[AMDGPU] Asm/disasm v_cndmask_b32_e64 with abs/neg source modifiers

This commit allows v_cndmask_b32_e64 with abs, neg source
modifiers on src0, src1 to be assembled and disassembled.

This does appear to be allowed, even though they are floating point
modifiers and the operand type is b32.

To do this, I added src0_modifiers and src1_modifiers to the
MachineInstr, which involved fixing up several places in codegen and mir
tests.

Differential Revision: https://reviews.llvm.org/D59191

Change-Id: I69bf4a8c73ebc65744f6110bb8fc4e937d79fbea
llvm-svn: 356398

diff --git a/llvm/lib/Target/AMDGPU/SIFoldOperands.cpp b/llvm/lib/Target/AMDGPU/SIFoldOperands.cpp
index 9361b25..4a82d3a 100644 (file)
--- a/llvm/lib/Target/AMDGPU/SIFoldOperands.cpp
+++ b/llvm/lib/Target/AMDGPU/SIFoldOperands.cpp
@@ -762,14 +762,23 @@ static bool tryFoldInst(const SIInstrInfo *TII,
       Opc == AMDGPU::V_CNDMASK_B64_PSEUDO) {
     const MachineOperand *Src0 = TII->getNamedOperand(*MI, AMDGPU::OpName::src0);
     const MachineOperand *Src1 = TII->getNamedOperand(*MI, AMDGPU::OpName::src1);
-    if (Src1->isIdenticalTo(*Src0)) {
+    int Src1ModIdx = AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src1_modifiers);
+    int Src0ModIdx = AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src0_modifiers);
+    if (Src1->isIdenticalTo(*Src0) &&
+        (Src1ModIdx == -1 || !MI->getOperand(Src1ModIdx).getImm()) &&
+        (Src0ModIdx == -1 || !MI->getOperand(Src0ModIdx).getImm())) {
       LLVM_DEBUG(dbgs() << "Folded " << *MI << " into ");
+      auto &NewDesc =
+          TII->get(Src0->isReg() ? (unsigned)AMDGPU::COPY : getMovOpc(false));
       int Src2Idx = AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src2);
       if (Src2Idx != -1)
         MI->RemoveOperand(Src2Idx);
       MI->RemoveOperand(AMDGPU::getNamedOperandIdx(Opc, AMDGPU::OpName::src1));
-      mutateCopyOp(*MI, TII->get(Src0->isReg() ? (unsigned)AMDGPU::COPY
-                                               : getMovOpc(false)));
+      if (Src1ModIdx != -1)
+        MI->RemoveOperand(Src1ModIdx);
+      if (Src0ModIdx != -1)
+        MI->RemoveOperand(Src0ModIdx);
+      mutateCopyOp(*MI, NewDesc);
       LLVM_DEBUG(dbgs() << *MI << '\n');
       return true;
     }

diff --git a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
index 6f48a2e..38f27c5 100644 (file)
--- a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
+++ b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
@@ -3437,11 +3437,15 @@ MachineBasicBlock *SITargetLowering::EmitInstrWithCustomInserter(
     BuildMI(*BB, MI, DL, TII->get(AMDGPU::COPY), SrcCondCopy)
       .addReg(SrcCond);
     BuildMI(*BB, MI, DL, TII->get(AMDGPU::V_CNDMASK_B32_e64), DstLo)
+      .addImm(0)
       .addReg(Src0, 0, AMDGPU::sub0)
+      .addImm(0)
       .addReg(Src1, 0, AMDGPU::sub0)
       .addReg(SrcCondCopy);
     BuildMI(*BB, MI, DL, TII->get(AMDGPU::V_CNDMASK_B32_e64), DstHi)
+      .addImm(0)
       .addReg(Src0, 0, AMDGPU::sub1)
+      .addImm(0)
       .addReg(Src1, 0, AMDGPU::sub1)
       .addReg(SrcCondCopy);
 

diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
index 0d90309..196ecd7 100644 (file)
--- a/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/SIInstrInfo.cpp
@@ -693,7 +693,9 @@ void SIInstrInfo::insertVectorSelect(MachineBasicBlock &MBB,
     BuildMI(MBB, I, DL, get(AMDGPU::COPY), SReg)
       .add(Cond[0]);
     BuildMI(MBB, I, DL, get(AMDGPU::V_CNDMASK_B32_e64), DstReg)
+      .addImm(0)
       .addReg(FalseReg)
+      .addImm(0)
       .addReg(TrueReg)
       .addReg(SReg);
   } else if (Cond.size() == 2) {
@@ -705,7 +707,9 @@ void SIInstrInfo::insertVectorSelect(MachineBasicBlock &MBB,
         .addImm(-1)
         .addImm(0);
       BuildMI(MBB, I, DL, get(AMDGPU::V_CNDMASK_B32_e64), DstReg)
+        .addImm(0)
         .addReg(FalseReg)
+        .addImm(0)
         .addReg(TrueReg)
         .addReg(SReg);
       break;
@@ -716,7 +720,9 @@ void SIInstrInfo::insertVectorSelect(MachineBasicBlock &MBB,
         .addImm(0)
         .addImm(-1);
       BuildMI(MBB, I, DL, get(AMDGPU::V_CNDMASK_B32_e64), DstReg)
+        .addImm(0)
         .addReg(FalseReg)
+        .addImm(0)
         .addReg(TrueReg)
         .addReg(SReg);
       break;
@@ -728,7 +734,9 @@ void SIInstrInfo::insertVectorSelect(MachineBasicBlock &MBB,
       BuildMI(MBB, I, DL, get(AMDGPU::COPY), SReg)
         .add(RegOp);
       BuildMI(MBB, I, DL, get(AMDGPU::V_CNDMASK_B32_e64), DstReg)
+          .addImm(0)
           .addReg(FalseReg)
+          .addImm(0)
           .addReg(TrueReg)
           .addReg(SReg);
       break;
@@ -740,7 +748,9 @@ void SIInstrInfo::insertVectorSelect(MachineBasicBlock &MBB,
       BuildMI(MBB, I, DL, get(AMDGPU::COPY), SReg)
         .add(RegOp);
       BuildMI(MBB, I, DL, get(AMDGPU::V_CNDMASK_B32_e64), DstReg)
+          .addImm(0)
           .addReg(TrueReg)
+          .addImm(0)
           .addReg(FalseReg)
           .addReg(SReg);
       break;
@@ -754,7 +764,9 @@ void SIInstrInfo::insertVectorSelect(MachineBasicBlock &MBB,
         .addImm(-1)
         .addImm(0);
       BuildMI(MBB, I, DL, get(AMDGPU::V_CNDMASK_B32_e64), DstReg)
+        .addImm(0)
         .addReg(FalseReg)
+        .addImm(0)
         .addReg(TrueReg)
         .addReg(SReg);
       break;
@@ -768,7 +780,9 @@ void SIInstrInfo::insertVectorSelect(MachineBasicBlock &MBB,
         .addImm(0)
         .addImm(-1);
       BuildMI(MBB, I, DL, get(AMDGPU::V_CNDMASK_B32_e64), DstReg)
+        .addImm(0)
         .addReg(FalseReg)
+        .addImm(0)
         .addReg(TrueReg)
         .addReg(SReg);
       llvm_unreachable("Unhandled branch predicate EXECZ");
@@ -2579,7 +2593,8 @@ bool SIInstrInfo::canShrink(const MachineInstr &MI,
   // Can't shrink instruction with three operands.
   // FIXME: v_cndmask_b32 has 3 operands and is shrinkable, but we need to add
   // a special case for it.  It can only be shrunk if the third operand
-  // is vcc.  We should handle this the same way we handle vopc, by addding
+  // is vcc, and src0_modifiers and src1_modifiers are not set.
+  // We should handle this the same way we handle vopc, by addding
   // a register allocation hint pre-regalloc and then do the shrinking
   // post-regalloc.
   if (Src2) {

diff --git a/llvm/lib/Target/AMDGPU/SIInstrInfo.td b/llvm/lib/Target/AMDGPU/SIInstrInfo.td
index c3d9ff7..680c287 100644 (file)
--- a/llvm/lib/Target/AMDGPU/SIInstrInfo.td
+++ b/llvm/lib/Target/AMDGPU/SIInstrInfo.td
@@ -1283,7 +1283,7 @@ class isModifierType<ValueType SrcVT> {
 }
 
 // Return type of input modifiers operand for specified input operand
-class getSrcMod <ValueType VT> {
+class getSrcMod <ValueType VT, bit EnableF32SrcMods> {
   bit isFP = !if(!eq(VT.Value, f16.Value), 1,
                !if(!eq(VT.Value, f32.Value), 1,
                !if(!eq(VT.Value, f64.Value), 1,
@@ -1296,7 +1296,7 @@ class getSrcMod <ValueType VT> {
                             FP16InputMods,
                             FP32InputMods
                           ),
-                         Int32InputMods)
+                         !if(EnableF32SrcMods, FP32InputMods, Int32InputMods))
                      );
 }
 
@@ -1331,7 +1331,7 @@ class getIns32 <RegisterOperand Src0RC, RegisterClass Src1RC, int NumSrcArgs> {
 // Returns the input arguments for VOP3 instructions for the given SrcVT.
 class getIns64 <RegisterOperand Src0RC, RegisterOperand Src1RC,
                 RegisterOperand Src2RC, int NumSrcArgs,
-                bit HasIntClamp, bit HasModifiers, bit HasOMod,
+                bit HasIntClamp, bit HasModifiers, bit HasSrc2Mods, bit HasOMod,
                 Operand Src0Mod, Operand Src1Mod, Operand Src2Mod> {
 
   dag ret =
@@ -1369,16 +1369,33 @@ class getIns64 <RegisterOperand Src0RC, RegisterOperand Src1RC,
       /* endif */ )
     /* NumSrcArgs == 3 */,
       !if (!eq(HasModifiers, 1),
-        // VOP3 with modifiers
-        !if (!eq(HasOMod, 1),
-          (ins Src0Mod:$src0_modifiers, Src0RC:$src0,
-               Src1Mod:$src1_modifiers, Src1RC:$src1,
-               Src2Mod:$src2_modifiers, Src2RC:$src2,
-               clampmod:$clamp, omod:$omod),
-          (ins Src0Mod:$src0_modifiers, Src0RC:$src0,
-               Src1Mod:$src1_modifiers, Src1RC:$src1,
-               Src2Mod:$src2_modifiers, Src2RC:$src2,
-               clampmod:$clamp))
+        !if (!eq(HasSrc2Mods, 1),
+          // VOP3 with modifiers
+          !if (!eq(HasOMod, 1),
+            (ins Src0Mod:$src0_modifiers, Src0RC:$src0,
+                 Src1Mod:$src1_modifiers, Src1RC:$src1,
+                 Src2Mod:$src2_modifiers, Src2RC:$src2,
+                 clampmod:$clamp, omod:$omod),
+            !if (!eq(HasIntClamp, 1),
+              (ins Src0Mod:$src0_modifiers, Src0RC:$src0,
+                   Src1Mod:$src1_modifiers, Src1RC:$src1,
+                   Src2Mod:$src2_modifiers, Src2RC:$src2,
+                   clampmod:$clamp),
+              (ins Src0Mod:$src0_modifiers, Src0RC:$src0,
+                   Src1Mod:$src1_modifiers, Src1RC:$src1,
+                   Src2Mod:$src2_modifiers, Src2RC:$src2))),
+          // VOP3 with modifiers except src2
+          !if (!eq(HasOMod, 1),
+            (ins Src0Mod:$src0_modifiers, Src0RC:$src0,
+                 Src1Mod:$src1_modifiers, Src1RC:$src1,
+                 Src2RC:$src2, clampmod:$clamp, omod:$omod),
+            !if (!eq(HasIntClamp, 1),
+              (ins Src0Mod:$src0_modifiers, Src0RC:$src0,
+                   Src1Mod:$src1_modifiers, Src1RC:$src1,
+                   Src2RC:$src2, clampmod:$clamp),
+              (ins Src0Mod:$src0_modifiers, Src0RC:$src0,
+                   Src1Mod:$src1_modifiers, Src1RC:$src1,
+                   Src2RC:$src2))))
       /* else */,
         // VOP3 without modifiers
         !if (!eq(HasIntClamp, 1),
@@ -1743,9 +1760,10 @@ def PatGenMode {
   int Pattern   = 1;
 }
 
-class VOPProfile <list<ValueType> _ArgVT> {
+class VOPProfile <list<ValueType> _ArgVT, bit _EnableF32SrcMods = 0> {
 
   field list<ValueType> ArgVT = _ArgVT;
+  field bit EnableF32SrcMods = _EnableF32SrcMods;
 
   field ValueType DstVT = ArgVT[0];
   field ValueType Src0VT = ArgVT[1];
@@ -1763,9 +1781,9 @@ class VOPProfile <list<ValueType> _ArgVT> {
   field RegisterClass Src1DPP = getVregSrcForVT<Src1VT>.ret;
   field RegisterOperand Src0SDWA = getSDWASrcForVT<Src0VT>.ret;
   field RegisterOperand Src1SDWA = getSDWASrcForVT<Src0VT>.ret;
-  field Operand Src0Mod = getSrcMod<Src0VT>.ret;
-  field Operand Src1Mod = getSrcMod<Src1VT>.ret;
-  field Operand Src2Mod = getSrcMod<Src2VT>.ret;
+  field Operand Src0Mod = getSrcMod<Src0VT, EnableF32SrcMods>.ret;
+  field Operand Src1Mod = getSrcMod<Src1VT, EnableF32SrcMods>.ret;
+  field Operand Src2Mod = getSrcMod<Src2VT, EnableF32SrcMods>.ret;
   field Operand Src0ModDPP = getSrcModExt<Src0VT>.ret;
   field Operand Src1ModDPP = getSrcModExt<Src1VT>.ret;
   field Operand Src0ModSDWA = getSrcModSDWA<Src0VT>.ret;
@@ -1781,12 +1799,16 @@ class VOPProfile <list<ValueType> _ArgVT> {
   field bit HasSrc2 = !if(!eq(Src2VT.Value, untyped.Value), 0, 1);
 
   // TODO: Modifiers logic is somewhat adhoc here, to be refined later
-  field bit HasModifiers = isModifierType<Src0VT>.ret;
+  // HasModifiers affects the normal and DPP encodings. We take note of EnableF32SrcMods, which
+  // enables modifiers for i32 type.
+  field bit HasModifiers = BitOr<isModifierType<Src0VT>.ret, EnableF32SrcMods>.ret;
 
+  // HasSrc*FloatMods affects the SDWA encoding. We ignore EnableF32SrcMods.
   field bit HasSrc0FloatMods = isFloatType<Src0VT>.ret;
   field bit HasSrc1FloatMods = isFloatType<Src1VT>.ret;
   field bit HasSrc2FloatMods = isFloatType<Src2VT>.ret;
 
+  // HasSrc*IntMods affects the SDWA encoding. We ignore EnableF32SrcMods.
   field bit HasSrc0IntMods = isIntType<Src0VT>.ret;
   field bit HasSrc1IntMods = isIntType<Src1VT>.ret;
   field bit HasSrc2IntMods = isIntType<Src2VT>.ret;
@@ -1795,7 +1817,7 @@ class VOPProfile <list<ValueType> _ArgVT> {
   field bit HasSrc1Mods = !if(HasModifiers, BitOr<HasSrc1FloatMods, HasSrc1IntMods>.ret, 0);
   field bit HasSrc2Mods = !if(HasModifiers, BitOr<HasSrc2FloatMods, HasSrc2IntMods>.ret, 0);
 
-  field bit HasClamp = HasModifiers;
+  field bit HasClamp = isModifierType<Src0VT>.ret;
   field bit HasSDWAClamp = EmitDst;
   field bit HasFPClamp = BitAnd<isFloatType<DstVT>.ret, HasClamp>.ret;
   field bit HasIntClamp = !if(isFloatType<DstVT>.ret, 0, HasClamp);
@@ -1829,8 +1851,8 @@ class VOPProfile <list<ValueType> _ArgVT> {
 
   field dag Ins32 = getIns32<Src0RC32, Src1RC32, NumSrcArgs>.ret;
   field dag Ins64 = getIns64<Src0RC64, Src1RC64, Src2RC64, NumSrcArgs,
-                             HasIntClamp, HasModifiers, HasOMod, Src0Mod, Src1Mod,
-                             Src2Mod>.ret;
+                             HasIntClamp, HasModifiers, HasSrc2Mods,
+                             HasOMod, Src0Mod, Src1Mod, Src2Mod>.ret;
   field dag InsVOP3P = getInsVOP3P<Src0RC64, Src1RC64, Src2RC64,
                                    NumSrcArgs, HasClamp,
                                    Src0PackedMod, Src1PackedMod, Src2PackedMod>.ret;

diff --git a/llvm/lib/Target/AMDGPU/SIInstructions.td b/llvm/lib/Target/AMDGPU/SIInstructions.td
index 88acebb..a2b6f76 100644 (file)
--- a/llvm/lib/Target/AMDGPU/SIInstructions.td
+++ b/llvm/lib/Target/AMDGPU/SIInstructions.td
@@ -698,7 +698,7 @@ def : FMADModsPat<V_MAD_F16, AMDGPUfmad_ftz, f16> {
 multiclass SelectPat <ValueType vt, Instruction inst> {
   def : GCNPat <
     (vt (select i1:$src0, vt:$src1, vt:$src2)),
-    (inst $src2, $src1, $src0)
+    (inst (i32 0), $src2, (i32 0), $src1, $src0)
   >;
 }
 
@@ -1104,12 +1104,14 @@ def : POW_Common <V_LOG_F32_e32, V_EXP_F32_e32, V_MUL_LEGACY_F32_e32>;
 
 def : GCNPat <
   (i32 (sext i1:$src0)),
-  (V_CNDMASK_B32_e64 (i32 0), (i32 -1), $src0)
+  (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                     /*src1mod*/(i32 0), /*src1*/(i32 -1), $src0)
 >;
 
 class Ext32Pat <SDNode ext> : GCNPat <
   (i32 (ext i1:$src0)),
-  (V_CNDMASK_B32_e64 (i32 0), (i32 1), $src0)
+  (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                     /*src1mod*/(i32 0), /*src1*/(i32 1), $src0)
 >;
 
 def : Ext32Pat <zext>;
@@ -1240,8 +1242,9 @@ def : GCNPat <
 class ZExt_i64_i1_Pat <SDNode ext> : GCNPat <
   (i64 (ext i1:$src)),
     (REG_SEQUENCE VReg_64,
-      (V_CNDMASK_B32_e64 (i32 0), (i32 1), $src), sub0,
-      (S_MOV_B32 (i32 0)), sub1)
+      (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                         /*src1mod*/(i32 0), /*src1*/(i32 1), $src),
+      sub0, (S_MOV_B32 (i32 0)), sub1)
 >;
 
 
@@ -1259,8 +1262,10 @@ def : GCNPat <
 def : GCNPat <
   (i64 (sext i1:$src)),
   (REG_SEQUENCE VReg_64,
-    (V_CNDMASK_B32_e64 (i32 0), (i32 -1), $src), sub0,
-    (V_CNDMASK_B32_e64 (i32 0), (i32 -1), $src), sub1)
+    (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                       /*src1mod*/(i32 0), /*src1*/(i32 -1), $src), sub0,
+    (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                       /*src1mod*/(i32 0), /*src1*/(i32 -1), $src), sub1)
 >;
 
 class FPToI1Pat<Instruction Inst, int KOne, ValueType kone_type, ValueType vt, SDPatternOperator fp_to_int> : GCNPat <
@@ -1318,32 +1323,46 @@ def : GCNPat <
 
 def : GCNPat <
   (f16 (sint_to_fp i1:$src)),
-  (V_CVT_F16_F32_e32 (V_CNDMASK_B32_e64 (i32 0), (i32 CONST.FP32_NEG_ONE), $src))
+  (V_CVT_F16_F32_e32 (
+      V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                        /*src1mod*/(i32 0), /*src1*/(i32 CONST.FP32_NEG_ONE),
+                        $src))
 >;
 
 def : GCNPat <
   (f16 (uint_to_fp i1:$src)),
-  (V_CVT_F16_F32_e32 (V_CNDMASK_B32_e64 (i32 0), (i32 CONST.FP32_ONE), $src))
+  (V_CVT_F16_F32_e32 (
+      V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                        /*src1mod*/(i32 0), /*src1*/(i32 CONST.FP32_ONE),
+                        $src))
 >;
 
 def : GCNPat <
   (f32 (sint_to_fp i1:$src)),
-  (V_CNDMASK_B32_e64 (i32 0), (i32 CONST.FP32_NEG_ONE), $src)
+  (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                        /*src1mod*/(i32 0), /*src1*/(i32 CONST.FP32_NEG_ONE),
+                        $src)
 >;
 
 def : GCNPat <
   (f32 (uint_to_fp i1:$src)),
-  (V_CNDMASK_B32_e64 (i32 0), (i32 CONST.FP32_ONE), $src)
+  (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                        /*src1mod*/(i32 0), /*src1*/(i32 CONST.FP32_ONE),
+                        $src)
 >;
 
 def : GCNPat <
   (f64 (sint_to_fp i1:$src)),
-  (V_CVT_F64_I32_e32 (V_CNDMASK_B32_e64 (i32 0), (i32 -1), $src))
+  (V_CVT_F64_I32_e32 (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                                        /*src1mod*/(i32 0), /*src1*/(i32 -1),
+                                        $src))
 >;
 
 def : GCNPat <
   (f64 (uint_to_fp i1:$src)),
-  (V_CVT_F64_U32_e32 (V_CNDMASK_B32_e64 (i32 0), (i32 1), $src))
+  (V_CVT_F64_U32_e32 (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                                        /*src1mod*/(i32 0), /*src1*/(i32 1),
+                                        $src))
 >;
 
 //===----------------------------------------------------------------------===//

diff --git a/llvm/lib/Target/AMDGPU/SILowerI1Copies.cpp b/llvm/lib/Target/AMDGPU/SILowerI1Copies.cpp
index 2553617..823c904 100644 (file)
--- a/llvm/lib/Target/AMDGPU/SILowerI1Copies.cpp
+++ b/llvm/lib/Target/AMDGPU/SILowerI1Copies.cpp
@@ -483,6 +483,8 @@ void SILowerI1Copies::lowerCopiesFromI1() {
       ConstrainRegs.insert(SrcReg);
       BuildMI(MBB, MI, DL, TII->get(AMDGPU::V_CNDMASK_B32_e64), DstReg)
           .addImm(0)
+          .addImm(0)
+          .addImm(0)
           .addImm(-1)
           .addReg(SrcReg);
       DeadCopies.push_back(&MI);

diff --git a/llvm/lib/Target/AMDGPU/SIOptimizeExecMaskingPreRA.cpp b/llvm/lib/Target/AMDGPU/SIOptimizeExecMaskingPreRA.cpp
index 7832320..bc30b29 100644 (file)
--- a/llvm/lib/Target/AMDGPU/SIOptimizeExecMaskingPreRA.cpp
+++ b/llvm/lib/Target/AMDGPU/SIOptimizeExecMaskingPreRA.cpp
@@ -171,6 +171,10 @@ static unsigned optimizeVcndVcmpPair(MachineBasicBlock &MBB,
   if (!Sel || Sel->getOpcode() != AMDGPU::V_CNDMASK_B32_e64)
     return AMDGPU::NoRegister;
 
+  if (TII->hasModifiersSet(*Sel, AMDGPU::OpName::src0_modifiers) ||
+      TII->hasModifiersSet(*Sel, AMDGPU::OpName::src0_modifiers))
+    return AMDGPU::NoRegister;
+
   Op1 = TII->getNamedOperand(*Sel, AMDGPU::OpName::src0);
   Op2 = TII->getNamedOperand(*Sel, AMDGPU::OpName::src1);
   MachineOperand *CC = TII->getNamedOperand(*Sel, AMDGPU::OpName::src2);

diff --git a/llvm/lib/Target/AMDGPU/VOP2Instructions.td b/llvm/lib/Target/AMDGPU/VOP2Instructions.td
index 9a0a81c..1cb9bdb 100644 (file)
--- a/llvm/lib/Target/AMDGPU/VOP2Instructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP2Instructions.td
@@ -245,7 +245,8 @@ def VOP_MADMK_F32 : VOP_MADMK <f32>;
 class VOP_MAC <ValueType vt> : VOPProfile <[vt, vt, vt, vt]> {
   let Ins32 = (ins Src0RC32:$src0, Src1RC32:$src1, VGPR_32:$src2);
   let Ins64 = getIns64<Src0RC64, Src1RC64, RegisterOperand<VGPR_32>, 3,
-                       0, HasModifiers, HasOMod, Src0Mod, Src1Mod, Src2Mod>.ret;
+                       0, HasModifiers, HasModifiers, HasOMod,
+                       Src0Mod, Src1Mod, Src2Mod>.ret;
   let InsDPP = (ins Src0ModDPP:$src0_modifiers, Src0DPP:$src0,
                     Src1ModDPP:$src1_modifiers, Src1DPP:$src1,
                     VGPR_32:$src2, // stub argument
@@ -324,11 +325,12 @@ def VOP2b_I32_I1_I32_I32_I1 : VOPProfile<[i32, i32, i32, i1]> {
   let HasExtSDWA9 = 1;
 }
 
-// Read in from vcc or arbitrary SGPR
-def VOP2e_I32_I32_I32_I1 : VOPProfile<[i32, i32, i32, i1]> {
+// Read in from vcc or arbitrary SGPR.
+// Enable f32 source modifiers on i32 input type.
+def VOP2e_I32_I32_I32_I1 : VOPProfile<[i32, i32, i32, i1], /*EnableF32SrcMods=*/1> {
   let Src0RC32 = VCSrc_b32; // See comment in def VOP2b_I32_I1_I32_I32_I1 above.
   let Asm32 = "$vdst, $src0, $src1, vcc";
-  let Asm64 = "$vdst, $src0, $src1, $src2";
+  let Asm64 = "$vdst, $src0_modifiers, $src1_modifiers, $src2";
   let AsmSDWA = "$vdst, $src0_modifiers, $src1_modifiers, vcc $clamp $dst_sel $dst_unused $src0_sel $src1_sel";
   let AsmSDWA9 = "$vdst, $src0_modifiers, $src1_modifiers, vcc $clamp $dst_sel $dst_unused $src0_sel $src1_sel";
   let AsmDPP = "$vdst, $src0, $src1, vcc $dpp_ctrl$row_mask$bank_mask$bound_ctrl";
@@ -347,8 +349,8 @@ def VOP2e_I32_I32_I32_I1 : VOPProfile<[i32, i32, i32, i1]> {
                      src0_sel:$src0_sel, src1_sel:$src1_sel);
 
   let InsDPP = (ins DstRCDPP:$old,
-                    Src0DPP:$src0,
-                    Src1DPP:$src1,
+                    Src0ModDPP:$src0_modifiers, Src0DPP:$src0,
+                    Src1ModDPP:$src1_modifiers, Src1DPP:$src1,
                     dpp_ctrl:$dpp_ctrl, row_mask:$row_mask,
                     bank_mask:$bank_mask, bound_ctrl:$bound_ctrl);
   let HasExt = 1;
@@ -644,7 +646,9 @@ def : GCNPat<
 
 class ZExt_i16_i1_Pat <SDNode ext> : GCNPat <
   (i16 (ext i1:$src)),
-  (V_CNDMASK_B32_e64 (i32 0), (i32 1), $src)
+  (V_CNDMASK_B32_e64 (i32 0/*src0mod*/), (i32 0/*src0*/),
+                     (i32 0/*src1mod*/), (i32 1/*src1*/),
+                     $src)
 >;
 
 let Predicates = [Has16BitInsts] in {
@@ -681,7 +685,8 @@ def : ZExt_i16_i1_Pat<anyext>;
 
 def : GCNPat <
   (i16 (sext i1:$src)),
-  (V_CNDMASK_B32_e64 (i32 0), (i32 -1), $src)
+  (V_CNDMASK_B32_e64 /*src0mod*/(i32 0), /*src0*/(i32 0),
+                     /*src1mod*/(i32 0), /*src1*/(i32 -1), $src)
 >;
 
 // Undo sub x, c -> add x, -c canonicalization since c is more likely

diff --git a/llvm/lib/Target/AMDGPU/VOP3Instructions.td b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
index babd041..cc3de25 100644 (file)
--- a/llvm/lib/Target/AMDGPU/VOP3Instructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
@@ -190,6 +190,7 @@ class VOP3_Profile<VOPProfile P, VOP3Features Features = VOP3_REGULAR> : VOPProf
 class VOP3b_Profile<ValueType vt> : VOPProfile<[vt, vt, vt, vt]> {
   // v_div_scale_{f32|f64} do not support input modifiers.
   let HasModifiers = 0;
+  let HasClamp = 0;
   let HasOMod = 0;
   let Outs64 = (outs DstRC:$vdst, SReg_64:$sdst);
   let Asm64 = " $vdst, $sdst, $src0, $src1, $src2";

diff --git a/llvm/test/CodeGen/AMDGPU/coalescer-extend-pruned-subrange.mir b/llvm/test/CodeGen/AMDGPU/coalescer-extend-pruned-subrange.mir
index 2e4624e..576d8df 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/coalescer-extend-pruned-subrange.mir
+++ b/llvm/test/CodeGen/AMDGPU/coalescer-extend-pruned-subrange.mir
@@ -29,7 +29,7 @@ body: |
     %13:vgpr_32 = V_OR_B32_e32 %11, %12.sub2, implicit $exec
     %14:vgpr_32 = V_AND_B32_e32 1, %13, implicit $exec
     %15:sreg_64_xexec = V_CMP_EQ_U32_e64 0, %14, implicit $exec
-    %16:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %15, implicit $exec
+    %16:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %15, implicit $exec
     BUFFER_STORE_DWORD_OFFEN_exact %16, undef %17:vgpr_32, undef %18:sreg_128, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store 4 into constant-pool, align 1, addrspace 4)
     S_ENDPGM 0
 
@@ -80,7 +80,7 @@ body: |
     successors: %bb.10
     %31:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN undef %32:vgpr_32, undef %33:sreg_128, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load 4 from constant-pool, align 1, addrspace 4)
     %34:sreg_64_xexec = V_CMP_NE_U32_e64 0, %31, implicit $exec
-    %35:vgpr_32 = V_CNDMASK_B32_e64 0, -1, %34, implicit $exec
+    %35:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, -1, %34, implicit $exec
     %28:vreg_1 = COPY %35
     S_BRANCH %bb.10
 
@@ -93,7 +93,7 @@ body: |
     $exec = S_OR_B64 $exec, %29, implicit-def $scc
     %36:vreg_1 = COPY %28
     %37:sreg_64_xexec = V_CMP_NE_U32_e64 0, %36, implicit $exec
-    %38:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %37, implicit $exec
+    %38:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %37, implicit $exec
     %39:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
     undef %40.sub0:vreg_128 = COPY %39
     %40.sub1:vreg_128 = COPY %39

diff --git a/llvm/test/CodeGen/AMDGPU/coalescer-subranges-another-prune-error.mir b/llvm/test/CodeGen/AMDGPU/coalescer-subranges-another-prune-error.mir
index 41af493..bb4fdc9 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/coalescer-subranges-another-prune-error.mir
+++ b/llvm/test/CodeGen/AMDGPU/coalescer-subranges-another-prune-error.mir
@@ -74,7 +74,7 @@ body:             |
     %29.sub0:vreg_128 = COPY %1
     %30:sreg_64 = V_CMP_NE_U32_e64 0, %28, implicit $exec
     %31:sreg_64_xexec = V_CMP_EQ_U32_e64 0, %28, implicit $exec
-    dead %32:vgpr_32 = V_CNDMASK_B32_e64 0, -1, killed %31, implicit $exec
+    dead %32:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, -1, killed %31, implicit $exec
     %33:vreg_128 = COPY %29
     %33.sub1:vreg_128 = COPY undef %32
     %34:vgpr_32 = V_MOV_B32_e32 -1, implicit $exec

diff --git a/llvm/test/CodeGen/AMDGPU/coalescer-with-subregs-bad-identical.mir b/llvm/test/CodeGen/AMDGPU/coalescer-with-subregs-bad-identical.mir
index b2d2064..15cc8f4 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/coalescer-with-subregs-bad-identical.mir
+++ b/llvm/test/CodeGen/AMDGPU/coalescer-with-subregs-bad-identical.mir
@@ -69,7 +69,7 @@ body: |
     %18:vgpr_32 = V_MIN_F32_e32 1065353216, killed %17, implicit $exec
     %19:sreg_64_xexec = V_CMP_NEQ_F32_e64 0, 1065353216, 0, killed %18, 0, implicit $exec
     %20:vgpr_32 = V_MOV_B32_e32 2143289344, implicit $exec
-    %21:vgpr_32 = V_CNDMASK_B32_e64 0, killed %20, killed %19, implicit $exec
+    %21:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, killed %20, killed %19, implicit $exec
     %22:sreg_64 = V_CMP_LT_F32_e64 0, 0, 0, killed %21, 0, implicit $exec
     %23:sreg_64 = COPY $exec, implicit-def $exec
     %24:sreg_64 = S_AND_B64 %23, %22, implicit-def dead $scc

diff --git a/llvm/test/CodeGen/AMDGPU/fold-cndmask.mir b/llvm/test/CodeGen/AMDGPU/fold-cndmask.mir
index ae7bdbe..f2072c0 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/fold-cndmask.mir
+++ b/llvm/test/CodeGen/AMDGPU/fold-cndmask.mir
@@ -22,12 +22,12 @@ registers:
 body:             |
   bb.0.entry:
     %0 = IMPLICIT_DEF
-    %1 = V_CNDMASK_B32_e64 0, 0, %0, implicit $exec
-    %2 = V_CNDMASK_B32_e64 %1, %1, %0, implicit $exec
+    %1 = V_CNDMASK_B32_e64 0, 0, 0, 0, %0, implicit $exec
+    %2 = V_CNDMASK_B32_e64 0, %1, 0, %1, %0, implicit $exec
     %3 = IMPLICIT_DEF
-    %4 = V_CNDMASK_B32_e64 %3, %3, %0, implicit $exec
+    %4 = V_CNDMASK_B32_e64 0, %3, 0, %3, %0, implicit $exec
     %5 = COPY %1
-    %6 = V_CNDMASK_B32_e64 %5, 0, %0, implicit $exec
+    %6 = V_CNDMASK_B32_e64 0, %5, 0, 0, %0, implicit $exec
     $vcc = IMPLICIT_DEF
     %7 = V_CNDMASK_B32_e32 %3, %3, implicit $exec, implicit $vcc
 

diff --git a/llvm/test/CodeGen/AMDGPU/macro-fusion-cluster-vcc-uses.mir b/llvm/test/CodeGen/AMDGPU/macro-fusion-cluster-vcc-uses.mir
index 5ab9ed9..f684afe 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/macro-fusion-cluster-vcc-uses.mir
+++ b/llvm/test/CodeGen/AMDGPU/macro-fusion-cluster-vcc-uses.mir
@@ -150,7 +150,7 @@ body: |
 # GCN-LABEL: name: cluster_cmp_cndmask
 # GCN: S_NOP 0, implicit-def $vcc
 # GCN-NEXT: %3:sreg_64_xexec = V_CMP_EQ_I32_e64 %0, %1, implicit $exec
-# GCN-NEXT: dead %4:vgpr_32 = V_CNDMASK_B32_e64 %0, %1, %3, implicit $exec
+# GCN-NEXT: dead %4:vgpr_32 = V_CNDMASK_B32_e64 0, %0, 0, %1, %3, implicit $exec
 name: cluster_cmp_cndmask
 registers:
   - { id: 0, class: vgpr_32 }
@@ -168,13 +168,13 @@ body: |
     %1 = V_MOV_B32_e32 0, implicit $exec
     %3 = V_CMP_EQ_I32_e64 %0, %1, implicit $exec
     S_NOP 0, implicit def $vcc
-    %4 = V_CNDMASK_B32_e64 %0, %1, %3, implicit $exec
+    %4 = V_CNDMASK_B32_e64 0, %0, 0, %1, %3, implicit $exec
 ...
 
 # GCN-LABEL: name: cluster_multi_use_cmp_cndmask
 # GCN: %4:sreg_64_xexec = V_CMP_EQ_I32_e64 %0, %1, implicit $exec
-# GCN-NEXT: dead %5:vgpr_32 = V_CNDMASK_B32_e64 %2, %1, %4, implicit $exec
-# GCN-NEXT: dead %6:vgpr_32 = V_CNDMASK_B32_e64 %1, %3, %4, implicit $exec
+# GCN-NEXT: dead %5:vgpr_32 = V_CNDMASK_B32_e64 0, %2, 0, %1, %4, implicit $exec
+# GCN-NEXT: dead %6:vgpr_32 = V_CNDMASK_B32_e64 0, %1, 0, %3, %4, implicit $exec
 name: cluster_multi_use_cmp_cndmask
 registers:
   - { id: 0, class: vgpr_32 }
@@ -195,15 +195,15 @@ body: |
 
     %4 = V_CMP_EQ_I32_e64 %0, %1, implicit $exec
     S_NOP 0, implicit def $vcc
-    %5 = V_CNDMASK_B32_e64 %2, %1, %4, implicit $exec
-    %6 = V_CNDMASK_B32_e64 %1, %3, %4, implicit $exec
+    %5 = V_CNDMASK_B32_e64 0, %2, 0, %1, %4, implicit $exec
+    %6 = V_CNDMASK_B32_e64 0, %1, 0, %3, %4, implicit $exec
 ...
 
 # GCN-LABEL: name: cluster_multi_use_cmp_cndmask2
 # GCN: %4:sreg_64_xexec = V_CMP_EQ_I32_e64 %0, %1, implicit $exec
-# GCN-NEXT: dead %5:vgpr_32 = V_CNDMASK_B32_e64 %2, %1, %4, implicit $exec
+# GCN-NEXT: dead %5:vgpr_32 = V_CNDMASK_B32_e64 0, %2, 0, %1, %4, implicit $exec
 # GCN-NEXT: %3:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
-# GCN-NEXT: dead %6:vgpr_32 = V_CNDMASK_B32_e64 %1, %3, %4, implicit $exec
+# GCN-NEXT: dead %6:vgpr_32 = V_CNDMASK_B32_e64 0, %1, 0, %3, %4, implicit $exec
 name: cluster_multi_use_cmp_cndmask2
 registers:
   - { id: 0, class: vgpr_32 }
@@ -221,7 +221,7 @@ body: |
     %1 = V_MOV_B32_e32 0, implicit $exec
     %4 = V_CMP_EQ_I32_e64 %0, %1, implicit $exec
     %2 = V_MOV_B32_e32 0, implicit $exec
-    %5 = V_CNDMASK_B32_e64 %2, %1, %4, implicit $exec
+    %5 = V_CNDMASK_B32_e64 0, %2, 0, %1, %4, implicit $exec
     %3 = V_MOV_B32_e32 0, implicit $exec
-    %6 = V_CNDMASK_B32_e64 %1, %3, %4, implicit $exec
+    %6 = V_CNDMASK_B32_e64 0, %1, 0, %3, %4, implicit $exec
 ...

diff --git a/llvm/test/CodeGen/AMDGPU/merge-load-store-vreg.mir b/llvm/test/CodeGen/AMDGPU/merge-load-store-vreg.mir
index b4d4662..0f3913e 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/merge-load-store-vreg.mir
+++ b/llvm/test/CodeGen/AMDGPU/merge-load-store-vreg.mir
@@ -77,7 +77,7 @@ body:             |
 
   bb.2:
     %1:sreg_64_xexec = V_CMP_NE_U32_e64 %0, 0, implicit $exec
-    %2:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %1, implicit $exec
+    %2:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %1, implicit $exec
     V_CMP_NE_U32_e32 1, %2, implicit-def $vcc, implicit $exec
     DS_WRITE_B32 %0, %0, 1024, 0, implicit $m0, implicit $exec :: (store 4 into %ir.tmp)
     %3:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
@@ -112,7 +112,7 @@ body:             |
 
   bb.2:
     %1:sreg_64_xexec = V_CMP_NE_U32_e64 %0.sub0, 0, implicit $exec
-    %2:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %1, implicit $exec
+    %2:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %1, implicit $exec
     V_CMP_NE_U32_e32 1, %2, implicit-def $vcc, implicit $exec
     DS_WRITE_B32 %0.sub0, %0.sub0, 1024, 0, implicit $m0, implicit $exec :: (store 4 into %ir.tmp)
     undef %3.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec
@@ -143,7 +143,7 @@ body:             |
 
   bb.2:
     %1:sreg_64_xexec = V_CMP_NE_U32_e64 %0.sub0, 0, implicit $exec
-    %2:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %1, implicit $exec
+    %2:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %1, implicit $exec
     V_CMP_NE_U32_e32 1, %2, implicit-def $vcc, implicit $exec
     DS_WRITE_B32 %0.sub0, %0.sub0, 0, 0, implicit $m0, implicit $exec :: (store 4 into %ir.tmp)
     undef %3.sub0:vreg_64 = V_MOV_B32_e32 0, implicit $exec

diff --git a/llvm/test/CodeGen/AMDGPU/optimize-negated-cond-exec-masking.mir b/llvm/test/CodeGen/AMDGPU/optimize-negated-cond-exec-masking.mir
index 9851321..a9b8d94 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/optimize-negated-cond-exec-masking.mir
+++ b/llvm/test/CodeGen/AMDGPU/optimize-negated-cond-exec-masking.mir
@@ -9,7 +9,7 @@ name:            negated_cond_vop2
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
     $vcc = S_AND_B64 $exec, killed $vcc, implicit-def dead $scc
     S_CBRANCH_VCCNZ %bb.2, implicit killed $vcc
@@ -31,7 +31,7 @@ name:            negated_cond_vop3
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1, 1, implicit $exec
     $vcc = S_AND_B64 killed %2, $exec, implicit-def dead $scc
     S_CBRANCH_VCCNZ %bb.2, implicit killed $vcc
@@ -46,7 +46,7 @@ body:             |
 
 # GCN: name: negated_cond_vop2_redef_vcc1
 # GCN:      %0:sreg_64_xexec = IMPLICIT_DEF
-# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN-NEXT: V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
 # GCN-NEXT: $vcc_lo = COPY $sgpr0
 # GCN-NEXT: $vcc = S_AND_B64 $exec, $vcc, implicit-def dead $scc
@@ -56,7 +56,7 @@ name:            negated_cond_vop2_redef_vcc1
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
     $vcc_lo = COPY $sgpr0
     $vcc = S_AND_B64 $exec, killed $vcc, implicit-def dead $scc
@@ -72,7 +72,7 @@ body:             |
 
 # GCN: name: negated_cond_vop2_redef_vcc2
 # GCN:      %0:sreg_64_xexec = IMPLICIT_DEF
-# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN-NEXT: V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
 # GCN-NEXT: $vcc_hi = COPY $sgpr0
 # GCN-NEXT: $vcc = S_AND_B64 $exec, $vcc, implicit-def dead $scc
@@ -82,7 +82,7 @@ name:            negated_cond_vop2_redef_vcc2
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
     $vcc_hi = COPY $sgpr0
     $vcc = S_AND_B64 $exec, killed $vcc, implicit-def dead $scc
@@ -98,7 +98,7 @@ body:             |
 
 # GCN: name: negated_cond_vop3_redef_cmp
 # GCN:      %0:sreg_64_xexec = IMPLICIT_DEF
-# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN-NEXT: %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1, 1, implicit $exec
 # GCN-NEXT: %2.sub1:sreg_64_xexec = COPY $sgpr0
 # GCN-NEXT: $vcc = S_AND_B64 %2, $exec, implicit-def dead $scc
@@ -108,7 +108,7 @@ name:            negated_cond_vop3_redef_cmp
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1, 1, implicit $exec
     %2.sub1 = COPY $sgpr0
     $vcc = S_AND_B64 killed %2, $exec, implicit-def dead $scc
@@ -149,7 +149,7 @@ name:            negated_cond_vop3_imp_vcc
 body:             |
   bb.0:
     $vcc = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, $vcc, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, $vcc, implicit $exec
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1, 1, implicit $exec
     $vcc = S_AND_B64 killed %2, $exec, implicit-def dead $scc
     S_CBRANCH_VCCNZ %bb.2, implicit killed $vcc
@@ -171,7 +171,7 @@ name:            negated_cond_vop2_imp_vcc
 body:             |
   bb.0:
     $vcc = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, $vcc, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, $vcc, implicit $exec
     V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
     $vcc = S_AND_B64 killed $vcc, $exec, implicit-def dead $scc
     S_CBRANCH_VCCNZ %bb.2, implicit killed $vcc
@@ -186,7 +186,7 @@ body:             |
 
 # GCN: name: negated_cond_vop3_redef_sel
 # GCN:      %0:sreg_64_xexec = IMPLICIT_DEF
-# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN-NEXT: %1:vgpr_32 = COPY $vgpr0
 # GCN-NEXT: %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1, 1, implicit $exec
 # GCN-NEXT: $vcc = S_AND_B64 %2, $exec, implicit-def dead $scc
@@ -196,7 +196,7 @@ name:            negated_cond_vop3_redef_sel
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     %1:vgpr_32 = COPY $vgpr0
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1, 1, implicit $exec
     $vcc = S_AND_B64 killed %2, $exec, implicit-def dead $scc
@@ -212,7 +212,7 @@ body:             |
 
 # GCN: name: negated_cond_vop2_used_sel
 # GCN:      %0:sreg_64_xexec = IMPLICIT_DEF
-# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN-NEXT: $vcc = S_ANDN2_B64 $exec, %0, implicit-def $scc
 # GCN-NEXT: S_CBRANCH_VCCNZ %bb.2, implicit $vcc
 ---
@@ -220,7 +220,7 @@ name:            negated_cond_vop2_used_sel
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
     $vcc = S_AND_B64 $exec, killed $vcc, implicit-def dead $scc
     S_CBRANCH_VCCNZ %bb.2, implicit killed $vcc
@@ -236,7 +236,7 @@ body:             |
 
 # GCN: name: negated_cond_vop2_used_vcc
 # GCN:      %0:sreg_64_xexec = IMPLICIT_DEF
-# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN-NEXT: %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN-NEXT: V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
 # GCN-NEXT: $sgpr0_sgpr1 = COPY $vcc
 # GCN-NEXT: $vcc = S_ANDN2_B64 $exec, %0, implicit-def $scc
@@ -246,7 +246,7 @@ name:            negated_cond_vop2_used_vcc
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
     $sgpr0_sgpr1 = COPY $vcc
     $vcc = S_AND_B64 $exec, killed $vcc, implicit-def dead $scc
@@ -263,7 +263,7 @@ body:             |
 # GCN: name: negated_cond_vop3_sel_wrong_subreg1
 # GCN:      %0:sreg_64_xexec = IMPLICIT_DEF
 # GCN-NEXT: %1.sub1:vreg_64 = IMPLICIT_DEF
-# GCN-NEXT: %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN-NEXT: %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN-NEXT: %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1.sub1, 1, implicit $exec
 # GCN-NEXT: $vcc = S_AND_B64 %2, $exec, implicit-def dead $scc
 # GCN-NEXT: S_CBRANCH_VCCNZ %bb.2, implicit $vcc
@@ -273,7 +273,7 @@ body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
     %1.sub1 = IMPLICIT_DEF
-    %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1.sub1, 1, implicit $exec
     $vcc = S_AND_B64 killed %2, $exec, implicit-def dead $scc
     S_CBRANCH_VCCNZ %bb.2, implicit killed $vcc
@@ -288,7 +288,7 @@ body:             |
 
 # GCN: name: negated_cond_vop3_sel_wrong_subreg2
 # GCN:      %0:sreg_64_xexec = IMPLICIT_DEF
-# GCN-NEXT: %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN-NEXT: %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN-NEXT: %1.sub1:vreg_64 = IMPLICIT_DEF
 # GCN-NEXT: %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1.sub1, 1, implicit $exec
 # GCN-NEXT: $vcc = S_AND_B64 %2, $exec, implicit-def dead $scc
@@ -298,7 +298,7 @@ name:            negated_cond_vop3_sel_wrong_subreg2
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     %1.sub1 = IMPLICIT_DEF
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1.sub1, 1, implicit $exec
     $vcc = S_AND_B64 killed %2, $exec, implicit-def dead $scc
@@ -323,7 +323,7 @@ body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
     %1.sub1 = IMPLICIT_DEF
-    %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1.sub0, 1, implicit $exec
     $vcc = S_AND_B64 killed %2, $exec, implicit-def dead $scc
     S_CBRANCH_VCCNZ %bb.2, implicit killed $vcc
@@ -346,7 +346,7 @@ name:            negated_cond_vop3_sel_right_subreg2
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1.sub0:vreg_64 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     %1.sub1 = IMPLICIT_DEF
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1.sub0, 1, implicit $exec
     $vcc = S_AND_B64 killed %2, $exec, implicit-def dead $scc
@@ -362,7 +362,7 @@ body:             |
 
 # GCN: name: negated_cond_vop3_sel_subreg_overlap
 # GCN:      %0:sreg_64_xexec = IMPLICIT_DEF
-# GCN-NEXT: %1.sub2:vreg_128 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN-NEXT: %1.sub2:vreg_128 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN-NEXT: %1.sub2_sub3:vreg_128 = IMPLICIT_DEF
 # GCN-NEXT: %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1.sub2, 1, implicit $exec
 # GCN-NEXT: $vcc = S_AND_B64 %2, $exec, implicit-def dead $scc
@@ -372,7 +372,7 @@ name:            negated_cond_vop3_sel_subreg_overlap
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1.sub2:vreg_128 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1.sub2:vreg_128 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     %1.sub2_sub3 = IMPLICIT_DEF
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1.sub2, 1, implicit $exec
     $vcc = S_AND_B64 killed %2, $exec, implicit-def dead $scc
@@ -395,7 +395,7 @@ name:            negated_cond_vop2_dominated_blocks
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 
   bb.1:
     V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec
@@ -411,7 +411,7 @@ body:             |
 ...
 
 # GCN: name: negated_cond_vop2_different_blocks_cmp_and
-# GCN:      %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+# GCN:      %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 # GCN:      $vcc = S_AND_B64 $exec, %2, implicit-def dead $scc
 # GCN-NEXT: S_CBRANCH_VCCNZ %bb.3, implicit $vcc
 ---
@@ -419,7 +419,7 @@ name:            negated_cond_vop2_different_blocks_cmp_and
 body:             |
   bb.0:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
     %2:sreg_64_xexec = V_CMP_NE_U32_e64 %1, 1, implicit $exec
 
   bb.1:
@@ -435,7 +435,7 @@ body:             |
 ...
 
 # GCN: name: negated_cond_vop2_not_dominated_blocks
-# GCN:      V_CNDMASK_B32_e64 0, 1,
+# GCN:      V_CNDMASK_B32_e64 0, 0, 0, 1,
 # GCN:      $vcc = S_AND_B64 $exec, $vcc, implicit-def dead $scc
 # GCN-NEXT: S_CBRANCH_VCCNZ %bb.4, implicit $vcc
 ---
@@ -449,7 +449,7 @@ body:             |
 
   bb.1:
     %0:sreg_64_xexec = IMPLICIT_DEF
-    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %0, implicit $exec
+    %1:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %0, implicit $exec
 
   bb.2:
     V_CMP_NE_U32_e32 1, %1, implicit-def $vcc, implicit $exec

diff --git a/llvm/test/CodeGen/AMDGPU/regcoal-subrange-join-seg.mir b/llvm/test/CodeGen/AMDGPU/regcoal-subrange-join-seg.mir
index f11d9de..8a16f50 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/regcoal-subrange-join-seg.mir
+++ b/llvm/test/CodeGen/AMDGPU/regcoal-subrange-join-seg.mir
@@ -189,7 +189,7 @@ body:             |
     %46 = V_AND_B32_e32 1, killed %45, implicit $exec
     %21 = S_BUFFER_LOAD_DWORD_SGPR undef %22, undef %23, 0 :: (dereferenceable invariant load 4)
     %25 = V_CMP_GE_F32_e64 0, 0, 0, killed %21, 0, implicit $exec
-    %26 = V_CNDMASK_B32_e64 0, -1, killed %25, implicit $exec
+    %26 = V_CNDMASK_B32_e64 0, 0, 0, -1, killed %25, implicit $exec
     %62 = IMPLICIT_DEF
 
   bb.29:

diff --git a/llvm/test/CodeGen/AMDGPU/shrink-vop3-carry-out.mir b/llvm/test/CodeGen/AMDGPU/shrink-vop3-carry-out.mir
index 30da2ec..f403b22 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/shrink-vop3-carry-out.mir
+++ b/llvm/test/CodeGen/AMDGPU/shrink-vop3-carry-out.mir
@@ -9,7 +9,7 @@
 ...
 # GCN-LABEL: name: shrink_add_vop3{{$}}
 # GCN: %29:vgpr_32, %9:sreg_64_xexec = V_ADD_I32_e64 %19, %17, implicit $exec
-# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 1, killed %9, implicit $exec
+# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed %9, implicit $exec
 name:            shrink_add_vop3
 alignment:       0
 exposesReturnsTwice: false
@@ -84,7 +84,7 @@ body:             |
     %17 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 0, 0, 0, 0, implicit $exec
     %19 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 4, 0, 0, 0, implicit $exec
     %29, %9 = V_ADD_I32_e64 %19, %17, implicit $exec
-    %24 = V_CNDMASK_B32_e64 0, 1, killed %9, implicit $exec
+    %24 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed %9, implicit $exec
     BUFFER_STORE_DWORD_ADDR64 %24, %28, killed %16, 0, 0, 0, 0, 0, implicit $exec
     S_ENDPGM 0
 
@@ -92,7 +92,7 @@ body:             |
 ---
 # GCN-LABEL: name: shrink_sub_vop3{{$}}
 # GCN: %29:vgpr_32, %9:sreg_64_xexec = V_SUB_I32_e64 %19, %17, implicit $exec
-# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 1, killed %9, implicit $exec
+# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed %9, implicit $exec
 
 name:            shrink_sub_vop3
 alignment:       0
@@ -168,7 +168,7 @@ body:             |
     %17 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 0, 0, 0, 0, implicit $exec
     %19 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 4, 0, 0, 0, implicit $exec
     %29, %9 = V_SUB_I32_e64 %19, %17, implicit $exec
-    %24 = V_CNDMASK_B32_e64 0, 1, killed %9, implicit $exec
+    %24 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed %9, implicit $exec
     BUFFER_STORE_DWORD_ADDR64 %24, %28, killed %16, 0, 0, 0, 0, 0, implicit $exec
     S_ENDPGM 0
 
@@ -176,7 +176,7 @@ body:             |
 ---
 # GCN-LABEL: name: shrink_subrev_vop3{{$}}
 # GCN: %29:vgpr_32, %9:sreg_64_xexec = V_SUBREV_I32_e64 %19, %17, implicit $exec
-# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 1, killed %9, implicit $exec
+# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed %9, implicit $exec
 
 name:            shrink_subrev_vop3
 alignment:       0
@@ -252,7 +252,7 @@ body:             |
     %17 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 0, 0, 0, 0, implicit $exec
     %19 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 4, 0, 0, 0, implicit $exec
     %29, %9 = V_SUBREV_I32_e64 %19, %17, implicit $exec
-    %24 = V_CNDMASK_B32_e64 0, 1, killed %9, implicit $exec
+    %24 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed %9, implicit $exec
     BUFFER_STORE_DWORD_ADDR64 %29, %28, killed %16, 0, 0, 0, 0, 0, implicit $exec
     S_ENDPGM 0
 
@@ -260,7 +260,7 @@ body:             |
 ---
 # GCN-LABEL: name: check_addc_src2_vop3{{$}}
 # GCN: %29:vgpr_32, $vcc = V_ADDC_U32_e64 %19, %17, %9, implicit $exec
-# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 1, killed $vcc, implicit $exec
+# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed $vcc, implicit $exec
 name: check_addc_src2_vop3
 alignment:       0
 exposesReturnsTwice: false
@@ -336,7 +336,7 @@ body:             |
     %19 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 4, 0, 0, 0, implicit $exec
     %9 = S_MOV_B64 0
     %29, $vcc = V_ADDC_U32_e64 %19, %17, %9, implicit $exec
-    %24 = V_CNDMASK_B32_e64 0, 1, killed $vcc, implicit $exec
+    %24 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed $vcc, implicit $exec
     BUFFER_STORE_DWORD_ADDR64 %24, %28, killed %16, 0, 0, 0, 0, 0, implicit $exec
     S_ENDPGM 0
 
@@ -344,7 +344,7 @@ body:             |
 ---
 # GCN-LABEL: name: shrink_addc_vop3{{$}}
 # GCN: %29:vgpr_32 = V_ADDC_U32_e32 %19, %17, implicit-def $vcc, implicit $vcc, implicit $exec
-# GCN %24 = V_CNDMASK_B32_e64 0, 1, killed $vcc, implicit $exec
+# GCN %24 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed $vcc, implicit $exec
 
 name:            shrink_addc_vop3
 alignment:       0
@@ -421,7 +421,7 @@ body:             |
     %19 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 4, 0, 0, 0, implicit $exec
     $vcc = S_MOV_B64 0
     %29, $vcc = V_ADDC_U32_e64 %19, %17, $vcc, implicit $exec
-    %24 = V_CNDMASK_B32_e64 0, 1, killed $vcc, implicit $exec
+    %24 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed $vcc, implicit $exec
     BUFFER_STORE_DWORD_ADDR64 %24, %28, killed %16, 0, 0, 0, 0, 0, implicit $exec
     S_ENDPGM 0
 
@@ -430,7 +430,7 @@ body:             |
 ---
 # GCN-LABEL: name: shrink_addc_undef_vcc{{$}}
 # GCN: %29:vgpr_32 = V_ADDC_U32_e32 %19, %17, implicit-def $vcc, implicit undef $vcc, implicit $exec
-# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 1, killed $vcc, implicit $exec
+# GCN: %24:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed $vcc, implicit $exec
 name:            shrink_addc_undef_vcc
 alignment:       0
 exposesReturnsTwice: false
@@ -505,7 +505,7 @@ body:             |
     %17 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 0, 0, 0, 0, implicit $exec
     %19 = BUFFER_LOAD_DWORD_ADDR64 %28, %13, 0, 4, 0, 0, 0, implicit $exec
     %29, $vcc = V_ADDC_U32_e64 %19, %17, undef $vcc, implicit $exec
-    %24 = V_CNDMASK_B32_e64 0, 1, killed $vcc, implicit $exec
+    %24 = V_CNDMASK_B32_e64 0, 0, 0, 1, killed $vcc, implicit $exec
     BUFFER_STORE_DWORD_ADDR64 %24, %28, killed %16, 0, 0, 0, 0, 0, implicit $exec
     S_ENDPGM 0
 

diff --git a/llvm/test/CodeGen/AMDGPU/subreg-split-live-in-error.mir b/llvm/test/CodeGen/AMDGPU/subreg-split-live-in-error.mir
index 3bb2737..eb42e2c 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/subreg-split-live-in-error.mir
+++ b/llvm/test/CodeGen/AMDGPU/subreg-split-live-in-error.mir
@@ -184,7 +184,7 @@ body: |
     %36:vgpr_32 = V_MAC_F32_e32 0, %33, %36, implicit $exec
     %37:vgpr_32 = V_MAD_F32 0, %35, 0, 0, 0, 0, 0, 0, implicit $exec
     %38:sreg_64_xexec = V_CMP_NE_U32_e64 0, %5, implicit $exec
-    %39:vgpr_32 = V_CNDMASK_B32_e64 0, 1, %38, implicit $exec
+    %39:vgpr_32 = V_CNDMASK_B32_e64 0, 0, 0, 1, %38, implicit $exec
     V_CMP_NE_U32_e32 1, %39, implicit-def $vcc, implicit $exec
     $vcc = S_AND_B64 $exec, $vcc, implicit-def dead $scc
     %40:vgpr_32 = V_ADD_F32_e32 %36, %37, implicit $exec

diff --git a/llvm/test/CodeGen/AMDGPU/undefined-physreg-sgpr-spill.mir b/llvm/test/CodeGen/AMDGPU/undefined-physreg-sgpr-spill.mir
index 5e72713..f94a042 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/undefined-physreg-sgpr-spill.mir
+++ b/llvm/test/CodeGen/AMDGPU/undefined-physreg-sgpr-spill.mir
@@ -47,7 +47,7 @@ body:             |
     $vgpr1 = GLOBAL_LOAD_UBYTE killed $vgpr1_vgpr2, 0, 0, 0, implicit $exec :: (non-temporal dereferenceable invariant load 1 from `i1 addrspace(4)* undef`)
     $vcc = V_CMP_NE_U32_e64 0, $vgpr0, implicit $exec
     $sgpr0_sgpr1 = V_CMP_EQ_U32_e64 1, killed $vgpr1, implicit $exec
-    $vgpr1 = V_CNDMASK_B32_e64 0, -1, killed $sgpr0_sgpr1, implicit $exec
+    $vgpr1 = V_CNDMASK_B32_e64 0, 0, 0, -1, killed $sgpr0_sgpr1, implicit $exec
     $sgpr0_sgpr1 = COPY $exec, implicit-def $exec
     SI_SPILL_S64_SAVE $sgpr0_sgpr1, %stack.0, implicit $exec, implicit $sgpr8_sgpr9_sgpr10_sgpr11, implicit $sgpr13, implicit-def dead $m0 :: (store 8 into %stack.0, align 4, addrspace 5)
     $sgpr2_sgpr3 = S_AND_B64 killed $sgpr0_sgpr1, killed $vcc, implicit-def dead $scc
@@ -111,7 +111,7 @@ body:             |
     $vgpr1 = GLOBAL_LOAD_UBYTE killed $vgpr1_vgpr2, 0, 0, 0, implicit $exec :: (non-temporal dereferenceable invariant load 1 from `i1 addrspace(4)* undef`)
     $vcc = V_CMP_NE_U32_e64 0, $vgpr0, implicit $exec
     $sgpr0_sgpr1 = V_CMP_EQ_U32_e64 1, killed $vgpr1, implicit $exec
-    $vgpr1 = V_CNDMASK_B32_e64 0, -1, killed $sgpr0_sgpr1, implicit $exec
+    $vgpr1 = V_CNDMASK_B32_e64 0, 0, 0, -1, killed $sgpr0_sgpr1, implicit $exec
     $sgpr0_sgpr1 = COPY $exec, implicit-def $exec
     $sgpr2_sgpr3 = S_AND_B64 $sgpr0_sgpr1, killed $vcc, implicit-def dead $scc
     SI_SPILL_S64_SAVE killed $sgpr0_sgpr1, %stack.0, implicit $exec, implicit $sgpr8_sgpr9_sgpr10_sgpr11, implicit $sgpr13, implicit-def dead $m0 :: (store 8 into %stack.0, align 4, addrspace 5)

diff --git a/llvm/test/CodeGen/AMDGPU/waitcnt-back-edge-loop.mir b/llvm/test/CodeGen/AMDGPU/waitcnt-back-edge-loop.mir
index 5dc6ca5..dae7a28 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/waitcnt-back-edge-loop.mir
+++ b/llvm/test/CodeGen/AMDGPU/waitcnt-back-edge-loop.mir
@@ -16,7 +16,7 @@ body:             |
     $vgpr4 = FLAT_LOAD_DWORD $vgpr1_vgpr2, 0, 0, 0, implicit $exec, implicit $flat_scr :: (volatile load 4 from `float addrspace(1)* null`, addrspace 1)
     $vgpr0 = FLAT_LOAD_DWORD $vgpr1_vgpr2, 0, 0, 0, implicit $exec, implicit $flat_scr :: (volatile load 4 from `float addrspace(1)* null`, addrspace 1)
     $sgpr0_sgpr1 = V_CMP_EQ_U32_e64 3, killed $sgpr4, implicit $exec
-    $vgpr3 = V_CNDMASK_B32_e64 -1082130432, 1065353216, killed $sgpr0_sgpr1, implicit $exec
+    $vgpr3 = V_CNDMASK_B32_e64 0, -1082130432, 0, 1065353216, killed $sgpr0_sgpr1, implicit $exec
     $vgpr5 = V_MOV_B32_e32 $vgpr0, implicit $exec, implicit $exec
     S_BRANCH %bb.1
 

diff --git a/llvm/test/MC/AMDGPU/vop3.s b/llvm/test/MC/AMDGPU/vop3.s
index b77c2f4..29341f6 100644 (file)
--- a/llvm/test/MC/AMDGPU/vop3.s
+++ b/llvm/test/MC/AMDGPU/vop3.s
@@ -199,7 +199,7 @@ v_add_f32_e64 v1, v3, v5
 // VI:   v_add_f32_e64 v1, v3, v5 ; encoding: [0x01,0x00,0x01,0xd1,0x03,0x0b,0x02,0x00]
 
 
-// TODO: Modifier tests
+// TODO: Modifier tests (v_cndmask done)
 
 v_cndmask_b32 v1, v3, v5, s[4:5]
 // SICI: v_cndmask_b32_e64 v1, v3, v5, s[4:5] ; encoding: [0x01,0x00,0x00,0xd2,0x03,0x0b,0x12,0x00]
@@ -213,6 +213,18 @@ v_cndmask_b32_e64 v1, v3, v5, vcc
 // SICI: v_cndmask_b32_e64 v1, v3, v5, vcc ; encoding: [0x01,0x00,0x00,0xd2,0x03,0x0b,0xaa,0x01]
 // VI:   v_cndmask_b32_e64 v1, v3, v5, vcc ; encoding: [0x01,0x00,0x00,0xd1,0x03,0x0b,0xaa,0x01]
 
+v_cndmask_b32 v1, -v3, v5, s[4:5]
+// SICI: v_cndmask_b32_e64 v1, -v3, v5, s[4:5] ; encoding: [0x01,0x00,0x00,0xd2,0x03,0x0b,0x12,0x20]
+// VI:   v_cndmask_b32_e64 v1, -v3, v5, s[4:5] ; encoding: [0x01,0x00,0x00,0xd1,0x03,0x0b,0x12,0x20]
+
+v_cndmask_b32_e64 v1, v3, |v5|, s[4:5]
+// SICI: v_cndmask_b32_e64 v1, v3, |v5|, s[4:5] ; encoding: [0x01,0x02,0x00,0xd2,0x03,0x0b,0x12,0x00]
+// VI:   v_cndmask_b32_e64 v1, v3, |v5|, s[4:5] ; encoding: [0x01,0x02,0x00,0xd1,0x03,0x0b,0x12,0x00]
+
+v_cndmask_b32_e64 v1, -abs(v3), v5, vcc
+// SICI: v_cndmask_b32_e64 v1, -|v3|, v5, vcc ; encoding: [0x01,0x01,0x00,0xd2,0x03,0x0b,0xaa,0x21]
+// VI:   v_cndmask_b32_e64 v1, -|v3|, v5, vcc ; encoding: [0x01,0x01,0x00,0xd1,0x03,0x0b,0xaa,0x21]
+
 //TODO: readlane, writelane
 
 v_add_f32 v1, v3, s5

diff --git a/llvm/test/MC/Disassembler/AMDGPU/vop3_vi.txt b/llvm/test/MC/Disassembler/AMDGPU/vop3_vi.txt
index abc24a1..99f940f 100644 (file)
--- a/llvm/test/MC/Disassembler/AMDGPU/vop3_vi.txt
+++ b/llvm/test/MC/Disassembler/AMDGPU/vop3_vi.txt
@@ -132,6 +132,15 @@
 # VI:   v_cndmask_b32_e64 v1, v3, v5, vcc ; encoding: [0x01,0x00,0x00,0xd1,0x03,0x0b,0xaa,0x01]
 0x01 0x00 0x00 0xd1 0x03 0x0b 0xaa 0x01
 
+# VI:   v_cndmask_b32_e64 v1, -v3, v5, s[4:5] ; encoding: [0x01,0x00,0x00,0xd1,0x03,0x0b,0x12,0x20]
+0x01,0x00,0x00,0xd1,0x03,0x0b,0x12,0x20
+
+# VI:   v_cndmask_b32_e64 v1, v3, |v5|, s[4:5] ; encoding: [0x01,0x02,0x00,0xd1,0x03,0x0b,0x12,0x00]
+0x01,0x02,0x00,0xd1,0x03,0x0b,0x12,0x00
+
+# VI:   v_cndmask_b32_e64 v1, -|v3|, v5, vcc ; encoding: [0x01,0x01,0x00,0xd1,0x03,0x0b,0xaa,0x21]
+0x01,0x01,0x00,0xd1,0x03,0x0b,0xaa,0x21
+
 # VI:   v_add_f32_e64 v1, v3, s5 ; encoding: [0x01,0x00,0x01,0xd1,0x03,0x0b,0x00,0x00]
 0x01 0x00 0x01 0xd1 0x03 0x0b 0x00 0x00