[10/11][POC][Clang][RISCV] Define vget for tuple type

[eopXD]

Summary

For the cover letter of this patch-set, please checkout D146872.

Depends on D147915.

This is the 10th patch of the patch-set.

This patch is a proof-of-concept and will be extended to full coverage
in the future. Only vget for tuple type of NF=2, EEW=32, LMUL=1 is
defined now.

[craig.topper]

clang/include/clang/Basic/riscv_vector.td

2545

Does SemaChecking.cpp already guarantee Index is less than Maxindex?

[craig.topper]

clang/include/clang/Basic/riscv_vector.td

2545

Actually it looks like maybe SemaChecking considers MaxIndex to be LMUL*NF? Does that mean vget can extract an LMUL1 value from an LMUL4 tuple?

[eopXD]

Add test case for out of range vget and address comment from Craig.

[eopXD]

clang/include/clang/Basic/riscv_vector.td

2545

Modified code under SemaChecking.cpp for vector tuple types.

[eopXD]

Rebase and fix test case failure by only defining intrinsics for vint32m1x2_t.

[craig.topper]

clang/include/clang/Basic/riscv_vector.td

2542

You can drop this assert. cast already has the same assert.

2545

Does this min do anything or does SemaChecking already guarantee it?

clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vget_tuple.c

19

index argument is unused

clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vget-index-out-of-range.c

343

index argument is unused

[eopXD]

Address comments from Craig.`

clang/include/clang/Basic/riscv_vector.td

2545

Yes, SemaChecking should be able to guard the value correctly and this line is redundant.

[eopXD]

Update code.

[craig.topper]

LGTM

[eopXD]

Rebase.

[eopXD]

Add unsigned cast to eliminate compile warning.

clang/include/clang/Basic/riscv_vector.td

foreach dst_lmul = ["(LFixedLog2LMUL:-2)", "(LFixedLog2LMUL:-1)", "(LFixedLog2LMUL:-0)",
       def vlmul_ext_u # dst_lmul : RVVBuiltin<"Uv" # dst_lmul # "Uv",
                                               dst_lmul # "UvUv", "csil", dst_lmul # "Uv">;
     }
   }
 
   let Name = "vget_v", MaskedPolicyScheme = NonePolicy,
       ManualCodegen = [{
       {
+        if (isa<StructType>(Ops[0]->getType())) // For tuple type
+          // Extract value from index (operand 1) of vtuple (operand 0)
+          return Builder.CreateExtractValue(
+            Ops[0],
+            {(unsigned)cast<ConstantInt>(Ops[1])->getZExtValue()});
         auto *VecTy = cast<ScalableVectorType>(ResultType);
         auto *OpVecTy = cast<ScalableVectorType>(Ops[0]->getType());
         // Mask to only valid indices.
         unsigned MaxIndex = OpVecTy->getMinNumElements() / VecTy->getMinNumElements();
         assert(isPowerOf2_32(MaxIndex));
         Ops[1] = Builder.CreateZExt(Ops[1], Builder.getInt64Ty());
         Ops[1] = Builder.CreateAnd(Ops[1], MaxIndex - 1);
         Ops[1] = Builder.CreateMul(Ops[1],
                                    ConstantInt::get(Ops[1]->getType(),
                                                     VecTy->getMinNumElements()));
         return Builder.CreateExtractVector(ResultType, Ops[0], Ops[1]);
       }
       }] in {
     foreach dst_lmul = ["(SFixedLog2LMUL:0)", "(SFixedLog2LMUL:1)", "(SFixedLog2LMUL:2)"] in {
       def : RVVBuiltin<"v" # dst_lmul # "v", dst_lmul # "vvKz", "csilxfd", dst_lmul # "v">;
       def : RVVBuiltin<"Uv" # dst_lmul # "Uv", dst_lmul # "UvUvKz", "csil", dst_lmul # "Uv">;
     }
+    foreach nf = [2] in {
+      let Log2LMUL = [0] in {
+        defvar T = "(Tuple:" # nf # ")";
+        def : RVVBuiltin<T # "vv", "v" # T # "vKz", "i", "v">;
+      }
+    }
   }
 
   let Name = "vset_v", Log2LMUL = [0, 1, 2], MaskedPolicyScheme = NonePolicy,
       ManualCodegen = [{
       {
         auto *ResVecTy = cast<ScalableVectorType>(ResultType);
         auto *VecTy = cast<ScalableVectorType>(Ops[2]->getType());
         // Mask to only valid indices.

clang/lib/Sema/SemaChecking.cpp

return SemaBuiltinConstantArgRange(TheCall, 0, 0, 3) ||
            CheckRISCVLMUL(TheCall, 1);
   case RISCVVector::BI__builtin_rvv_vget_v: {
     ASTContext::BuiltinVectorTypeInfo ResVecInfo =
         Context.getBuiltinVectorTypeInfo(cast<BuiltinType>(
             TheCall->getType().getCanonicalType().getTypePtr()));
     ASTContext::BuiltinVectorTypeInfo VecInfo =
         Context.getBuiltinVectorTypeInfo(cast<BuiltinType>(
             TheCall->getArg(0)->getType().getCanonicalType().getTypePtr()));
-    unsigned MaxIndex =
-        (VecInfo.EC.getKnownMinValue() * VecInfo.NumVectors) /
+    unsigned MaxIndex;
+    if (VecInfo.NumVectors != 1) // vget for tuple type
+      MaxIndex = VecInfo.NumVectors;
+    else // vget for non-tuple type
+      MaxIndex = (VecInfo.EC.getKnownMinValue() * VecInfo.NumVectors) /
                  (ResVecInfo.EC.getKnownMinValue() * ResVecInfo.NumVectors);
     return SemaBuiltinConstantArgRange(TheCall, 1, 0, MaxIndex - 1);
   }
   case RISCVVector::BI__builtin_rvv_vset_v: {
     ASTContext::BuiltinVectorTypeInfo ResVecInfo =
         Context.getBuiltinVectorTypeInfo(cast<BuiltinType>(
             TheCall->getType().getCanonicalType().getTypePtr()));
     ASTContext::BuiltinVectorTypeInfo VecInfo =
         Context.getBuiltinVectorTypeInfo(cast<BuiltinType>(

clang/lib/Support/RISCVVIntrinsicUtils.cpp

void RVVType::initShortStr() {
   case ScalarTypeKind::UnsignedInteger:
     ShortStr = "u" + utostr(ElementBitwidth);
     break;
   default:
     llvm_unreachable("Unhandled case!");
   }
   if (isVector())
     ShortStr += LMUL.str();
+  if (isTuple())
+    ShortStr += "x" + utostr(NF);
 }
 
 void RVVType::applyBasicType() {
   switch (BT) {
   case BasicType::Int8:
     ElementBitwidth = 8;
     ScalarType = ScalarTypeKind::SignedInteger;
     break;

clang/test/CodeGen/RISCV/rvv-intrinsics-autogenerated/non-policy/non-overloaded/vget_tuple.c

+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// REQUIRES: riscv-registered-target
+// RUN: %clang_cc1 -triple riscv64 -target-feature +v -target-feature +zfh \
+// RUN:   -target-feature +experimental-zvfh -disable-O0-optnone  \
+// RUN:   -emit-llvm %s -o - | opt -S -passes=mem2reg | \
+// RUN:   FileCheck --check-prefix=CHECK-RV64 %s
+
+#include <riscv_vector.h>
+
+// CHECK-RV64-LABEL: define dso_local <vscale x 2 x i32> @test_vget_v_i32m1x2_i32m1
+// CHECK-RV64-SAME: (<vscale x 2 x i32> [[SRC_COERCE0:%.*]], <vscale x 2 x i32> [[SRC_COERCE1:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-RV64-NEXT:  entry:
+// CHECK-RV64-NEXT:    [[TMP0:%.*]] = insertvalue { <vscale x 2 x i32>, <vscale x 2 x i32> } poison, <vscale x 2 x i32> [[SRC_COERCE0]], 0
+// CHECK-RV64-NEXT:    [[TMP1:%.*]] = insertvalue { <vscale x 2 x i32>, <vscale x 2 x i32> } [[TMP0]], <vscale x 2 x i32> [[SRC_COERCE1]], 1
+// CHECK-RV64-NEXT:    [[TMP2:%.*]] = extractvalue { <vscale x 2 x i32>, <vscale x 2 x i32> } [[TMP1]], 0
+// CHECK-RV64-NEXT:    ret <vscale x 2 x i32> [[TMP2]]
+//
+vint32m1_t test_vget_v_i32m1x2_i32m1(vint32m1x2_t src) {
+  return __riscv_vget_v_i32m1x2_i32m1(src, 0);
+}

clang/test/CodeGen/RISCV/rvv-intrinsics-handcrafted/vget-index-out-of-range.c

vfloat16m2_t test_vget_v_f16m8_f16m2(vfloat16m8_t src) {
   // expected-error@+1 {{argument value 4 is outside the valid range [0, 3]}}
   return __riscv_vget_v_f16m8_f16m2(src, 4);
 }
 
 vfloat16m4_t test_vget_v_f16m8_f16m4(vfloat16m8_t src) {
   // expected-error@+1 {{argument value 2 is outside the valid range [0, 1]}}
   return __riscv_vget_v_f16m8_f16m4(src, 2);
 }
+
+vint32m1_t test_vget_v_i32m1x2_i32m1(vint32m1x2_t src) {
+  // expected-error@+1 {{argument value 2 is outside the valid range [0, 1]}}
+  return __riscv_vget_v_i32m1x2_i32m1(src, 2);
+}