &&"
+                                 " Subtarget->hasVFP4()) || "
                                  "Subtarget->isTargetDarwin()">;
 
 // VGETLNi32 is microcoded on Swift - prefer VMOV.
@@ -275,8 +297,8 @@ def HasSlowVDUP32 : Predicate<"Subtarget->isSwift()">;
 def UseVMOVSR : Predicate<"Subtarget->isCortexA9() || !Subtarget->useNEONForSinglePrecisionFP()">;
 def DontUseVMOVSR : Predicate<"!Subtarget->isCortexA9() && Subtarget->useNEONForSinglePrecisionFP()">;
 
-def IsLE             : Predicate<"TLI.isLittleEndian()">;
-def IsBE             : Predicate<"TLI.isBigEndian()">;
+def IsLE             : Predicate<"getTargetLowering()->isLittleEndian()">;
+def IsBE             : Predicate<"getTargetLowering()->isBigEndian()">;
 
 //===----------------------------------------------------------------------===//
 // ARM Flag Definitions.
@@ -456,7 +478,7 @@ def AdrLabelAsmOperand : AsmOperandClass { let Name = "AdrLabel"; }
 def adrlabel : Operand<i32> {
   let EncoderMethod = "getAdrLabelOpValue";
   let ParserMatchClass = AdrLabelAsmOperand;
-  let PrintMethod = "printAdrLabelOperand";
+  let PrintMethod = "printAdrLabelOperand<0>";
 }
 
 def neon_vcvt_imm32 : Operand<i32> {
@@ -581,17 +603,6 @@ def imm0_1 : Operand<i32> { let ParserMatchClass = Imm0_1AsmOperand; }
 def Imm0_3AsmOperand: ImmAsmOperand { let Name = "Imm0_3"; }
 def imm0_3 : Operand<i32> { let ParserMatchClass = Imm0_3AsmOperand; }
 
-/// imm0_4 predicate - Immediate in the range [0,4].
-def Imm0_4AsmOperand : ImmAsmOperand
-{ 
-  let Name = "Imm0_4"; 
-  let DiagnosticType = "ImmRange0_4";  
-}
-def imm0_4 : Operand<i32>, ImmLeaf<i32, [{ return Imm >= 0 && Imm < 5; }]> {
-  let ParserMatchClass = Imm0_4AsmOperand;
-  let DecoderMethod = "DecodeImm0_4";
-}
-
 /// imm0_7 predicate - Immediate in the range [0,7].
 def Imm0_7AsmOperand: ImmAsmOperand { let Name = "Imm0_7"; }
 def imm0_7 : Operand<i32>, ImmLeaf<i32, [{
@@ -671,6 +682,15 @@ def imm0_63 : Operand<i32>, ImmLeaf<i32, [{
   let ParserMatchClass = Imm0_63AsmOperand;
 }
 
+/// imm0_239 predicate - Immediate in the range [0,239].
+def Imm0_239AsmOperand : ImmAsmOperand {
+  let Name = "Imm0_239";
+  let DiagnosticType = "ImmRange0_239";
+}
+def imm0_239 : Operand<i32>, ImmLeaf<i32, [{ return Imm >= 0 && Imm < 240; }]> {
+  let ParserMatchClass = Imm0_239AsmOperand;
+}
+
 /// imm0_255 predicate - Immediate in the range [0,255].
 def Imm0_255AsmOperand : ImmAsmOperand { let Name = "Imm0_255"; }
 def imm0_255 : Operand<i32>, ImmLeaf<i32, [{ return Imm >= 0 && Imm < 256; }]> {
@@ -702,6 +722,11 @@ def imm0_65535_expr : Operand<i32> {
   let ParserMatchClass = Imm0_65535ExprAsmOperand;
 }
 
+def Imm256_65535ExprAsmOperand: ImmAsmOperand { let Name = "Imm256_65535Expr"; }
+def imm256_65535_expr : Operand<i32> {
+  let ParserMatchClass = Imm256_65535ExprAsmOperand;
+}
+
 /// imm24b - True if the 32-bit immediate is encodable in 24 bits.
 def Imm24bitAsmOperand: ImmAsmOperand { let Name = "Imm24bit"; }
 def imm24b : Operand<i32>, ImmLeaf<i32, [{
@@ -1007,11 +1032,6 @@ def p_imm : Operand<i32> {
   let DecoderMethod = "DecodeCoprocessor";
 }
 
-def pf_imm : Operand<i32> {
-  let PrintMethod = "printPImmediate";
-  let ParserMatchClass = CoprocNumAsmOperand;
-}
-
 def CoprocRegAsmOperand : AsmOperandClass {
   let Name = "CoprocReg";
   let ParserMethod = "parseCoprocRegOperand";
@@ -1327,7 +1347,7 @@ class AI_ext_rrot<bits<8> opcod, string opc, PatFrag opnode>
   : AExtI<opcod, (outs GPRnopc:$Rd), (ins GPRnopc:$Rm, rot_imm:$rot),
           IIC_iEXTr, opc, "\t$Rd, $Rm$rot",
           [(set GPRnopc:$Rd, (opnode (rotr GPRnopc:$Rm, rot_imm:$rot)))]>,
-       Requires<[IsARM, HasV6]> {
+       Requires<[IsARM, HasV6]>, Sched<[WriteALUsi]> {
   bits<4> Rd;
   bits<4> Rm;
   bits<2> rot;
@@ -1340,11 +1360,11 @@ class AI_ext_rrot<bits<8> opcod, string opc, PatFrag opnode>
 class AI_ext_rrot_np<bits<8> opcod, string opc>
   : AExtI<opcod, (outs GPRnopc:$Rd), (ins GPRnopc:$Rm, rot_imm:$rot),
           IIC_iEXTr, opc, "\t$Rd, $Rm$rot", []>,
-       Requires<[IsARM, HasV6]> {
+       Requires<[IsARM, HasV6]>, Sched<[WriteALUsi]> {
   bits<2> rot;
   let Inst{19-16} = 0b1111;
   let Inst{11-10} = rot;
-}
+ }
 
 /// AI_exta_rrot - A binary operation with two forms: one whose operand is a
 /// register and one whose operand is a register rotated by 8/16/24.
@@ -1353,7 +1373,7 @@ class AI_exta_rrot<bits<8> opcod, string opc, PatFrag opnode>
           IIC_iEXTAr, opc, "\t$Rd, $Rn, $Rm$rot",
           [(set GPRnopc:$Rd, (opnode GPR:$Rn,
                                      (rotr GPRnopc:$Rm, rot_imm:$rot)))]>,
-        Requires<[IsARM, HasV6]> {
+        Requires<[IsARM, HasV6]>, Sched<[WriteALUsr]> {
   bits<4> Rd;
   bits<4> Rm;
   bits<4> Rn;
@@ -1368,7 +1388,7 @@ class AI_exta_rrot<bits<8> opcod, string opc, PatFrag opnode>
 class AI_exta_rrot_np<bits<8> opcod, string opc>
   : AExtI<opcod, (outs GPRnopc:$Rd), (ins GPR:$Rn, GPRnopc:$Rm, rot_imm:$rot),
           IIC_iEXTAr, opc, "\t$Rd, $Rn, $Rm$rot", []>,
-       Requires<[IsARM, HasV6]> {
+       Requires<[IsARM, HasV6]>, Sched<[WriteALUsr]> {
   bits<4> Rn;
   bits<2> rot;
   let Inst{19-16} = Rn;
@@ -1664,53 +1684,11 @@ PseudoInst<(outs), (ins i32imm:$amt, pred:$p), NoItinerary,
            [(ARMcallseq_start timm:$amt)]>;
 }
 
-// Atomic pseudo-insts which will be lowered to ldrexd/strexd loops.
-// (These pseudos use a hand-written selection code).
-let usesCustomInserter = 1, Defs = [CPSR], mayLoad = 1, mayStore = 1 in {
-def ATOMOR6432   : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMXOR6432  : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMADD6432  : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMSUB6432  : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMNAND6432 : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMAND6432  : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMSWAP6432 : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMCMPXCHG6432 : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                                 (ins GPR:$addr, GPR:$cmp1, GPR:$cmp2,
-                                      GPR:$set1, GPR:$set2),
-                                 NoItinerary, []>;
-def ATOMMIN6432  : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMUMIN6432  : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMMAX6432  : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-def ATOMUMAX6432  : PseudoInst<(outs GPR:$dst1, GPR:$dst2),
-                              (ins GPR:$addr, GPR:$src1, GPR:$src2),
-                              NoItinerary, []>;
-}
-
-def HINT : AI<(outs), (ins imm0_4:$imm), MiscFrm, NoItinerary,
+def HINT : AI<(outs), (ins imm0_239:$imm), MiscFrm, NoItinerary,
               "hint", "\t$imm", []>, Requires<[IsARM, HasV6]> {
-  bits<3> imm;
-  let Inst{27-3} = 0b0011001000001111000000000;
-  let Inst{2-0} = imm;
+  bits<8> imm;
+  let Inst{27-8} = 0b00110010000011110000;
+  let Inst{7-0} = imm;
 }
 
 def : InstAlias<"nop$p", (HINT 0, pred:$p)>, Requires<[IsARM, HasV6T2]>;
@@ -1718,6 +1696,9 @@ def : InstAlias<"yield$p", (HINT 1, pred:$p)>, Requires<[IsARM, HasV6T2]>;
 def : InstAlias<"wfe$p", (HINT 2, pred:$p)>, Requires<[IsARM, HasV6T2]>;
 def : InstAlias<"wfi$p", (HINT 3, pred:$p)>, Requires<[IsARM, HasV6T2]>;
 def : InstAlias<"sev$p", (HINT 4, pred:$p)>, Requires<[IsARM, HasV6T2]>;
+def : InstAlias<"sevl$p", (HINT 5, pred:$p)>, Requires<[IsARM, HasV8]>;
+
+def : Pat<(int_arm_sevl), (HINT 5)>;
 
 def SEL : AI<(outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm), DPFrm, NoItinerary, "sel",
              "\t$Rd, $Rn, $Rm", []>, Requires<[IsARM, HasV6]> {
@@ -1735,12 +1716,23 @@ def SEL : AI<(outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm), DPFrm, NoItinerary, "sel",
 
 // The 16-bit operand $val can be used by a debugger to store more information
 // about the breakpoint.
-def BKPT : AI<(outs), (ins imm0_65535:$val), MiscFrm, NoItinerary,
-              "bkpt", "\t$val", []>, Requires<[IsARM]> {
+def BKPT : AInoP<(outs), (ins imm0_65535:$val), MiscFrm, NoItinerary,
+                 "bkpt", "\t$val", []>, Requires<[IsARM]> {
   bits<16> val;
   let Inst{3-0} = val{3-0};
   let Inst{19-8} = val{15-4};
   let Inst{27-20} = 0b00010010;
+  let Inst{31-28} = 0xe; // AL
+  let Inst{7-4} = 0b0111;
+}
+
+def HLT : AInoP<(outs), (ins imm0_65535:$val), MiscFrm, NoItinerary,
+                 "hlt", "\t$val", []>, Requires<[IsARM, HasV8]> {
+  bits<16> val;
+  let Inst{3-0} = val{3-0};
+  let Inst{19-8} = val{15-4};
+  let Inst{27-20} = 0b00010000;
+  let Inst{31-28} = 0xe; // AL
   let Inst{7-4} = 0b0111;
 }
 
@@ -1780,7 +1772,8 @@ multiclass APreLoad<bits<1> read, bits<1> data, string opc> {
 
   def i12 : AXI<(outs), (ins addrmode_imm12:$addr), MiscFrm, IIC_Preload,
                 !strconcat(opc, "\t$addr"),
-                [(ARMPreload addrmode_imm12:$addr, (i32 read), (i32 data))]> {
+                [(ARMPreload addrmode_imm12:$addr, (i32 read), (i32 data))]>,
+                Sched<[WritePreLd]> {
     bits<4> Rt;
     bits<17> addr;
     let Inst{31-26} = 0b111101;
@@ -1796,7 +1789,8 @@ multiclass APreLoad<bits<1> read, bits<1> data, string opc> {
 
   def rs : AXI<(outs), (ins ldst_so_reg:$shift), MiscFrm, IIC_Preload,
                !strconcat(opc, "\t$shift"),
-               [(ARMPreload ldst_so_reg:$shift, (i32 read), (i32 data))]> {
+               [(ARMPreload ldst_so_reg:$shift, (i32 read), (i32 data))]>,
+               Sched<[WritePreLd]> {
     bits<17> shift;
     let Inst{31-26} = 0b111101;
     let Inst{25} = 1; // 1 for register form
@@ -1816,7 +1810,7 @@ defm PLDW : APreLoad<0, 1, "pldw">, Requires<[IsARM,HasV7,HasMP]>;
 defm PLI  : APreLoad<1, 0, "pli">,  Requires<[IsARM,HasV7]>;
 
 def SETEND : AXI<(outs), (ins setend_op:$end), MiscFrm, NoItinerary,
-                 "setend\t$end", []>, Requires<[IsARM]> {
+                 "setend\t$end", []>, Requires<[IsARM]>, Deprecated<HasV8Ops> {
   bits<1> end;
   let Inst{31-10} = 0b1111000100000001000000;
   let Inst{9} = end;
@@ -1863,7 +1857,8 @@ def TRAP : AXI<(outs), (ins), MiscFrm, NoItinerary,
 let isNotDuplicable = 1 in {
 def PICADD  : ARMPseudoInst<(outs GPR:$dst), (ins GPR:$a, pclabel:$cp, pred:$p),
                             4, IIC_iALUr,
-                            [(set GPR:$dst, (ARMpic_add GPR:$a, imm:$cp))]>;
+                            [(set GPR:$dst, (ARMpic_add GPR:$a, imm:$cp))]>,
+                            Sched<[WriteALU, ReadALU]>;
 
 let AddedComplexity = 10 in {
 def PICLDR  : ARMPseudoInst<(outs GPR:$dst), (ins addrmodepc:$addr, pred:$p),
@@ -1923,11 +1918,11 @@ def ADR : AI1<{0,?,?,0}, (outs GPR:$Rd), (ins adrlabel:$label),
 
 let hasSideEffects = 1 in {
 def LEApcrel : ARMPseudoInst<(outs GPR:$Rd), (ins i32imm:$label, pred:$p),
-                    4, IIC_iALUi, []>;
+                    4, IIC_iALUi, []>, Sched<[WriteALU, ReadALU]>;
 
 def LEApcrelJT : ARMPseudoInst<(outs GPR:$Rd),
                       (ins i32imm:$label, nohash_imm:$id, pred:$p),
-                      4, IIC_iALUi, []>;
+                      4, IIC_iALUi, []>, Sched<[WriteALU, ReadALU]>;
 }
 
 //===----------------------------------------------------------------------===//
@@ -1938,16 +1933,22 @@ let isReturn = 1, isTerminator = 1, isBarrier = 1 in {
   // ARMV4T and above
   def BX_RET : AI<(outs), (ins), BrMiscFrm, IIC_Br,
                   "bx", "\tlr", [(ARMretflag)]>,
-               Requires<[IsARM, HasV4T]> {
+               Requires<[IsARM, HasV4T]>, Sched<[WriteBr]> {
     let Inst{27-0}  = 0b0001001011111111111100011110;
   }
 
   // ARMV4 only
   def MOVPCLR : AI<(outs), (ins), BrMiscFrm, IIC_Br,
                   "mov", "\tpc, lr", [(ARMretflag)]>,
-               Requires<[IsARM, NoV4T]> {
+               Requires<[IsARM, NoV4T]>, Sched<[WriteBr]> {
     let Inst{27-0} = 0b0001101000001111000000001110;
   }
+
+  // Exception return: N.b. doesn't set CPSR as far as we're concerned (it sets
+  // the user-space one).
+  def SUBS_PC_LR : ARMPseudoInst<(outs), (ins i32imm:$offset, pred:$p),
+                                 4, IIC_Br,
+                                 [(ARMintretflag imm:$offset)]>;
 }
 
 // Indirect branches
@@ -1955,7 +1956,7 @@ let isBranch = 1, isTerminator = 1, isBarrier = 1, isIndirectBranch = 1 in {
   // ARMV4T and above
   def BX : AXI<(outs), (ins GPR:$dst), BrMiscFrm, IIC_Br, "bx\t$dst",
                   [(brind GPR:$dst)]>,
-              Requires<[IsARM, HasV4T]> {
+              Requires<[IsARM, HasV4T]>, Sched<[WriteBr]> {
     bits<4> dst;
     let Inst{31-4} = 0b1110000100101111111111110001;
     let Inst{3-0}  = dst;
@@ -1963,7 +1964,7 @@ let isBranch = 1, isTerminator = 1, isBarrier = 1, isIndirectBranch = 1 in {
 
   def BX_pred : AI<(outs), (ins GPR:$dst), BrMiscFrm, IIC_Br,
                   "bx", "\t$dst", [/* pattern left blank */]>,
-              Requires<[IsARM, HasV4T]> {
+              Requires<[IsARM, HasV4T]>, Sched<[WriteBr]> {
     bits<4> dst;
     let Inst{27-4} = 0b000100101111111111110001;
     let Inst{3-0}  = dst;
@@ -1980,7 +1981,7 @@ let isCall = 1,
   def BL  : ABXI<0b1011, (outs), (ins bl_target:$func),
                 IIC_Br, "bl\t$func",
                 [(ARMcall tglobaladdr:$func)]>,
-            Requires<[IsARM]> {
+            Requires<[IsARM]>, Sched<[WriteBrL]> {
     let Inst{31-28} = 0b1110;
     bits<24> func;
     let Inst{23-0} = func;
@@ -1990,7 +1991,7 @@ let isCall = 1,
   def BL_pred : ABI<0b1011, (outs), (ins bl_target:$func),
                    IIC_Br, "bl", "\t$func",
                    [(ARMcall_pred tglobaladdr:$func)]>,
-                Requires<[IsARM]> {
+                Requires<[IsARM]>, Sched<[WriteBrL]> {
     bits<24> func;
     let Inst{23-0} = func;
     let DecoderMethod = "DecodeBranchImmInstruction";
@@ -2000,7 +2001,7 @@ let isCall = 1,
   def BLX : AXI<(outs), (ins GPR:$func), BrMiscFrm,
                 IIC_Br, "blx\t$func",
                 [(ARMcall GPR:$func)]>,
-            Requires<[IsARM, HasV5T]> {
+            Requires<[IsARM, HasV5T]>, Sched<[WriteBrL]> {
     bits<4> func;
     let Inst{31-4} = 0b1110000100101111111111110011;
     let Inst{3-0}  = func;
@@ -2009,7 +2010,7 @@ let isCall = 1,
   def BLX_pred : AI<(outs), (ins GPR:$func), BrMiscFrm,
                     IIC_Br, "blx", "\t$func",
                     [(ARMcall_pred GPR:$func)]>,
-                 Requires<[IsARM, HasV5T]> {
+                 Requires<[IsARM, HasV5T]>, Sched<[WriteBrL]> {
     bits<4> func;
     let Inst{27-4} = 0b000100101111111111110011;
     let Inst{3-0}  = func;
@@ -2019,18 +2020,18 @@ let isCall = 1,
   // Note: Restrict $func to the tGPR regclass to prevent it being in LR.
   def BX_CALL : ARMPseudoInst<(outs), (ins tGPR:$func),
                    8, IIC_Br, [(ARMcall_nolink tGPR:$func)]>,
-                   Requires<[IsARM, HasV4T]>;
+                   Requires<[IsARM, HasV4T]>, Sched<[WriteBr]>;
 
   // ARMv4
   def BMOVPCRX_CALL : ARMPseudoInst<(outs), (ins tGPR:$func),
                    8, IIC_Br, [(ARMcall_nolink tGPR:$func)]>,
-                   Requires<[IsARM, NoV4T]>;
+                   Requires<[IsARM, NoV4T]>, Sched<[WriteBr]>;
 
   // mov lr, pc; b if callee is marked noreturn to avoid confusing the
   // return stack predictor.
   def BMOVPCB_CALL : ARMPseudoInst<(outs), (ins bl_target:$func),
                                8, IIC_Br, [(ARMcall_nolink tglobaladdr:$func)]>,
-                      Requires<[IsARM]>;
+                      Requires<[IsARM]>, Sched<[WriteBr]>;
 }
 
 let isBranch = 1, isTerminator = 1 in {
@@ -2038,7 +2039,8 @@ let isBranch = 1, isTerminator = 1 in {
   // a two-value operand where a dag node expects two operands. :(
   def Bcc : ABI<0b1010, (outs), (ins br_target:$target),
                IIC_Br, "b", "\t$target",
-               [/*(ARMbrcond bb:$target, imm:$cc, CCR:$ccr)*/]> {
+               [/*(ARMbrcond bb:$target, imm:$cc, CCR:$ccr)*/]>,
+               Sched<[WriteBr]>  {
     bits<24> target;
     let Inst{23-0} = target;
     let DecoderMethod = "DecodeBranchImmInstruction";
@@ -2051,25 +2053,27 @@ let isBranch = 1, isTerminator = 1 in {
     // should be sufficient.
     // FIXME: Is B really a Barrier? That doesn't seem right.
     def B : ARMPseudoExpand<(outs), (ins br_target:$target), 4, IIC_Br,
-                [(br bb:$target)], (Bcc br_target:$target, (ops 14, zero_reg))>;
+                [(br bb:$target)], (Bcc br_target:$target, (ops 14, zero_reg))>,
+                Sched<[WriteBr]>;
 
     let isNotDuplicable = 1, isIndirectBranch = 1 in {
     def BR_JTr : ARMPseudoInst<(outs),
                       (ins GPR:$target, i32imm:$jt, i32imm:$id),
                       0, IIC_Br,
-                      [(ARMbrjt GPR:$target, tjumptable:$jt, imm:$id)]>;
+                      [(ARMbrjt GPR:$target, tjumptable:$jt, imm:$id)]>,
+                      Sched<[WriteBr]>;
     // FIXME: This shouldn't use the generic "addrmode2," but rather be split
     // into i12 and rs suffixed versions.
     def BR_JTm : ARMPseudoInst<(outs),
                      (ins addrmode2:$target, i32imm:$jt, i32imm:$id),
                      0, IIC_Br,
                      [(ARMbrjt (i32 (load addrmode2:$target)), tjumptable:$jt,
-                       imm:$id)]>;
+                       imm:$id)]>, Sched<[WriteBrTbl]>;
     def BR_JTadd : ARMPseudoInst<(outs),
                    (ins GPR:$target, GPR:$idx, i32imm:$jt, i32imm:$id),
                    0, IIC_Br,
                    [(ARMbrjt (add GPR:$target, GPR:$idx), tjumptable:$jt,
-                     imm:$id)]>;
+                     imm:$id)]>, Sched<[WriteBrTbl]>;
     } // isNotDuplicable = 1, isIndirectBranch = 1
   } // isBarrier = 1
 
@@ -2078,7 +2082,7 @@ let isBranch = 1, isTerminator = 1 in {
 // BLX (immediate)
 def BLXi : AXI<(outs), (ins blx_target:$target), BrMiscFrm, NoItinerary,
                "blx\t$target", []>,
-           Requires<[IsARM, HasV5T]> {
+           Requires<[IsARM, HasV5T]>, Sched<[WriteBrL]> {
   let Inst{31-25} = 0b1111101;
   bits<25> target;
   let Inst{23-0} = target{24-1};
@@ -2087,7 +2091,7 @@ def BLXi : AXI<(outs), (ins blx_target:$target), BrMiscFrm, NoItinerary,
 
 // Branch and Exchange Jazelle
 def BXJ : ABI<0b0001, (outs), (ins GPR:$func), NoItinerary, "bxj", "\t$func",
-              [/* pattern left blank */]> {
+              [/* pattern left blank */]>, Sched<[WriteBr]> {
   bits<4> func;
   let Inst{23-20} = 0b0010;
   let Inst{19-8} = 0xfff;
@@ -2098,18 +2102,20 @@ def BXJ : ABI<0b0001, (outs), (ins GPR:$func), NoItinerary, "bxj", "\t$func",
 // Tail calls.
 
 let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1, Uses = [SP] in {
-  def TCRETURNdi : PseudoInst<(outs), (ins i32imm:$dst), IIC_Br, []>;
+  def TCRETURNdi : PseudoInst<(outs), (ins i32imm:$dst), IIC_Br, []>,
+                   Sched<[WriteBr]>;
 
-  def TCRETURNri : PseudoInst<(outs), (ins tcGPR:$dst), IIC_Br, []>;
+  def TCRETURNri : PseudoInst<(outs), (ins tcGPR:$dst), IIC_Br, []>,
+                   Sched<[WriteBr]>;
 
   def TAILJMPd : ARMPseudoExpand<(outs), (ins br_target:$dst),
                                  4, IIC_Br, [],
                                  (Bcc br_target:$dst, (ops 14, zero_reg))>,
-                                 Requires<[IsARM]>;
+                                 Requires<[IsARM]>, Sched<[WriteBr]>;
 
   def TAILJMPr : ARMPseudoExpand<(outs), (ins tcGPR:$dst),
                                  4, IIC_Br, [],
-                                 (BX GPR:$dst)>,
+                                 (BX GPR:$dst)>, Sched<[WriteBr]>,
                                  Requires<[IsARM]>;
 }
 
@@ -2123,7 +2129,8 @@ def SMC : ABI<0b0001, (outs), (ins imm0_15:$opt), NoItinerary, "smc", "\t$opt",
 
 // Supervisor Call (Software Interrupt)
 let isCall = 1, Uses = [SP] in {
-def SVC : ABI<0b1111, (outs), (ins imm24b:$svc), IIC_Br, "svc", "\t$svc", []> {
+def SVC : ABI<0b1111, (outs), (ins imm24b:$svc), IIC_Br, "svc", "\t$svc", []>,
+          Sched<[WriteBr]> {
   bits<24> svc;
   let Inst{23-0} = svc;
 }
@@ -2272,6 +2279,13 @@ def LDRD : AI3ld<0b1101, 0, (outs GPR:$Rd, GPR:$dst2),
                  []>, Requires<[IsARM, HasV5TE]>;
 }
 
+def LDA : AIldracq<0b00, (outs GPR:$Rt), (ins addr_offset_none:$addr),
+                    NoItinerary, "lda", "\t$Rt, $addr", []>;
+def LDAB : AIldracq<0b10, (outs GPR:$Rt), (ins addr_offset_none:$addr),
+                    NoItinerary, "ldab", "\t$Rt, $addr", []>;
+def LDAH : AIldracq<0b11, (outs GPR:$Rt), (ins addr_offset_none:$addr),
+                    NoItinerary, "ldah", "\t$Rt, $addr", []>;
+
 // Indexed loads
 multiclass AI2_ldridx<bit isByte, string opc,
                       InstrItinClass iii, InstrItinClass iir> {
@@ -2284,7 +2298,6 @@ multiclass AI2_ldridx<bit isByte, string opc,
     let Inst{19-16} = addr{16-13};
     let Inst{11-0} = addr{11-0};
     let DecoderMethod = "DecodeLDRPreImm";
-    let AsmMatchConverter = "cvtLdWriteBackRegAddrModeImm12";
   }
 
   def _PRE_REG  : AI2ldstidx<1, isByte, 1, (outs GPR:$Rt, GPR:$Rn_wb),
@@ -2297,7 +2310,6 @@ multiclass AI2_ldridx<bit isByte, string opc,
     let Inst{11-0} = addr{11-0};
     let Inst{4} = 0;
     let DecoderMethod = "DecodeLDRPreReg";
-    let AsmMatchConverter = "cvtLdWriteBackRegAddrMode2";
   }
 
   def _POST_REG : AI2ldstidx<1, isByte, 0, (outs GPR:$Rt, GPR:$Rn_wb),
@@ -2355,7 +2367,6 @@ multiclass AI3_ldridx<bits<4> op, string opc, InstrItinClass itin> {
     let Inst{19-16} = addr{12-9};   // Rn
     let Inst{11-8}  = addr{7-4};    // imm7_4/zero
     let Inst{3-0}   = addr{3-0};    // imm3_0/Rm
-    let AsmMatchConverter = "cvtLdWriteBackRegAddrMode3";
     let DecoderMethod = "DecodeAddrMode3Instruction";
   }
   def _POST : AI3ldstidx<op, 1, 0, (outs GPR:$Rt, GPR:$Rn_wb),
@@ -2391,7 +2402,6 @@ def LDRD_PRE : AI3ldstidx<0b1101, 0, 1, (outs GPR:$Rt, GPR:$Rt2, GPR:$Rn_wb),
   let Inst{11-8}  = addr{7-4};    // imm7_4/zero
   let Inst{3-0}   = addr{3-0};    // imm3_0/Rm
   let DecoderMethod = "DecodeAddrMode3Instruction";
-  let AsmMatchConverter = "cvtLdrdPre";
 }
 def LDRD_POST: AI3ldstidx<0b1101, 0, 0, (outs GPR:$Rt, GPR:$Rt2, GPR:$Rn_wb),
                           (ins addr_offset_none:$addr, am3offset:$offset),
@@ -2494,7 +2504,6 @@ multiclass AI3ldrT<bits<4> op, string opc> {
     let Inst{22} = 1;
     let Inst{11-8} = offset{7-4};
     let Inst{3-0} = offset{3-0};
-    let AsmMatchConverter = "cvtLdExtTWriteBackImm";
   }
   def r : AI3ldstidxT<op, 1, (outs GPRnopc:$Rt, GPRnopc:$base_wb),
                       (ins addr_offset_none:$addr, postidx_reg:$Rm),
@@ -2506,7 +2515,6 @@ multiclass AI3ldrT<bits<4> op, string opc> {
     let Inst{11-8} = 0;
     let Unpredictable{11-8} = 0b1111;
     let Inst{3-0} = Rm{3-0};
-    let AsmMatchConverter = "cvtLdExtTWriteBackReg";
     let DecoderMethod = "DecodeLDR";
   }
 }
@@ -2544,7 +2552,6 @@ multiclass AI2_stridx<bit isByte, string opc,
     let Inst{23}    = addr{12};     // U (add = ('U' == 1))
     let Inst{19-16} = addr{16-13};  // Rn
     let Inst{11-0}  = addr{11-0};   // imm12
-    let AsmMatchConverter = "cvtStWriteBackRegAddrModeImm12";
     let DecoderMethod = "DecodeSTRPreImm";
   }
 
@@ -2558,7 +2565,6 @@ multiclass AI2_stridx<bit isByte, string opc,
     let Inst{19-16} = addr{16-13}; // Rn
     let Inst{11-0}  = addr{11-0};
     let Inst{4}     = 0;           // Inst{4} = 0
-    let AsmMatchConverter = "cvtStWriteBackRegAddrMode2";
     let DecoderMethod = "DecodeSTRPreReg";
   }
   def _POST_REG : AI2ldstidx<0, isByte, 0, (outs GPR:$Rn_wb),
@@ -2667,7 +2673,6 @@ def STRH_PRE  : AI3ldstidx<0b1011, 0, 1, (outs GPR:$Rn_wb),
   let Inst{19-16} = addr{12-9};   // Rn
   let Inst{11-8}  = addr{7-4};    // imm7_4/zero
   let Inst{3-0}   = addr{3-0};    // imm3_0/Rm
-  let AsmMatchConverter = "cvtStWriteBackRegAddrMode3";
   let DecoderMethod = "DecodeAddrMode3Instruction";
 }
 
@@ -2701,7 +2706,6 @@ def STRD_PRE : AI3ldstidx<0b1111, 0, 1, (outs GPR:$Rn_wb),
   let Inst{11-8}  = addr{7-4};    // imm7_4/zero
   let Inst{3-0}   = addr{3-0};    // imm3_0/Rm
   let DecoderMethod = "DecodeAddrMode3Instruction";
-  let AsmMatchConverter = "cvtStrdPre";
 }
 
 def STRD_POST: AI3ldstidx<0b1111, 0, 0, (outs GPR:$Rn_wb),
@@ -2808,7 +2812,6 @@ multiclass AI3strT<bits<4> op, string opc> {
     let Inst{22} = 1;
     let Inst{11-8} = offset{7-4};
     let Inst{3-0} = offset{3-0};
-    let AsmMatchConverter = "cvtStExtTWriteBackImm";
   }
   def r : AI3ldstidxT<op, 0, (outs GPR:$base_wb),
                       (ins GPR:$Rt, addr_offset_none:$addr, postidx_reg:$Rm),
@@ -2819,13 +2822,18 @@ multiclass AI3strT<bits<4> op, string opc> {
     let Inst{22} = 0;
     let Inst{11-8} = 0;
     let Inst{3-0} = Rm{3-0};
-    let AsmMatchConverter = "cvtStExtTWriteBackReg";
   }
 }
 
 
 defm STRHT : AI3strT<0b1011, "strht">;
 
+def STL : AIstrrel<0b00, (outs), (ins GPR:$Rt, addr_offset_none:$addr),
+                   NoItinerary, "stl", "\t$Rt, $addr", []>;
+def STLB : AIstrrel<0b10, (outs), (ins GPR:$Rt, addr_offset_none:$addr),
+                    NoItinerary, "stlb", "\t$Rt, $addr", []>;
+def STLH : AIstrrel<0b11, (outs), (ins GPR:$Rt, addr_offset_none:$addr),
+                    NoItinerary, "stlh", "\t$Rt, $addr", []>;
 
 //===----------------------------------------------------------------------===//
 //  Load / store multiple Instructions.
@@ -2955,7 +2963,7 @@ defm sysSTM : arm_ldst_mult<"stm", " ^", 0, 1, LdStMulFrm, IIC_iStore_m,
 
 let neverHasSideEffects = 1 in
 def MOVr : AsI1<0b1101, (outs GPR:$Rd), (ins GPR:$Rm), DPFrm, IIC_iMOVr,
-                "mov", "\t$Rd, $Rm", []>, UnaryDP {
+                "mov", "\t$Rd, $Rm", []>, UnaryDP, Sched<[WriteALU]> {
   bits<4> Rd;
   bits<4> Rm;
 
@@ -2969,7 +2977,7 @@ def MOVr : AsI1<0b1101, (outs GPR:$Rd), (ins GPR:$Rm), DPFrm, IIC_iMOVr,
 // A version for the smaller set of tail call registers.
 let neverHasSideEffects = 1 in
 def MOVr_TC : AsI1<0b1101, (outs tcGPR:$Rd), (ins tcGPR:$Rm), DPFrm,
-                IIC_iMOVr, "mov", "\t$Rd, $Rm", []>, UnaryDP {
+                IIC_iMOVr, "mov", "\t$Rd, $Rm", []>, UnaryDP, Sched<[WriteALU]> {
   bits<4> Rd;
   bits<4> Rm;
 
@@ -2982,7 +2990,8 @@ def MOVr_TC : AsI1<0b1101, (outs tcGPR:$Rd), (ins tcGPR:$Rm), DPFrm,
 def MOVsr : AsI1<0b1101, (outs GPRnopc:$Rd), (ins shift_so_reg_reg:$src),
                 DPSoRegRegFrm, IIC_iMOVsr,
                 "mov", "\t$Rd, $src",
-                [(set GPRnopc:$Rd, shift_so_reg_reg:$src)]>, UnaryDP {
+                [(set GPRnopc:$Rd, shift_so_reg_reg:$src)]>, UnaryDP,
+                Sched<[WriteALU]> {
   bits<4> Rd;
   bits<12> src;
   let Inst{15-12} = Rd;
@@ -2998,7 +3007,7 @@ def MOVsr : AsI1<0b1101, (outs GPRnopc:$Rd), (ins shift_so_reg_reg:$src),
 def MOVsi : AsI1<0b1101, (outs GPR:$Rd), (ins shift_so_reg_imm:$src),
                 DPSoRegImmFrm, IIC_iMOVsr,
                 "mov", "\t$Rd, $src", [(set GPR:$Rd, shift_so_reg_imm:$src)]>,
-                UnaryDP {
+                UnaryDP, Sched<[WriteALU]> {
   bits<4> Rd;
   bits<12> src;
   let Inst{15-12} = Rd;
@@ -3011,7 +3020,8 @@ def MOVsi : AsI1<0b1101, (outs GPR:$Rd), (ins shift_so_reg_imm:$src),
 
 let isReMaterializable = 1, isAsCheapAsAMove = 1, isMoveImm = 1 in
 def MOVi : AsI1<0b1101, (outs GPR:$Rd), (ins so_imm:$imm), DPFrm, IIC_iMOVi,
-                "mov", "\t$Rd, $imm", [(set GPR:$Rd, so_imm:$imm)]>, UnaryDP {
+                "mov", "\t$Rd, $imm", [(set GPR:$Rd, so_imm:$imm)]>, UnaryDP,
+                Sched<[WriteALU]> {
   bits<4> Rd;
   bits<12> imm;
   let Inst{25} = 1;
@@ -3025,7 +3035,7 @@ def MOVi16 : AI1<0b1000, (outs GPR:$Rd), (ins imm0_65535_expr:$imm),
                  DPFrm, IIC_iMOVi,
                  "movw", "\t$Rd, $imm",
                  [(set GPR:$Rd, imm0_65535:$imm)]>,
-                 Requires<[IsARM, HasV6T2]>, UnaryDP {
+                 Requires<[IsARM, HasV6T2]>, UnaryDP, Sched<[WriteALU]> {
   bits<4> Rd;
   bits<16> imm;
   let Inst{15-12} = Rd;
@@ -3041,7 +3051,8 @@ def : InstAlias<"mov${p} $Rd, $imm",
         Requires<[IsARM]>;
 
 def MOVi16_ga_pcrel : PseudoInst<(outs GPR:$Rd),
-                                (ins i32imm:$addr, pclabel:$id), IIC_iMOVi, []>;
+                                (ins i32imm:$addr, pclabel:$id), IIC_iMOVi, []>,
+                      Sched<[WriteALU]>;
 
 let Constraints = "$src = $Rd" in {
 def MOVTi16 : AI1<0b1010, (outs GPRnopc:$Rd),
@@ -3051,7 +3062,7 @@ def MOVTi16 : AI1<0b1010, (outs GPRnopc:$Rd),
                   [(set GPRnopc:$Rd,
                         (or (and GPR:$src, 0xffff),
                             lo16AllZero:$imm))]>, UnaryDP,
-                  Requires<[IsARM, HasV6T2]> {
+                  Requires<[IsARM, HasV6T2]>, Sched<[WriteALU]> {
   bits<4> Rd;
   bits<16> imm;
   let Inst{15-12} = Rd;
@@ -3063,7 +3074,8 @@ def MOVTi16 : AI1<0b1010, (outs GPRnopc:$Rd),
 }
 
 def MOVTi16_ga_pcrel : PseudoInst<(outs GPR:$Rd),
-                      (ins GPR:$src, i32imm:$addr, pclabel:$id), IIC_iMOVi, []>;
+                      (ins GPR:$src, i32imm:$addr, pclabel:$id), IIC_iMOVi, []>,
+                      Sched<[WriteALU]>;
 
 } // Constraints
 
@@ -3073,7 +3085,7 @@ def : ARMPat<(or GPR:$src, 0xffff0000), (MOVTi16 GPR:$src, 0xffff)>,
 let Uses = [CPSR] in
 def RRX: PseudoInst<(outs GPR:$Rd), (ins GPR:$Rm), IIC_iMOVsi,
                     [(set GPR:$Rd, (ARMrrx GPR:$Rm))]>, UnaryDP,
-                    Requires<[IsARM]>;
+                    Requires<[IsARM]>, Sched<[WriteALU]>;
 
 // These aren't really mov instructions, but we have to define them this way
 // due to flag operands.
@@ -3081,10 +3093,10 @@ def RRX: PseudoInst<(outs GPR:$Rd), (ins GPR:$Rm), IIC_iMOVsi,
 let Defs = [CPSR] in {
 def MOVsrl_flag : PseudoInst<(outs GPR:$dst), (ins GPR:$src), IIC_iMOVsi,
                       [(set GPR:$dst, (ARMsrl_flag GPR:$src))]>, UnaryDP,
-                      Requires<[IsARM]>;
+                      Sched<[WriteALU]>, Requires<[IsARM]>;
 def MOVsra_flag : PseudoInst<(outs GPR:$dst), (ins GPR:$src), IIC_iMOVsi,
                       [(set GPR:$dst, (ARMsra_flag GPR:$src))]>, UnaryDP,
-                      Requires<[IsARM]>;
+                      Sched<[WriteALU]>, Requires<[IsARM]>;
 }
 
 //===----------------------------------------------------------------------===//
@@ -3250,7 +3262,8 @@ class AAI<bits<8> op27_20, bits<8> op11_4, string opc,
           list<dag> pattern = [],
           dag iops = (ins GPRnopc:$Rn, GPRnopc:$Rm),
           string asm = "\t$Rd, $Rn, $Rm">
-  : AI<(outs GPRnopc:$Rd), iops, DPFrm, IIC_iALUr, opc, asm, pattern> {
+  : AI<(outs GPRnopc:$Rd), iops, DPFrm, IIC_iALUr, opc, asm, pattern>,
+    Sched<[WriteALU, ReadALU, ReadALU]> {
   bits<4> Rn;
   bits<4> Rd;
   bits<4> Rm;
@@ -3265,9 +3278,11 @@ class AAI<bits<8> op27_20, bits<8> op11_4, string opc,
 
 // Saturating add/subtract
 
+let DecoderMethod = "DecodeQADDInstruction" in
 def QADD    : AAI<0b00010000, 0b00000101, "qadd",
                   [(set GPRnopc:$Rd, (int_arm_qadd GPRnopc:$Rm, GPRnopc:$Rn))],
                   (ins GPRnopc:$Rm, GPRnopc:$Rn), "\t$Rd, $Rm, $Rn">;
+
 def QSUB    : AAI<0b00010010, 0b00000101, "qsub",
                   [(set GPRnopc:$Rd, (int_arm_qsub GPRnopc:$Rm, GPRnopc:$Rn))],
                   (ins GPRnopc:$Rm, GPRnopc:$Rn), "\t$Rd, $Rm, $Rn">;
@@ -3326,7 +3341,7 @@ def UHSUB8  : AAI<0b01100111, 0b11111111, "uhsub8">;
 def USAD8  : AI<(outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
                 MulFrm /* for convenience */, NoItinerary, "usad8",
                 "\t$Rd, $Rn, $Rm", []>,
-             Requires<[IsARM, HasV6]> {
+             Requires<[IsARM, HasV6]>, Sched<[WriteALU, ReadALU, ReadALU]> {
   bits<4> Rd;
   bits<4> Rn;
   bits<4> Rm;
@@ -3340,7 +3355,7 @@ def USAD8  : AI<(outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm),
 def USADA8 : AI<(outs GPR:$Rd), (ins GPR:$Rn, GPR:$Rm, GPR:$Ra),
                 MulFrm /* for convenience */, NoItinerary, "usada8",
                 "\t$Rd, $Rn, $Rm, $Ra", []>,
-             Requires<[IsARM, HasV6]> {
+             Requires<[IsARM, HasV6]>, Sched<[WriteALU, ReadALU, ReadALU]>{
   bits<4> Rd;
   bits<4> Rn;
   bits<4> Rm;
@@ -3473,7 +3488,7 @@ def BFI:I<(outs GPRnopc:$Rd), (ins GPRnopc:$src, GPR:$Rn, bf_inv_mask_imm:$imm),
 
 def  MVNr  : AsI1<0b1111, (outs GPR:$Rd), (ins GPR:$Rm), DPFrm, IIC_iMVNr,
                   "mvn", "\t$Rd, $Rm",
-                  [(set GPR:$Rd, (not GPR:$Rm))]>, UnaryDP {
+                  [(set GPR:$Rd, (not GPR:$Rm))]>, UnaryDP, Sched<[WriteALU]> {
   bits<4> Rd;
   bits<4> Rm;
   let Inst{25} = 0;
@@ -3484,7 +3499,8 @@ def  MVNr  : AsI1<0b1111, (outs GPR:$Rd), (ins GPR:$Rm), DPFrm, IIC_iMVNr,
 }
 def  MVNsi  : AsI1<0b1111, (outs GPR:$Rd), (ins so_reg_imm:$shift),
                   DPSoRegImmFrm, IIC_iMVNsr, "mvn", "\t$Rd, $shift",
-                  [(set GPR:$Rd, (not so_reg_imm:$shift))]>, UnaryDP {
+                  [(set GPR:$Rd, (not so_reg_imm:$shift))]>, UnaryDP,
+                  Sched<[WriteALU]> {
   bits<4> Rd;
   bits<12> shift;
   let Inst{25} = 0;
@@ -3496,7 +3512,8 @@ def  MVNsi  : AsI1<0b1111, (outs GPR:$Rd), (ins so_reg_imm:$shift),
 }
 def  MVNsr  : AsI1<0b1111, (outs GPR:$Rd), (ins so_reg_reg:$shift),
                   DPSoRegRegFrm, IIC_iMVNsr, "mvn", "\t$Rd, $shift",
-                  [(set GPR:$Rd, (not so_reg_reg:$shift))]>, UnaryDP {
+                  [(set GPR:$Rd, (not so_reg_reg:$shift))]>, UnaryDP,
+                  Sched<[WriteALU]> {
   bits<4> Rd;
   bits<12> shift;
   let Inst{25} = 0;
@@ -3511,7 +3528,7 @@ def  MVNsr  : AsI1<0b1111, (outs GPR:$Rd), (ins so_reg_reg:$shift),
 let isReMaterializable = 1, isAsCheapAsAMove = 1, isMoveImm = 1 in
 def  MVNi  : AsI1<0b1111, (outs GPR:$Rd), (ins so_imm:$imm), DPFrm,
                   IIC_iMVNi, "mvn", "\t$Rd, $imm",
-                  [(set GPR:$Rd, so_imm_not:$imm)]>,UnaryDP {
+                  [(set GPR:$Rd, so_imm_not:$imm)]>,UnaryDP, Sched<[WriteALU]> {
   bits<4> Rd;
   bits<12> imm;
   let Inst{25} = 1;
@@ -3993,14 +4010,58 @@ def PKHTB : APKHI<0b01101000, 1, (outs GPRnopc:$Rd),
 
 // Alternate cases for PKHTB where identities eliminate some nodes.  Note that
 // a shift amount of 0 is *not legal* here, it is PKHBT instead.
+// We also can not replace a srl (17..31) by an arithmetic shift we would use in
+// pkhtb src1, src2, asr (17..31).
 def : ARMV6Pat<(or (and GPRnopc:$src1, 0xFFFF0000),
-                   (srl GPRnopc:$src2, imm16_31:$sh)),
+                   (srl GPRnopc:$src2, imm16:$sh)),
+               (PKHTB GPRnopc:$src1, GPRnopc:$src2, imm16:$sh)>;
+def : ARMV6Pat<(or (and GPRnopc:$src1, 0xFFFF0000),
+                   (sra GPRnopc:$src2, imm16_31:$sh)),
                (PKHTB GPRnopc:$src1, GPRnopc:$src2, imm16_31:$sh)>;
 def : ARMV6Pat<(or (and GPRnopc:$src1, 0xFFFF0000),
                    (and (srl GPRnopc:$src2, imm1_15:$sh), 0xFFFF)),
                (PKHTB GPRnopc:$src1, GPRnopc:$src2, imm1_15:$sh)>;
 
 //===----------------------------------------------------------------------===//
+// CRC Instructions
+//
+// Polynomials:
+// + CRC32{B,H,W}       0x04C11DB7
+// + CRC32C{B,H,W}      0x1EDC6F41
+//
+
+class AI_crc32<bit C, bits<2> sz, string suffix, SDPatternOperator builtin>
+  : AInoP<(outs GPRnopc:$Rd), (ins GPRnopc:$Rn, GPRnopc:$Rm), MiscFrm, NoItinerary,
+               !strconcat("crc32", suffix), "\t$Rd, $Rn, $Rm",
+               [(set GPRnopc:$Rd, (builtin GPRnopc:$Rn, GPRnopc:$Rm))]>,
+               Requires<[IsARM, HasV8, HasCRC]> {
+  bits<4> Rd;
+  bits<4> Rn;
+  bits<4> Rm;
+
+  let Inst{31-28} = 0b1110;
+  let Inst{27-23} = 0b00010;
+  let Inst{22-21} = sz;
+  let Inst{20}    = 0;
+  let Inst{19-16} = Rn;
+  let Inst{15-12} = Rd;
+  let Inst{11-10} = 0b00;
+  let Inst{9}     = C;
+  let Inst{8}     = 0;
+  let Inst{7-4}   = 0b0100;
+  let Inst{3-0}   = Rm;
+
+  let Unpredictable{11-8} = 0b1101;
+}
+
+def CRC32B  : AI_crc32<0, 0b00, "b", int_arm_crc32b>;
+def CRC32CB : AI_crc32<1, 0b00, "cb", int_arm_crc32cb>;
+def CRC32H  : AI_crc32<0, 0b01, "h", int_arm_crc32h>;
+def CRC32CH : AI_crc32<1, 0b01, "ch", int_arm_crc32ch>;
+def CRC32W  : AI_crc32<0, 0b10, "w", int_arm_crc32w>;
+def CRC32CW : AI_crc32<1, 0b10, "cw", int_arm_crc32cw>;
+
+//===----------------------------------------------------------------------===//
 //  Comparison Instructions...
 //
 
@@ -4022,7 +4083,8 @@ def : ARMPat<(ARMcmpZ GPR:$src, so_reg_reg:$rhs),
 let isCompare = 1, Defs = [CPSR] in {
 def CMNri : AI1<0b1011, (outs), (ins GPR:$Rn, so_imm:$imm), DPFrm, IIC_iCMPi,
                 "cmn", "\t$Rn, $imm",
-                [(ARMcmn GPR:$Rn, so_imm:$imm)]> {
+                [(ARMcmn GPR:$Rn, so_imm:$imm)]>,
+                Sched<[WriteCMP, ReadALU]> {
   bits<4> Rn;
   bits<12> imm;
   let Inst{25} = 1;
@@ -4038,7 +4100,7 @@ def CMNri : AI1<0b1011, (outs), (ins GPR:$Rn, so_imm:$imm), DPFrm, IIC_iCMPi,
 def CMNzrr : AI1<0b1011, (outs), (ins GPR:$Rn, GPR:$Rm), DPFrm, IIC_iCMPr,
                  "cmn", "\t$Rn, $Rm",
                  [(BinOpFrag<(ARMcmpZ node:$LHS,(ineg node:$RHS))>
-                   GPR:$Rn, GPR:$Rm)]> {
+                   GPR:$Rn, GPR:$Rm)]>, Sched<[WriteCMP, ReadALU, ReadALU]> {
   bits<4> Rn;
   bits<4> Rm;
   let isCommutable = 1;
@@ -4056,7 +4118,8 @@ def CMNzrsi : AI1<0b1011, (outs),
                   (ins GPR:$Rn, so_reg_imm:$shift), DPSoRegImmFrm, IIC_iCMPsr,
                   "cmn", "\t$Rn, $shift",
                   [(BinOpFrag<(ARMcmpZ node:$LHS,(ineg node:$RHS))>
-                    GPR:$Rn, so_reg_imm:$shift)]> {
+                    GPR:$Rn, so_reg_imm:$shift)]>,
+                    Sched<[WriteCMPsi, ReadALU]> {
   bits<4> Rn;
   bits<12> shift;
   let Inst{25} = 0;
@@ -4074,7 +4137,8 @@ def CMNzrsr : AI1<0b1011, (outs),
                   (ins GPRnopc:$Rn, so_reg_reg:$shift), DPSoRegRegFrm, IIC_iCMPsr,
                   "cmn", "\t$Rn, $shift",
                   [(BinOpFrag<(ARMcmpZ node:$LHS,(ineg node:$RHS))>
-                    GPRnopc:$Rn, so_reg_reg:$shift)]> {
+                    GPRnopc:$Rn, so_reg_reg:$shift)]>,
+                    Sched<[WriteCMPsr, ReadALU]> {
   bits<4> Rn;
   bits<12> shift;
   let Inst{25} = 0;
@@ -4112,65 +4176,77 @@ let usesCustomInserter = 1, isBranch = 1, isTerminator = 1,
 def BCCi64 : PseudoInst<(outs),
     (ins i32imm:$cc, GPR:$lhs1, GPR:$lhs2, GPR:$rhs1, GPR:$rhs2, brtarget:$dst),
      IIC_Br,
-    [(ARMBcci64 imm:$cc, GPR:$lhs1, GPR:$lhs2, GPR:$rhs1, GPR:$rhs2, bb:$dst)]>;
+    [(ARMBcci64 imm:$cc, GPR:$lhs1, GPR:$lhs2, GPR:$rhs1, GPR:$rhs2, bb:$dst)]>,
+    Sched<[WriteBr]>;
 
 def BCCZi64 : PseudoInst<(outs),
      (ins i32imm:$cc, GPR:$lhs1, GPR:$lhs2, brtarget:$dst), IIC_Br,
-    [(ARMBcci64 imm:$cc, GPR:$lhs1, GPR:$lhs2, 0, 0, bb:$dst)]>;
+    [(ARMBcci64 imm:$cc, GPR:$lhs1, GPR:$lhs2, 0, 0, bb:$dst)]>,
+    Sched<[WriteBr]>;
 } // usesCustomInserter
 
 
 // Conditional moves
-// FIXME: should be able to write a pattern for ARMcmov, but can't use
-// a two-value operand where a dag node expects two operands. :(
 let neverHasSideEffects = 1 in {
 
 let isCommutable = 1, isSelect = 1 in
-def MOVCCr : ARMPseudoInst<(outs GPR:$Rd), (ins GPR:$false, GPR:$Rm, pred:$p),
+def MOVCCr : ARMPseudoInst<(outs GPR:$Rd),
+                           (ins GPR:$false, GPR:$Rm, cmovpred:$p),
                            4, IIC_iCMOVr,
-  [/*(set GPR:$Rd, (ARMcmov GPR:$false, GPR:$Rm, imm:$cc, CCR:$ccr))*/]>,
-      RegConstraint<"$false = $Rd">;
+                           [(set GPR:$Rd, (ARMcmov GPR:$false, GPR:$Rm,
+                                                   cmovpred:$p))]>,
+             RegConstraint<"$false = $Rd">, Sched<[WriteALU]>;
 
 def MOVCCsi : ARMPseudoInst<(outs GPR:$Rd),
-                           (ins GPR:$false, so_reg_imm:$shift, pred:$p),
-                           4, IIC_iCMOVsr,
-  [/*(set GPR:$Rd, (ARMcmov GPR:$false, so_reg_imm:$shift,
-                            imm:$cc, CCR:$ccr))*/]>,
-      RegConstraint<"$false = $Rd">;
+                            (ins GPR:$false, so_reg_imm:$shift, cmovpred:$p),
+                            4, IIC_iCMOVsr,
+                            [(set GPR:$Rd,
+                                  (ARMcmov GPR:$false, so_reg_imm:$shift,
+                                           cmovpred:$p))]>,
+      RegConstraint<"$false = $Rd">, Sched<[WriteALU]>;
 def MOVCCsr : ARMPseudoInst<(outs GPR:$Rd),
-                           (ins GPR:$false, so_reg_reg:$shift, pred:$p),
+                            (ins GPR:$false, so_reg_reg:$shift, cmovpred:$p),
                            4, IIC_iCMOVsr,
-  [/*(set GPR:$Rd, (ARMcmov GPR:$false, so_reg_reg:$shift,
-                            imm:$cc, CCR:$ccr))*/]>,
-      RegConstraint<"$false = $Rd">;
+  [(set GPR:$Rd, (ARMcmov GPR:$false, so_reg_reg:$shift,
+                            cmovpred:$p))]>,
+      RegConstraint<"$false = $Rd">, Sched<[WriteALU]>;
 
 
 let isMoveImm = 1 in
-def MOVCCi16 : ARMPseudoInst<(outs GPR:$Rd),
-                             (ins GPR:$false, imm0_65535_expr:$imm, pred:$p),
-                             4, IIC_iMOVi,
-                             []>,
-      RegConstraint<"$false = $Rd">, Requires<[IsARM, HasV6T2]>;
+def MOVCCi16
+    : ARMPseudoInst<(outs GPR:$Rd),
+                    (ins GPR:$false, imm0_65535_expr:$imm, cmovpred:$p),
+                    4, IIC_iMOVi,
+                    [(set GPR:$Rd, (ARMcmov GPR:$false, imm0_65535:$imm,
+                                            cmovpred:$p))]>,
+      RegConstraint<"$false = $Rd">, Requires<[IsARM, HasV6T2]>,
+      Sched<[WriteALU]>;
 
 let isMoveImm = 1 in
 def MOVCCi : ARMPseudoInst<(outs GPR:$Rd),
-                           (ins GPR:$false, so_imm:$imm, pred:$p),
+                           (ins GPR:$false, so_imm:$imm, cmovpred:$p),
                            4, IIC_iCMOVi,
-   [/*(set GPR:$Rd, (ARMcmov GPR:$false, so_imm:$imm, imm:$cc, CCR:$ccr))*/]>,
-      RegConstraint<"$false = $Rd">;
+                           [(set GPR:$Rd, (ARMcmov GPR:$false, so_imm:$imm,
+                                                   cmovpred:$p))]>,
+      RegConstraint<"$false = $Rd">, Sched<[WriteALU]>;
 
 // Two instruction predicate mov immediate.
 let isMoveImm = 1 in
-def MOVCCi32imm : ARMPseudoInst<(outs GPR:$Rd),
-                                (ins GPR:$false, i32imm:$src, pred:$p),
-                  8, IIC_iCMOVix2, []>, RegConstraint<"$false = $Rd">;
+def MOVCCi32imm
+    : ARMPseudoInst<(outs GPR:$Rd),
+                    (ins GPR:$false, i32imm:$src, cmovpred:$p),
+                    8, IIC_iCMOVix2,
+                    [(set GPR:$Rd, (ARMcmov GPR:$false, imm:$src,
+                                            cmovpred:$p))]>,
+      RegConstraint<"$false = $Rd">, Requires<[IsARM, HasV6T2]>;
 
 let isMoveImm = 1 in
 def MVNCCi : ARMPseudoInst<(outs GPR:$Rd),
-                           (ins GPR:$false, so_imm:$imm, pred:$p),
+                           (ins GPR:$false, so_imm:$imm, cmovpred:$p),
                            4, IIC_iCMOVi,
- [/*(set GPR:$Rd, (ARMcmov GPR:$false, so_imm_not:$imm, imm:$cc, CCR:$ccr))*/]>,
-                RegConstraint<"$false = $Rd">;
+                           [(set GPR:$Rd, (ARMcmov GPR:$false, so_imm_not:$imm,
+                                                   cmovpred:$p))]>,
+                RegConstraint<"$false = $Rd">, Sched<[WriteALU]>;
 
 } // neverHasSideEffects
 
@@ -4189,10 +4265,20 @@ def memb_opt : Operand<i32> {
   let DecoderMethod = "DecodeMemBarrierOption";
 }
 
+def InstSyncBarrierOptOperand : AsmOperandClass {
+  let Name = "InstSyncBarrierOpt";
+  let ParserMethod = "parseInstSyncBarrierOptOperand";
+}
+def instsyncb_opt : Operand<i32> {
+  let PrintMethod = "printInstSyncBOption";
+  let ParserMatchClass = InstSyncBarrierOptOperand;
+  let DecoderMethod = "DecodeInstSyncBarrierOption";
+}
+
 // memory barriers protect the atomic sequences
 let hasSideEffects = 1 in {
 def DMB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
-                "dmb", "\t$opt", [(ARMMemBarrier (i32 imm:$opt))]>,
+                "dmb", "\t$opt", [(int_arm_dmb (i32 imm0_15:$opt))]>,
                 Requires<[IsARM, HasDB]> {
   bits<4> opt;
   let Inst{31-4} = 0xf57ff05;
@@ -4201,7 +4287,7 @@ def DMB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
 }
 
 def DSB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
-                "dsb", "\t$opt", []>,
+                "dsb", "\t$opt", [(int_arm_dsb (i32 imm0_15:$opt))]>,
                 Requires<[IsARM, HasDB]> {
   bits<4> opt;
   let Inst{31-4} = 0xf57ff04;
@@ -4209,7 +4295,7 @@ def DSB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
 }
 
 // ISB has only full system option
-def ISB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
+def ISB : AInoP<(outs), (ins instsyncb_opt:$opt), MiscFrm, NoItinerary,
                 "isb", "\t$opt", []>,
                 Requires<[IsARM, HasDB]> {
   bits<4> opt;
@@ -4217,124 +4303,219 @@ def ISB : AInoP<(outs), (ins memb_opt:$opt), MiscFrm, NoItinerary,
   let Inst{3-0} = opt;
 }
 
+let usesCustomInserter = 1, Defs = [CPSR] in {
+
 // Pseudo instruction that combines movs + predicated rsbmi
 // to implement integer ABS
-let usesCustomInserter = 1, Defs = [CPSR] in
-def ABS : ARMPseudoInst<(outs GPR:$dst), (ins GPR:$src), 8, NoItinerary, []>;
+  def ABS : ARMPseudoInst<(outs GPR:$dst), (ins GPR:$src), 8, NoItinerary, []>;
 
-let usesCustomInserter = 1 in {
-  let Defs = [CPSR] in {
+// Atomic pseudo-insts which will be lowered to ldrex/strex loops.
+// (64-bit pseudos use a hand-written selection code).
+  let mayLoad = 1, mayStore = 1 in {
     def ATOMIC_LOAD_ADD_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_add_8 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_SUB_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_sub_8 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_AND_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_and_8 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_OR_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_or_8 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_XOR_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_xor_8 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_NAND_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_nand_8 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_MIN_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_min_8 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_MAX_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_max_8 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_UMIN_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_umin_8 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_UMAX_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_umax_8 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_SWAP_I8 : PseudoInst<
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$new, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_CMP_SWAP_I8 : PseudoInst<
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$old, GPR:$new, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_ADD_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_add_16 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_SUB_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_sub_16 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_AND_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_and_16 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_OR_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_or_16 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_XOR_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_xor_16 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_NAND_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_nand_16 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_MIN_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_min_16 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_MAX_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_max_16 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_UMIN_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_umin_16 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_UMAX_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_umax_16 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_SWAP_I16 : PseudoInst<
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$new, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_CMP_SWAP_I16 : PseudoInst<
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$old, GPR:$new, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_ADD_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_add_32 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_SUB_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_sub_32 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_AND_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_and_32 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_OR_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_or_32 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_XOR_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_xor_32 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_NAND_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$incr), NoItinerary,
-      [(set GPR:$dst, (atomic_load_nand_32 GPR:$ptr, GPR:$incr))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$incr, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_MIN_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_min_32 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_MAX_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_max_32 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_UMIN_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_umin_32 GPR:$ptr, GPR:$val))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_LOAD_UMAX_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$val), NoItinerary,
-      [(set GPR:$dst, (atomic_load_umax_32 GPR:$ptr, GPR:$val))]>;
-
-    def ATOMIC_SWAP_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$new), NoItinerary,
-      [(set GPR:$dst, (atomic_swap_8 GPR:$ptr, GPR:$new))]>;
-    def ATOMIC_SWAP_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$new), NoItinerary,
-      [(set GPR:$dst, (atomic_swap_16 GPR:$ptr, GPR:$new))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$val, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_SWAP_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$new), NoItinerary,
-      [(set GPR:$dst, (atomic_swap_32 GPR:$ptr, GPR:$new))]>;
-
-    def ATOMIC_CMP_SWAP_I8 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$old, GPR:$new), NoItinerary,
-      [(set GPR:$dst, (atomic_cmp_swap_8 GPR:$ptr, GPR:$old, GPR:$new))]>;
-    def ATOMIC_CMP_SWAP_I16 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$old, GPR:$new), NoItinerary,
-      [(set GPR:$dst, (atomic_cmp_swap_16 GPR:$ptr, GPR:$old, GPR:$new))]>;
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$new, i32imm:$ordering),
+      NoItinerary, []>;
     def ATOMIC_CMP_SWAP_I32 : PseudoInst<
-      (outs GPR:$dst), (ins GPR:$ptr, GPR:$old, GPR:$new), NoItinerary,
-      [(set GPR:$dst, (atomic_cmp_swap_32 GPR:$ptr, GPR:$old, GPR:$new))]>;
-}
+      (outs GPR:$dst),
+      (ins GPR:$ptr, GPR:$old, GPR:$new, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_ADD_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_SUB_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_AND_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_OR_I64 :  PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_XOR_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_NAND_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_MIN_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_MAX_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_UMIN_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_LOAD_UMAX_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_SWAP_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
+    def ATOMIC_CMP_SWAP_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$cmp1, GPR:$cmp2,
+           GPR:$set1, GPR:$set2, i32imm:$ordering),
+      NoItinerary, []>;
+  }
+  let mayLoad = 1 in
+    def ATOMIC_LOAD_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, i32imm:$ordering),
+      NoItinerary, []>;
+  let mayStore = 1 in
+    def ATOMIC_STORE_I64 : PseudoInst<
+      (outs GPR:$dst1, GPR:$dst2),
+      (ins GPR:$addr, GPR:$src1, GPR:$src2, i32imm:$ordering),
+      NoItinerary, []>;
 }
 
 let usesCustomInserter = 1 in {
@@ -4344,48 +4525,147 @@ let usesCustomInserter = 1 in {
       [(ARMcopystructbyval GPR:$dst, GPR:$src, imm:$size, imm:$alignment)]>;
 }
 
+def ldrex_1 : PatFrag<(ops node:$ptr), (int_arm_ldrex node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i8;
+}]>;
+
+def ldrex_2 : PatFrag<(ops node:$ptr), (int_arm_ldrex node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i16;
+}]>;
+
+def ldrex_4 : PatFrag<(ops node:$ptr), (int_arm_ldrex node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i32;
+}]>;
+
+def strex_1 : PatFrag<(ops node:$val, node:$ptr),
+                      (int_arm_strex node:$val, node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i8;
+}]>;
+
+def strex_2 : PatFrag<(ops node:$val, node:$ptr),
+                      (int_arm_strex node:$val, node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i16;
+}]>;
+
+def strex_4 : PatFrag<(ops node:$val, node:$ptr),
+                      (int_arm_strex node:$val, node:$ptr), [{
+  return cast<MemIntrinsicSDNode>(N)->getMemoryVT() == MVT::i32;
+}]>;
+
 let mayLoad = 1 in {
 def LDREXB : AIldrex<0b10, (outs GPR:$Rt), (ins addr_offset_none:$addr),
-                     NoItinerary,
-                    "ldrexb", "\t$Rt, $addr", []>;
+                     NoItinerary, "ldrexb", "\t$Rt, $addr",
+                     [(set GPR:$Rt, (ldrex_1 addr_offset_none:$addr))]>;
 def LDREXH : AIldrex<0b11, (outs GPR:$Rt), (ins addr_offset_none:$addr),
-                     NoItinerary, "ldrexh", "\t$Rt, $addr", []>;
+                     NoItinerary, "ldrexh", "\t$Rt, $addr",
+                     [(set GPR:$Rt, (ldrex_2 addr_offset_none:$addr))]>;
 def LDREX  : AIldrex<0b00, (outs GPR:$Rt), (ins addr_offset_none:$addr),
-                     NoItinerary, "ldrex", "\t$Rt, $addr", []>;
+                     NoItinerary, "ldrex", "\t$Rt, $addr",
+                     [(set GPR:$Rt, (ldrex_4 addr_offset_none:$addr))]>;
 let hasExtraDefRegAllocReq = 1 in
-def LDREXD: AIldrex<0b01, (outs GPRPairOp:$Rt),(ins addr_offset_none:$addr),
+def LDREXD : AIldrex<0b01, (outs GPRPairOp:$Rt),(ins addr_offset_none:$addr),
                       NoItinerary, "ldrexd", "\t$Rt, $addr", []> {
   let DecoderMethod = "DecodeDoubleRegLoad";
 }
+
+def LDAEXB : AIldaex<0b10, (outs GPR:$Rt), (ins addr_offset_none:$addr),
+                     NoItinerary, "ldaexb", "\t$Rt, $addr", []>;
+def LDAEXH : AIldaex<0b11, (outs GPR:$Rt), (ins addr_offset_none:$addr),
+                     NoItinerary, "ldaexh", "\t$Rt, $addr", []>;
+def LDAEX  : AIldaex<0b00, (outs GPR:$Rt), (ins addr_offset_none:$addr),
+                     NoItinerary, "ldaex", "\t$Rt, $addr", []>;
+let hasExtraDefRegAllocReq = 1 in
+def LDAEXD : AIldaex<0b01, (outs GPRPairOp:$Rt),(ins addr_offset_none:$addr),
+                      NoItinerary, "ldaexd", "\t$Rt, $addr", []> {
+  let DecoderMethod = "DecodeDoubleRegLoad";
+}
 }
 
 let mayStore = 1, Constraints = "@earlyclobber $Rd" in {
 def STREXB: AIstrex<0b10, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
-                    NoItinerary, "strexb", "\t$Rd, $Rt, $addr", []>;
+                    NoItinerary, "strexb", "\t$Rd, $Rt, $addr",
+                    [(set GPR:$Rd, (strex_1 GPR:$Rt, addr_offset_none:$addr))]>;
 def STREXH: AIstrex<0b11, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
-                    NoItinerary, "strexh", "\t$Rd, $Rt, $addr", []>;
+                    NoItinerary, "strexh", "\t$Rd, $Rt, $addr",
+                    [(set GPR:$Rd, (strex_2 GPR:$Rt, addr_offset_none:$addr))]>;
 def STREX : AIstrex<0b00, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
-                    NoItinerary, "strex", "\t$Rd, $Rt, $addr", []>;
+                    NoItinerary, "strex", "\t$Rd, $Rt, $addr",
+                    [(set GPR:$Rd, (strex_4 GPR:$Rt, addr_offset_none:$addr))]>;
 let hasExtraSrcRegAllocReq = 1 in
 def STREXD : AIstrex<0b01, (outs GPR:$Rd),
                     (ins GPRPairOp:$Rt, addr_offset_none:$addr),
                     NoItinerary, "strexd", "\t$Rd, $Rt, $addr", []> {
   let DecoderMethod = "DecodeDoubleRegStore";
 }
+def STLEXB: AIstlex<0b10, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
+                    NoItinerary, "stlexb", "\t$Rd, $Rt, $addr",
+                    []>;
+def STLEXH: AIstlex<0b11, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
+                    NoItinerary, "stlexh", "\t$Rd, $Rt, $addr",
+                    []>;
+def STLEX : AIstlex<0b00, (outs GPR:$Rd), (ins GPR:$Rt, addr_offset_none:$addr),
+                    NoItinerary, "stlex", "\t$Rd, $Rt, $addr",
+                    []>;
+let hasExtraSrcRegAllocReq = 1 in
+def STLEXD : AIstlex<0b01, (outs GPR:$Rd),
+                    (ins GPRPairOp:$Rt, addr_offset_none:$addr),
+                    NoItinerary, "stlexd", "\t$Rd, $Rt, $addr", []> {
+  let DecoderMethod = "DecodeDoubleRegStore";
+}
 }
 
-
-def CLREX : AXI<(outs), (ins), MiscFrm, NoItinerary, "clrex", []>,
+def CLREX : AXI<(outs), (ins), MiscFrm, NoItinerary, "clrex",
+                [(int_arm_clrex)]>,
             Requires<[IsARM, HasV7]>  {
   let Inst{31-0} = 0b11110101011111111111000000011111;
 }
 
+def : ARMPat<(and (ldrex_1 addr_offset_none:$addr), 0xff),
+             (LDREXB addr_offset_none:$addr)>;
+def : ARMPat<(and (ldrex_2 addr_offset_none:$addr), 0xffff),
+             (LDREXH addr_offset_none:$addr)>;
+def : ARMPat<(strex_1 (and GPR:$Rt, 0xff), addr_offset_none:$addr),
+             (STREXB GPR:$Rt, addr_offset_none:$addr)>;
+def : ARMPat<(strex_2 (and GPR:$Rt, 0xffff), addr_offset_none:$addr),
+             (STREXH GPR:$Rt, addr_offset_none:$addr)>;
+
+class acquiring_load<PatFrag base>
+  : PatFrag<(ops node:$ptr), (base node:$ptr), [{
+  AtomicOrdering Ordering = cast<AtomicSDNode>(N)->getOrdering();
+  return Ordering == Acquire || Ordering == SequentiallyConsistent;
+}]>;
+
+def atomic_load_acquire_8  : acquiring_load<atomic_load_8>;
+def atomic_load_acquire_16 : acquiring_load<atomic_load_16>;
+def atomic_load_acquire_32 : acquiring_load<atomic_load_32>;
+
+class releasing_store<PatFrag base>
+  : PatFrag<(ops node:$ptr, node:$val), (base node:$ptr, node:$val), [{
+  AtomicOrdering Ordering = cast<AtomicSDNode>(N)->getOrdering();
+  return Ordering == Release || Ordering == SequentiallyConsistent;
+}]>;
+
+def atomic_store_release_8  : releasing_store<atomic_store_8>;
+def atomic_store_release_16 : releasing_store<atomic_store_16>;
+def atomic_store_release_32 : releasing_store<atomic_store_32>;
+
+let AddedComplexity = 8 in {
+  def : ARMPat<(atomic_load_acquire_8 addr_offset_none:$addr),  (LDAB addr_offset_none:$addr)>;
+  def : ARMPat<(atomic_load_acquire_16 addr_offset_none:$addr), (LDAH addr_offset_none:$addr)>;
+  def : ARMPat<(atomic_load_acquire_32 addr_offset_none:$addr), (LDA  addr_offset_none:$addr)>;
+  def : ARMPat<(atomic_store_release_8 addr_offset_none:$addr, GPR:$val),  (STLB GPR:$val, addr_offset_none:$addr)>;
+  def : ARMPat<(atomic_store_release_16 addr_offset_none:$addr, GPR:$val), (STLH GPR:$val, addr_offset_none:$addr)>;
+  def : ARMPat<(atomic_store_release_32 addr_offset_none:$addr, GPR:$val), (STL  GPR:$val, addr_offset_none:$addr)>;
+}
+
 // SWP/SWPB are deprecated in V6/V7.
 let mayLoad = 1, mayStore = 1 in {
 def SWP : AIswp<0, (outs GPRnopc:$Rt),
-                (ins GPRnopc:$Rt2, addr_offset_none:$addr), "swp", []>;
+                (ins GPRnopc:$Rt2, addr_offset_none:$addr), "swp", []>,
+                Requires<[PreV8]>;
 def SWPB: AIswp<1, (outs GPRnopc:$Rt),
-                (ins GPRnopc:$Rt2, addr_offset_none:$addr), "swpb", []>;
+                (ins GPRnopc:$Rt2, addr_offset_none:$addr), "swpb", []>,
+                Requires<[PreV8]>;
 }
 
 //===----------------------------------------------------------------------===//
@@ -4396,7 +4676,8 @@ def CDP : ABI<0b1110, (outs), (ins p_imm:$cop, imm0_15:$opc1,
             c_imm:$CRd, c_imm:$CRn, c_imm:$CRm, imm0_7:$opc2),
             NoItinerary, "cdp", "\t$cop, $opc1, $CRd, $CRn, $CRm, $opc2",
             [(int_arm_cdp imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
-                          imm:$CRm, imm:$opc2)]> {
+                          imm:$CRm, imm:$opc2)]>,
+            Requires<[PreV8]> {
   bits<4> opc1;
   bits<4> CRn;
   bits<4> CRd;
@@ -4413,11 +4694,12 @@ def CDP : ABI<0b1110, (outs), (ins p_imm:$cop, imm0_15:$opc1,
   let Inst{23-20} = opc1;
 }
 
-def CDP2 : ABXI<0b1110, (outs), (ins pf_imm:$cop, imm0_15:$opc1,
+def CDP2 : ABXI<0b1110, (outs), (ins p_imm:$cop, imm0_15:$opc1,
                c_imm:$CRd, c_imm:$CRn, c_imm:$CRm, imm0_7:$opc2),
                NoItinerary, "cdp2\t$cop, $opc1, $CRd, $CRn, $CRm, $opc2",
                [(int_arm_cdp2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
-                              imm:$CRm, imm:$opc2)]> {
+                              imm:$CRm, imm:$opc2)]>,
+               Requires<[PreV8]> {
   let Inst{31-28} = 0b1111;
   bits<4> opc1;
   bits<4> CRn;
@@ -4595,10 +4877,10 @@ defm LDC   : LdStCop <1, 0, "ldc">;
 defm LDCL  : LdStCop <1, 1, "ldcl">;
 defm STC   : LdStCop <0, 0, "stc">;
 defm STCL  : LdStCop <0, 1, "stcl">;
-defm LDC2  : LdSt2Cop<1, 0, "ldc2">;
-defm LDC2L : LdSt2Cop<1, 1, "ldc2l">;
-defm STC2  : LdSt2Cop<0, 0, "stc2">;
-defm STC2L : LdSt2Cop<0, 1, "stc2l">;
+defm LDC2  : LdSt2Cop<1, 0, "ldc2">, Requires<[PreV8]>;
+defm LDC2L : LdSt2Cop<1, 1, "ldc2l">, Requires<[PreV8]>;
+defm STC2  : LdSt2Cop<0, 0, "stc2">, Requires<[PreV8]>;
+defm STC2L : LdSt2Cop<0, 1, "stc2l">, Requires<[PreV8]>;
 
 //===----------------------------------------------------------------------===//
 // Move between coprocessor and ARM core register.
@@ -4631,16 +4913,17 @@ def MCR : MovRCopro<"mcr", 0 /* from ARM core register to coprocessor */,
                     (ins p_imm:$cop, imm0_7:$opc1, GPR:$Rt, c_imm:$CRn,
                          c_imm:$CRm, imm0_7:$opc2),
                     [(int_arm_mcr imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
-                                  imm:$CRm, imm:$opc2)]>;
+                                  imm:$CRm, imm:$opc2)]>,
+                    ComplexDeprecationPredicate<"MCR">;
 def : ARMInstAlias<"mcr${p} $cop, $opc1, $Rt, $CRn, $CRm",
                    (MCR p_imm:$cop, imm0_7:$opc1, GPR:$Rt, c_imm:$CRn,
                         c_imm:$CRm, 0, pred:$p)>;
 def MRC : MovRCopro<"mrc", 1 /* from coprocessor to ARM core register */,
-                    (outs GPR:$Rt),
+                    (outs GPRwithAPSR:$Rt),
                     (ins p_imm:$cop, imm0_7:$opc1, c_imm:$CRn, c_imm:$CRm,
                          imm0_7:$opc2), []>;
 def : ARMInstAlias<"mrc${p} $cop, $opc1, $Rt, $CRn, $CRm",
-                   (MRC GPR:$Rt, p_imm:$cop, imm0_7:$opc1, c_imm:$CRn,
+                   (MRC GPRwithAPSR:$Rt, p_imm:$cop, imm0_7:$opc1, c_imm:$CRn,
                         c_imm:$CRm, 0, pred:$p)>;
 
 def : ARMPat<(int_arm_mrc imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2),
@@ -4650,7 +4933,7 @@ class MovRCopro2<string opc, bit direction, dag oops, dag iops,
                  list<dag> pattern>
   : ABXI<0b1110, oops, iops, NoItinerary,
          !strconcat(opc, "\t$cop, $opc1, $Rt, $CRn, $CRm, $opc2"), pattern> {
-  let Inst{31-28} = 0b1111;
+  let Inst{31-24} = 0b11111110;
   let Inst{20} = direction;
   let Inst{4} = 1;
 
@@ -4674,16 +4957,18 @@ def MCR2 : MovRCopro2<"mcr2", 0 /* from ARM core register to coprocessor */,
                       (ins p_imm:$cop, imm0_7:$opc1, GPR:$Rt, c_imm:$CRn,
                            c_imm:$CRm, imm0_7:$opc2),
                       [(int_arm_mcr2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
-                                     imm:$CRm, imm:$opc2)]>;
+                                     imm:$CRm, imm:$opc2)]>,
+                      Requires<[PreV8]>;
 def : ARMInstAlias<"mcr2$ $cop, $opc1, $Rt, $CRn, $CRm",
                    (MCR2 p_imm:$cop, imm0_7:$opc1, GPR:$Rt, c_imm:$CRn,
                          c_imm:$CRm, 0)>;
 def MRC2 : MovRCopro2<"mrc2", 1 /* from coprocessor to ARM core register */,
-                      (outs GPR:$Rt),
+                      (outs GPRwithAPSR:$Rt),
                       (ins p_imm:$cop, imm0_7:$opc1, c_imm:$CRn, c_imm:$CRm,
-                           imm0_7:$opc2), []>;
+                           imm0_7:$opc2), []>,
+                      Requires<[PreV8]>;
 def : ARMInstAlias<"mrc2$ $cop, $opc1, $Rt, $CRn, $CRm",
-                   (MRC2 GPR:$Rt, p_imm:$cop, imm0_7:$opc1, c_imm:$CRn,
+                   (MRC2 GPRwithAPSR:$Rt, p_imm:$cop, imm0_7:$opc1, c_imm:$CRn,
                          c_imm:$CRm, 0)>;
 
 def : ARMV5TPat<(int_arm_mrc2 imm:$cop, imm:$opc1, imm:$CRn,
@@ -4718,7 +5003,8 @@ def MRRC : MovRRCopro<"mrrc", 1 /* from coprocessor to ARM core register */>;
 class MovRRCopro2<string opc, bit direction, list<dag> pattern = []>
   : ABXI<0b1100, (outs), (ins p_imm:$cop, imm0_15:$opc1,
          GPRnopc:$Rt, GPRnopc:$Rt2, c_imm:$CRm), NoItinerary,
-         !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"), pattern> {
+         !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"), pattern>,
+    Requires<[PreV8]> {
   let Inst{31-28} = 0b1111;
   let Inst{23-21} = 0b010;
   let Inst{20} = direction;
@@ -4820,7 +5106,7 @@ def MSRi : ABI<0b0011, (outs), (ins msr_mask:$mask,  so_imm:$a), NoItinerary,
 let isCall = 1,
   Defs = [R0, R12, LR, CPSR], Uses = [SP] in {
   def TPsoft : PseudoInst<(outs), (ins), IIC_Br,
-               [(set R0, ARMthread_pointer)]>;
+               [(set R0, ARMthread_pointer)]>, Sched<[WriteBr]>;
 }
 
 //===----------------------------------------------------------------------===//
@@ -4884,7 +5170,7 @@ let isBranch = 1, isTerminator = 1, isBarrier = 1, isIndirectBranch = 1 in
   def MOVPCRX : ARMPseudoExpand<(outs), (ins GPR:$dst),
                     4, IIC_Br, [(brind GPR:$dst)],
                     (MOVr PC, GPR:$dst, (ops 14, zero_reg), zero_reg)>,
-                  Requires<[IsARM, NoV4T]>;
+                  Requires<[IsARM, NoV4T]>, Sched<[WriteBr]>;
 
 // Large immediate handling.
 
@@ -5153,10 +5439,10 @@ def : MnemonicAlias<"rfeed", "rfeib">;
 def : MnemonicAlias<"rfe", "rfeia">;
 
 // SRS aliases
-def : MnemonicAlias<"srsfa", "srsda">;
-def : MnemonicAlias<"srsea", "srsdb">;
-def : MnemonicAlias<"srsfd", "srsia">;
-def : MnemonicAlias<"srsed", "srsib">;
+def : MnemonicAlias<"srsfa", "srsib">;
+def : MnemonicAlias<"srsea", "srsia">;
+def : MnemonicAlias<"srsfd", "srsdb">;
+def : MnemonicAlias<"srsed", "srsda">;
 def : MnemonicAlias<"srs", "srsia">;
 
 // QSAX == QSUBADDX
@@ -5233,7 +5519,7 @@ def RORi : ARMAsmPseudo<"ror${s}${p} $Rd, $Rm, $imm",
                              cc_out:$s)>;
 }
 def RRXi : ARMAsmPseudo<"rrx${s}${p} $Rd, $Rm",
-                        (ins GPRnopc:$Rd, GPRnopc:$Rm, pred:$p, cc_out:$s)>;
+                        (ins GPR:$Rd, GPR:$Rm, pred:$p, cc_out:$s)>;
 let TwoOperandAliasConstraint = "$Rn = $Rd" in {
 def ASRr : ARMAsmPseudo<"asr${s}${p} $Rd, $Rn, $Rm",
                         (ins GPRnopc:$Rd, GPRnopc:$Rn, GPRnopc:$Rm, pred:$p,
@@ -5269,4 +5555,5 @@ def : InstAlias<"umull${s}${p} $RdLo, $RdHi, $Rn, $Rm",
 
 // 'it' blocks in ARM mode just validate the predicates. The IT itself
 // is discarded.
-def ITasm : ARMAsmPseudo<"it$mask $cc", (ins it_pred:$cc, it_mask:$mask)>;
+def ITasm : ARMAsmPseudo<"it$mask $cc", (ins it_pred:$cc, it_mask:$mask)>,
+         ComplexDeprecationPredicate<"IT">;