[X86] Add builtins for pshufd, pshuflw, and pshufhw to enable target feature and immediate range checking.
llvm-svn: 334265
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 025b34e..741f36b 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -9325,6 +9325,57 @@
makeArrayRef(Indices, NumElts),
"blend");
}
+ case X86::BI__builtin_ia32_pshuflw:
+ case X86::BI__builtin_ia32_pshuflw256:
+ case X86::BI__builtin_ia32_pshuflw512: {
+ uint32_t Imm = cast<llvm::ConstantInt>(Ops[1])->getZExtValue();
+ llvm::Type *Ty = Ops[0]->getType();
+ unsigned NumElts = Ty->getVectorNumElements();
+
+ // Splat the 8-bits of immediate 4 times to help the loop wrap around.
+ Imm = (Imm & 0xff) * 0x01010101;
+
+ uint32_t Indices[32];
+ for (unsigned l = 0; l != NumElts; l += 8) {
+ for (unsigned i = 0; i != 4; ++i) {
+ Indices[l + i] = l + (Imm & 3);
+ Imm >>= 2;
+ }
+ for (unsigned i = 4; i != 8; ++i)
+ Indices[l + i] = l + i;
+ }
+
+ return Builder.CreateShuffleVector(Ops[0], UndefValue::get(Ty),
+ makeArrayRef(Indices, NumElts),
+ "pshuflw");
+ }
+ case X86::BI__builtin_ia32_pshufhw:
+ case X86::BI__builtin_ia32_pshufhw256:
+ case X86::BI__builtin_ia32_pshufhw512: {
+ uint32_t Imm = cast<llvm::ConstantInt>(Ops[1])->getZExtValue();
+ llvm::Type *Ty = Ops[0]->getType();
+ unsigned NumElts = Ty->getVectorNumElements();
+
+ // Splat the 8-bits of immediate 4 times to help the loop wrap around.
+ Imm = (Imm & 0xff) * 0x01010101;
+
+ uint32_t Indices[32];
+ for (unsigned l = 0; l != NumElts; l += 8) {
+ for (unsigned i = 0; i != 4; ++i)
+ Indices[l + i] = l + i;
+ for (unsigned i = 4; i != 8; ++i) {
+ Indices[l + i] = l + 4 + (Imm & 3);
+ Imm >>= 2;
+ }
+ }
+
+ return Builder.CreateShuffleVector(Ops[0], UndefValue::get(Ty),
+ makeArrayRef(Indices, NumElts),
+ "pshufhw");
+ }
+ case X86::BI__builtin_ia32_pshufd:
+ case X86::BI__builtin_ia32_pshufd256:
+ case X86::BI__builtin_ia32_pshufd512:
case X86::BI__builtin_ia32_vpermilpd:
case X86::BI__builtin_ia32_vpermilps:
case X86::BI__builtin_ia32_vpermilpd256: