[x86] Add the (v)cmpps/pd/ss/sd builtins to match gcc. Use them in the sse intrinsic files.

This still lower to the same intrinsics as before.

This is preparation for bounds checking the immediate on the avx version of the builtin so we don't pass illegal immediates into the backend. Since SSE uses a smaller size immediate its not possible to bounds check when using a shared builtin. Rather than creating a clang specific builtin for the different immediate, I decided (after consulting with Chandler) that it was better to match gcc.

llvm-svn: 224879

1 files changed, 150 insertions, 2 deletions

diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index fca60d50c56..1665460ee82 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -5986,8 +5986,8 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
   // 3DNow!
   case X86::BI__builtin_ia32_pswapdsf:
   case X86::BI__builtin_ia32_pswapdsi: {
-    const char *name = nullptr;
-    Intrinsic::ID ID = Intrinsic::not_intrinsic;
+    const char *name;
+    Intrinsic::ID ID;
     switch(BuiltinID) {
     default: llvm_unreachable("Unsupported intrinsic!");
     case X86::BI__builtin_ia32_pswapdsf:
@@ -6041,6 +6041,154 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
     Value *F = CGM.getIntrinsic(Intrinsic::x86_avx2_vbroadcasti128);
     return Builder.CreateCall(F, Builder.CreateBitCast(VecTmp, Int8PtrTy));
   }
+  // SSE comparison intrisics
+  case X86::BI__builtin_ia32_cmpeqps:
+  case X86::BI__builtin_ia32_cmpltps:
+  case X86::BI__builtin_ia32_cmpleps:
+  case X86::BI__builtin_ia32_cmpunordps:
+  case X86::BI__builtin_ia32_cmpneqps:
+  case X86::BI__builtin_ia32_cmpnltps:
+  case X86::BI__builtin_ia32_cmpnleps:
+  case X86::BI__builtin_ia32_cmpordps:
+  case X86::BI__builtin_ia32_cmpeqss:
+  case X86::BI__builtin_ia32_cmpltss:
+  case X86::BI__builtin_ia32_cmpless:
+  case X86::BI__builtin_ia32_cmpunordss:
+  case X86::BI__builtin_ia32_cmpneqss:
+  case X86::BI__builtin_ia32_cmpnltss:
+  case X86::BI__builtin_ia32_cmpnless:
+  case X86::BI__builtin_ia32_cmpordss:
+  case X86::BI__builtin_ia32_cmpeqpd:
+  case X86::BI__builtin_ia32_cmpltpd:
+  case X86::BI__builtin_ia32_cmplepd:
+  case X86::BI__builtin_ia32_cmpunordpd:
+  case X86::BI__builtin_ia32_cmpneqpd:
+  case X86::BI__builtin_ia32_cmpnltpd:
+  case X86::BI__builtin_ia32_cmpnlepd:
+  case X86::BI__builtin_ia32_cmpordpd:
+  case X86::BI__builtin_ia32_cmpeqsd:
+  case X86::BI__builtin_ia32_cmpltsd:
+  case X86::BI__builtin_ia32_cmplesd:
+  case X86::BI__builtin_ia32_cmpunordsd:
+  case X86::BI__builtin_ia32_cmpneqsd:
+  case X86::BI__builtin_ia32_cmpnltsd:
+  case X86::BI__builtin_ia32_cmpnlesd:
+  case X86::BI__builtin_ia32_cmpordsd:
+    // These exist so that the builtin that takes an immediate can be bounds
+    // checked by clang to avoid passing bad immediates to the backend. Since
+    // AVX has a larger immediate than SSE we would need separate builtins to
+    // do the different bounds checking. Rather than create a clang specific
+    // SSE only builtin, this implements eight separate builtins to match gcc
+    // implementation.
+
+    // Choose the immediate.
+    unsigned Imm;
+    switch (BuiltinID) {
+    default: llvm_unreachable("Unsupported intrinsic!");
+    case X86::BI__builtin_ia32_cmpeqps:
+    case X86::BI__builtin_ia32_cmpeqss:
+    case X86::BI__builtin_ia32_cmpeqpd:
+    case X86::BI__builtin_ia32_cmpeqsd:
+      Imm = 0;
+      break;
+    case X86::BI__builtin_ia32_cmpltps:
+    case X86::BI__builtin_ia32_cmpltss:
+    case X86::BI__builtin_ia32_cmpltpd:
+    case X86::BI__builtin_ia32_cmpltsd:
+      Imm = 1;
+      break;
+    case X86::BI__builtin_ia32_cmpleps:
+    case X86::BI__builtin_ia32_cmpless:
+    case X86::BI__builtin_ia32_cmplepd:
+    case X86::BI__builtin_ia32_cmplesd:
+      Imm = 2;
+      break;
+    case X86::BI__builtin_ia32_cmpunordps:
+    case X86::BI__builtin_ia32_cmpunordss:
+    case X86::BI__builtin_ia32_cmpunordpd:
+    case X86::BI__builtin_ia32_cmpunordsd:
+      Imm = 3;
+      break;
+    case X86::BI__builtin_ia32_cmpneqps:
+    case X86::BI__builtin_ia32_cmpneqss:
+    case X86::BI__builtin_ia32_cmpneqpd:
+    case X86::BI__builtin_ia32_cmpneqsd:
+      Imm = 4;
+      break;
+    case X86::BI__builtin_ia32_cmpnltps:
+    case X86::BI__builtin_ia32_cmpnltss:
+    case X86::BI__builtin_ia32_cmpnltpd:
+    case X86::BI__builtin_ia32_cmpnltsd:
+      Imm = 5;
+      break;
+    case X86::BI__builtin_ia32_cmpnleps:
+    case X86::BI__builtin_ia32_cmpnless:
+    case X86::BI__builtin_ia32_cmpnlepd:
+    case X86::BI__builtin_ia32_cmpnlesd:
+      Imm = 6;
+      break;
+    case X86::BI__builtin_ia32_cmpordps:
+    case X86::BI__builtin_ia32_cmpordss:
+    case X86::BI__builtin_ia32_cmpordpd:
+    case X86::BI__builtin_ia32_cmpordsd:
+      Imm = 7;
+      break;
+    }
+
+    // Choose the intrinsic ID.
+    const char *name;
+    Intrinsic::ID ID;
+    switch (BuiltinID) {
+    default: llvm_unreachable("Unsupported intrinsic!");
+    case X86::BI__builtin_ia32_cmpeqps:
+    case X86::BI__builtin_ia32_cmpltps:
+    case X86::BI__builtin_ia32_cmpleps:
+    case X86::BI__builtin_ia32_cmpunordps:
+    case X86::BI__builtin_ia32_cmpneqps:
+    case X86::BI__builtin_ia32_cmpnltps:
+    case X86::BI__builtin_ia32_cmpnleps:
+    case X86::BI__builtin_ia32_cmpordps:
+      name = "cmpps";
+      ID = Intrinsic::x86_sse_cmp_ps;
+      break;
+    case X86::BI__builtin_ia32_cmpeqss:
+    case X86::BI__builtin_ia32_cmpltss:
+    case X86::BI__builtin_ia32_cmpless:
+    case X86::BI__builtin_ia32_cmpunordss:
+    case X86::BI__builtin_ia32_cmpneqss:
+    case X86::BI__builtin_ia32_cmpnltss:
+    case X86::BI__builtin_ia32_cmpnless:
+    case X86::BI__builtin_ia32_cmpordss:
+      name = "cmpss";
+      ID = Intrinsic::x86_sse_cmp_ss;
+      break;
+    case X86::BI__builtin_ia32_cmpeqpd:
+    case X86::BI__builtin_ia32_cmpltpd:
+    case X86::BI__builtin_ia32_cmplepd:
+    case X86::BI__builtin_ia32_cmpunordpd:
+    case X86::BI__builtin_ia32_cmpneqpd:
+    case X86::BI__builtin_ia32_cmpnltpd:
+    case X86::BI__builtin_ia32_cmpnlepd:
+    case X86::BI__builtin_ia32_cmpordpd:
+      name = "cmppd";
+      ID = Intrinsic::x86_sse2_cmp_pd;
+      break;
+    case X86::BI__builtin_ia32_cmpeqsd:
+    case X86::BI__builtin_ia32_cmpltsd:
+    case X86::BI__builtin_ia32_cmplesd:
+    case X86::BI__builtin_ia32_cmpunordsd:
+    case X86::BI__builtin_ia32_cmpneqsd:
+    case X86::BI__builtin_ia32_cmpnltsd:
+    case X86::BI__builtin_ia32_cmpnlesd:
+    case X86::BI__builtin_ia32_cmpordsd:
+      name = "cmpsd";
+      ID = Intrinsic::x86_sse2_cmp_sd;
+      break;
+    }
+
+    Ops.push_back(llvm::ConstantInt::get(Int8Ty, Imm));
+    llvm::Function *F = CGM.getIntrinsic(ID);
+    return Builder.CreateCall(F, Ops, name);
   }
 }