6 files changed, 27 insertions, 34 deletions

diff --git a/contrib/llvm-project/llvm/lib/Target/AArch64/AArch64InstrInfo.td b/contrib/llvm-project/llvm/lib/Target/AArch64/AArch64InstrInfo.td
index 03baa7497615..ac61dd8745d4 100644
--- a/contrib/llvm-project/llvm/lib/Target/AArch64/AArch64InstrInfo.td
+++ b/contrib/llvm-project/llvm/lib/Target/AArch64/AArch64InstrInfo.td
@@ -4885,19 +4885,9 @@ defm UABDL   : SIMDLongThreeVectorBHSabdl<1, 0b0111, "uabdl",
 def : Pat<(abs (v8i16 (sub (zext (v8i8 V64:$opA)),
                            (zext (v8i8 V64:$opB))))),
           (UABDLv8i8_v8i16 V64:$opA, V64:$opB)>;
-def : Pat<(xor (v8i16 (AArch64vashr v8i16:$src, (i32 15))),
-               (v8i16 (add (sub (zext (v8i8 V64:$opA)),
-                                (zext (v8i8 V64:$opB))),
-                           (AArch64vashr v8i16:$src, (i32 15))))),
-          (UABDLv8i8_v8i16 V64:$opA, V64:$opB)>;
 def : Pat<(abs (v8i16 (sub (zext (extract_high_v16i8 (v16i8 V128:$opA))),
                            (zext (extract_high_v16i8 (v16i8 V128:$opB)))))),
           (UABDLv16i8_v8i16 V128:$opA, V128:$opB)>;
-def : Pat<(xor (v8i16 (AArch64vashr v8i16:$src, (i32 15))),
-               (v8i16 (add (sub (zext (extract_high_v16i8 (v16i8 V128:$opA))),
-                                (zext (extract_high_v16i8 (v16i8 V128:$opB)))),
-                           (AArch64vashr v8i16:$src, (i32 15))))),
-          (UABDLv16i8_v8i16 V128:$opA, V128:$opB)>;
 def : Pat<(abs (v4i32 (sub (zext (v4i16 V64:$opA)),
                            (zext (v4i16 V64:$opB))))),
           (UABDLv4i16_v4i32 V64:$opA, V64:$opB)>;
diff --git a/contrib/llvm-project/llvm/lib/Target/AMDGPU/SIMemoryLegalizer.cpp b/contrib/llvm-project/llvm/lib/Target/AMDGPU/SIMemoryLegalizer.cpp
index 84b9330ef963..50d8bfa87508 100644
--- a/contrib/llvm-project/llvm/lib/Target/AMDGPU/SIMemoryLegalizer.cpp
+++ b/contrib/llvm-project/llvm/lib/Target/AMDGPU/SIMemoryLegalizer.cpp
@@ -2358,6 +2358,11 @@ bool SIGfx12CacheControl::enableVolatileAndOrNonTemporal(
 
   bool Changed = false;
 
+  if (IsNonTemporal) {
+    // Set non-temporal hint for all cache levels.
+    Changed |= setTH(MI, AMDGPU::CPol::TH_NT);
+  }
+
   if (IsVolatile) {
     Changed |= setScope(MI, AMDGPU::CPol::SCOPE_SYS);
 
@@ -2370,11 +2375,6 @@ bool SIGfx12CacheControl::enableVolatileAndOrNonTemporal(
                           Position::AFTER);
   }
 
-  if (IsNonTemporal) {
-    // Set non-temporal hint for all cache levels.
-    Changed |= setTH(MI, AMDGPU::CPol::TH_NT);
-  }
-
   return Changed;
 }
 
diff --git a/contrib/llvm-project/llvm/lib/Target/RISCV/RISCVISelLowering.cpp b/contrib/llvm-project/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
index a0cec426002b..d46093b9e260 100644
--- a/contrib/llvm-project/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
+++ b/contrib/llvm-project/llvm/lib/Target/RISCV/RISCVISelLowering.cpp
@@ -14559,7 +14559,7 @@ static SDValue tryFoldSelectIntoOp(SDNode *N, SelectionDAG &DAG,
   EVT VT = N->getValueType(0);
   SDLoc DL(N);
   SDValue OtherOp = TrueVal.getOperand(1 - OpToFold);
-  EVT OtherOpVT = OtherOp->getValueType(0);
+  EVT OtherOpVT = OtherOp.getValueType();
   SDValue IdentityOperand =
       DAG.getNeutralElement(Opc, DL, OtherOpVT, N->getFlags());
   if (!Commutative)
diff --git a/contrib/llvm-project/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp b/contrib/llvm-project/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
index 833f058253d8..553d338b7790 100644
--- a/contrib/llvm-project/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
+++ b/contrib/llvm-project/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
@@ -2923,11 +2923,10 @@ bool X86DAGToDAGISel::selectAddr(SDNode *Parent, SDValue N, SDValue &Base,
 }
 
 bool X86DAGToDAGISel::selectMOV64Imm32(SDValue N, SDValue &Imm) {
-  // Cannot use 32 bit constants to reference objects in kernel code model.
-  // Cannot use 32 bit constants to reference objects in large PIC mode since
-  // GOTOFF is 64 bits.
+  // Cannot use 32 bit constants to reference objects in kernel/large code
+  // model.
   if (TM.getCodeModel() == CodeModel::Kernel ||
-      (TM.getCodeModel() == CodeModel::Large && TM.isPositionIndependent()))
+      TM.getCodeModel() == CodeModel::Large)
     return false;
 
   // In static codegen with small code model, we can get the address of a label
diff --git a/contrib/llvm-project/llvm/lib/Target/X86/X86ISelLowering.cpp b/contrib/llvm-project/llvm/lib/Target/X86/X86ISelLowering.cpp
index 96bbd981ff24..201dd8382536 100644
--- a/contrib/llvm-project/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/contrib/llvm-project/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -47038,10 +47038,13 @@ static SDValue combineShiftRightArithmetic(SDNode *N, SelectionDAG &DAG,
   if (SDValue V = combineShiftToPMULH(N, DAG, Subtarget))
     return V;
 
-  // fold (ashr (shl, a, [56,48,32,24,16]), SarConst)
-  // into (shl, (sext (a), [56,48,32,24,16] - SarConst)) or
-  // into (lshr, (sext (a), SarConst - [56,48,32,24,16]))
-  // depending on sign of (SarConst - [56,48,32,24,16])
+  // fold (SRA (SHL X, ShlConst), SraConst)
+  // into (SHL (sext_in_reg X), ShlConst - SraConst)
+  //   or (sext_in_reg X)
+  //   or (SRA (sext_in_reg X), SraConst - ShlConst)
+  // depending on relation between SraConst and ShlConst.
+  // We only do this if (Size - ShlConst) is equal to 8, 16 or 32. That allows
+  // us to do the sext_in_reg from corresponding bit.
 
   // sexts in X86 are MOVs. The MOVs have the same code size
   // as above SHIFTs (only SHIFT on 1 has lower code size).
@@ -47057,29 +47060,29 @@ static SDValue combineShiftRightArithmetic(SDNode *N, SelectionDAG &DAG,
   SDValue N00 = N0.getOperand(0);
   SDValue N01 = N0.getOperand(1);
   APInt ShlConst = N01->getAsAPIntVal();
-  APInt SarConst = N1->getAsAPIntVal();
+  APInt SraConst = N1->getAsAPIntVal();
   EVT CVT = N1.getValueType();
 
-  if (SarConst.isNegative())
+  if (CVT != N01.getValueType())
+    return SDValue();
+  if (SraConst.isNegative())
     return SDValue();
 
   for (MVT SVT : { MVT::i8, MVT::i16, MVT::i32 }) {
     unsigned ShiftSize = SVT.getSizeInBits();
-    // skipping types without corresponding sext/zext and
-    // ShlConst that is not one of [56,48,32,24,16]
+    // Only deal with (Size - ShlConst) being equal to 8, 16 or 32.
     if (ShiftSize >= Size || ShlConst != Size - ShiftSize)
       continue;
     SDLoc DL(N);
     SDValue NN =
         DAG.getNode(ISD::SIGN_EXTEND_INREG, DL, VT, N00, DAG.getValueType(SVT));
-    SarConst = SarConst - (Size - ShiftSize);
-    if (SarConst == 0)
+    if (SraConst.eq(ShlConst))
       return NN;
-    if (SarConst.isNegative())
+    if (SraConst.ult(ShlConst))
       return DAG.getNode(ISD::SHL, DL, VT, NN,
-                         DAG.getConstant(-SarConst, DL, CVT));
+                         DAG.getConstant(ShlConst - SraConst, DL, CVT));
     return DAG.getNode(ISD::SRA, DL, VT, NN,
-                       DAG.getConstant(SarConst, DL, CVT));
+                       DAG.getConstant(SraConst - ShlConst, DL, CVT));
   }
   return SDValue();
 }
diff --git a/contrib/llvm-project/llvm/lib/Target/X86/X86Subtarget.h b/contrib/llvm-project/llvm/lib/Target/X86/X86Subtarget.h
index a458b5f9ec8f..4d55a084b730 100644
--- a/contrib/llvm-project/llvm/lib/Target/X86/X86Subtarget.h
+++ b/contrib/llvm-project/llvm/lib/Target/X86/X86Subtarget.h
@@ -244,7 +244,8 @@ public:
   // TODO: Currently we're always allowing widening on CPUs without VLX,
   // because for many cases we don't have a better option.
   bool canExtendTo512DQ() const {
-    return hasAVX512() && (!hasVLX() || getPreferVectorWidth() >= 512);
+    return hasAVX512() && hasEVEX512() &&
+           (!hasVLX() || getPreferVectorWidth() >= 512);
   }
   bool canExtendTo512BW() const  {
     return hasBWI() && canExtendTo512DQ();