[ARM] Pass half or i16 types for NEON intrinsics
For generating NEON intrinsics, this determines the NEON data type, and whether
it should be a half type or an i16 type. I.e., we always pass a half type for
AArch64, this hasn't changed, but now also for ARM but only when FullFP16 is
enabled, and i16 otherwise.
This is intended to be non-functional change, but together with the backend
work in D44538 which adds support for f16 vectors, this enables adding the
AArch32 FP16 (vector) intrinsics.
Differential Revision: https://reviews.llvm.org/D44561
llvm-svn: 327836
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 49dfc60..7186e88 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -3441,7 +3441,7 @@
static llvm::VectorType *GetNeonType(CodeGenFunction *CGF,
NeonTypeFlags TypeFlags,
- llvm::Triple::ArchType Arch,
+ bool HasLegalHalfType=true,
bool V1Ty=false) {
int IsQuad = TypeFlags.isQuad();
switch (TypeFlags.getEltType()) {
@@ -3452,9 +3452,7 @@
case NeonTypeFlags::Poly16:
return llvm::VectorType::get(CGF->Int16Ty, V1Ty ? 1 : (4 << IsQuad));
case NeonTypeFlags::Float16:
- // FIXME: Only AArch64 backend can so far properly handle half types.
- // Remove else part once ARM backend support for half is complete.
- if (Arch == llvm::Triple::aarch64)
+ if (HasLegalHalfType)
return llvm::VectorType::get(CGF->HalfTy, V1Ty ? 1 : (4 << IsQuad));
else
return llvm::VectorType::get(CGF->Int16Ty, V1Ty ? 1 : (4 << IsQuad));
@@ -4338,8 +4336,9 @@
NeonTypeFlags Type(NeonTypeConst.getZExtValue());
bool Usgn = Type.isUnsigned();
bool Quad = Type.isQuad();
+ const bool HasLegalHalfType = getTarget().hasLegalHalfType();
- llvm::VectorType *VTy = GetNeonType(this, Type, Arch);
+ llvm::VectorType *VTy = GetNeonType(this, Type, HasLegalHalfType);
llvm::Type *Ty = VTy;
if (!Ty)
return nullptr;
@@ -4413,13 +4412,15 @@
case NEON::BI__builtin_neon_vcvt_f32_v:
case NEON::BI__builtin_neon_vcvtq_f32_v:
Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
- Ty = GetNeonType(this, NeonTypeFlags(NeonTypeFlags::Float32, false, Quad), Arch);
+ Ty = GetNeonType(this, NeonTypeFlags(NeonTypeFlags::Float32, false, Quad),
+ HasLegalHalfType);
return Usgn ? Builder.CreateUIToFP(Ops[0], Ty, "vcvt")
: Builder.CreateSIToFP(Ops[0], Ty, "vcvt");
case NEON::BI__builtin_neon_vcvt_f16_v:
case NEON::BI__builtin_neon_vcvtq_f16_v:
Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
- Ty = GetNeonType(this, NeonTypeFlags(NeonTypeFlags::Float16, false, Quad), Arch);
+ Ty = GetNeonType(this, NeonTypeFlags(NeonTypeFlags::Float16, false, Quad),
+ HasLegalHalfType);
return Usgn ? Builder.CreateUIToFP(Ops[0], Ty, "vcvt")
: Builder.CreateSIToFP(Ops[0], Ty, "vcvt");
case NEON::BI__builtin_neon_vcvt_n_f16_v:
@@ -5528,7 +5529,8 @@
bool usgn = Type.isUnsigned();
bool rightShift = false;
- llvm::VectorType *VTy = GetNeonType(this, Type, Arch);
+ llvm::VectorType *VTy = GetNeonType(this, Type,
+ getTarget().hasLegalHalfType());
llvm::Type *Ty = VTy;
if (!Ty)
return nullptr;
@@ -5774,7 +5776,7 @@
// Determine the type of this overloaded NEON intrinsic.
NeonTypeFlags Type(Result.getZExtValue());
- llvm::VectorType *Ty = GetNeonType(&CGF, Type, Arch);
+ llvm::VectorType *Ty = GetNeonType(&CGF, Type);
if (!Ty)
return nullptr;
@@ -6828,7 +6830,7 @@
}
}
- llvm::VectorType *VTy = GetNeonType(this, Type, Arch);
+ llvm::VectorType *VTy = GetNeonType(this, Type);
llvm::Type *Ty = VTy;
if (!Ty)
return nullptr;
@@ -6893,7 +6895,7 @@
Ops[0] = Builder.CreateBitCast(Ops[0], DoubleTy);
Ops[1] = Builder.CreateBitCast(Ops[1], DoubleTy);
llvm::Type *VTy = GetNeonType(this,
- NeonTypeFlags(NeonTypeFlags::Float64, false, true), Arch);
+ NeonTypeFlags(NeonTypeFlags::Float64, false, true));
Ops[2] = Builder.CreateBitCast(Ops[2], VTy);
Ops[2] = Builder.CreateExtractElement(Ops[2], Ops[3], "extract");
Value *F = CGM.getIntrinsic(Intrinsic::fma, DoubleTy);
@@ -7132,14 +7134,14 @@
case NEON::BI__builtin_neon_vcvt_f64_v:
case NEON::BI__builtin_neon_vcvtq_f64_v:
Ops[0] = Builder.CreateBitCast(Ops[0], Ty);
- Ty = GetNeonType(this, NeonTypeFlags(NeonTypeFlags::Float64, false, quad), Arch);
+ Ty = GetNeonType(this, NeonTypeFlags(NeonTypeFlags::Float64, false, quad));
return usgn ? Builder.CreateUIToFP(Ops[0], Ty, "vcvt")
: Builder.CreateSIToFP(Ops[0], Ty, "vcvt");
case NEON::BI__builtin_neon_vcvt_f64_f32: {
assert(Type.getEltType() == NeonTypeFlags::Float64 && quad &&
"unexpected vcvt_f64_f32 builtin");
NeonTypeFlags SrcFlag = NeonTypeFlags(NeonTypeFlags::Float32, false, false);
- Ops[0] = Builder.CreateBitCast(Ops[0], GetNeonType(this, SrcFlag, Arch));
+ Ops[0] = Builder.CreateBitCast(Ops[0], GetNeonType(this, SrcFlag));
return Builder.CreateFPExt(Ops[0], Ty, "vcvt");
}
@@ -7147,7 +7149,7 @@
assert(Type.getEltType() == NeonTypeFlags::Float32 &&
"unexpected vcvt_f32_f64 builtin");
NeonTypeFlags SrcFlag = NeonTypeFlags(NeonTypeFlags::Float64, false, true);
- Ops[0] = Builder.CreateBitCast(Ops[0], GetNeonType(this, SrcFlag, Arch));
+ Ops[0] = Builder.CreateBitCast(Ops[0], GetNeonType(this, SrcFlag));
return Builder.CreateFPTrunc(Ops[0], Ty, "vcvt");
}
@@ -7244,7 +7246,7 @@
Quad = true;
Ops[0] = Builder.CreateBitCast(Ops[0], DoubleTy);
llvm::Type *VTy = GetNeonType(this,
- NeonTypeFlags(NeonTypeFlags::Float64, false, Quad), Arch);
+ NeonTypeFlags(NeonTypeFlags::Float64, false, Quad));
Ops[1] = Builder.CreateBitCast(Ops[1], VTy);
Ops[1] = Builder.CreateExtractElement(Ops[1], Ops[2], "extract");
Value *Result = Builder.CreateFMul(Ops[0], Ops[1]);