Remove the 'unaligned load' builtins now that they're no longer used in the *mmintrin...

git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@131300 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/clang/Basic/BuiltinsX86.def b/include/clang/Basic/BuiltinsX86.def
index ce376dd3d71939cdd5a00b4139966025c2717824..6ef667db7beb424ad7d4bebb64a8e3105b282d42 100644 (file)
--- a/include/clang/Basic/BuiltinsX86.def
+++ b/include/clang/Basic/BuiltinsX86.def
@@ -240,7 +240,6 @@ BUILTIN(__builtin_ia32_cvtps2pi, "V2iV4f", "")
 BUILTIN(__builtin_ia32_cvtss2si, "iV4f", "")
 BUILTIN(__builtin_ia32_cvtss2si64, "LLiV4f", "")
 BUILTIN(__builtin_ia32_cvttps2pi, "V2iV4f", "")
-BUILTIN(__builtin_ia32_loadups, "V4ffC*", "")
 BUILTIN(__builtin_ia32_storeups, "vf*V4f", "")
 BUILTIN(__builtin_ia32_storehps, "vV2i*V4f", "")
 BUILTIN(__builtin_ia32_storelps, "vV2i*V4f", "")
@@ -254,7 +253,6 @@ BUILTIN(__builtin_ia32_rsqrtss, "V4fV4f", "")
 BUILTIN(__builtin_ia32_sqrtps, "V4fV4f", "")
 BUILTIN(__builtin_ia32_sqrtss, "V4fV4f", "")
 BUILTIN(__builtin_ia32_maskmovdqu, "vV16cV16cc*", "")
-BUILTIN(__builtin_ia32_loadupd, "V2ddC*", "")
 BUILTIN(__builtin_ia32_storeupd, "vd*V2d", "")
 BUILTIN(__builtin_ia32_movmskpd, "iV2d", "")
 BUILTIN(__builtin_ia32_pmovmskb128, "iV16c", "")
@@ -280,7 +278,6 @@ BUILTIN(__builtin_ia32_cvttps2dq, "V4iV4f", "")
 BUILTIN(__builtin_ia32_clflush, "vvC*", "")
 BUILTIN(__builtin_ia32_lfence, "v", "")
 BUILTIN(__builtin_ia32_mfence, "v", "")
-BUILTIN(__builtin_ia32_loaddqu, "V16ccC*", "")
 BUILTIN(__builtin_ia32_storedqu, "vc*V16c", "")
 BUILTIN(__builtin_ia32_pmuludq128, "V2LLiV4iV4i", "")
 BUILTIN(__builtin_ia32_psraw128, "V8sV8sV8s", "")

diff --git a/lib/CodeGen/CGBuiltin.cpp b/lib/CodeGen/CGBuiltin.cpp
index 494dfaeff7759b5697bd98615381f766036a5514..9d6c20f12f0cde154c92bc5a065abc8d210e7573 100644 (file)
--- a/lib/CodeGen/CGBuiltin.cpp
+++ b/lib/CodeGen/CGBuiltin.cpp
@@ -2143,19 +2143,6 @@ Value *CodeGenFunction::EmitX86BuiltinExpr(unsigned BuiltinID,
     // If palignr is shifting the pair of vectors more than 32 bytes, emit zero.
     return llvm::Constant::getNullValue(ConvertType(E->getType()));
   }
-  case X86::BI__builtin_ia32_loadups:
-  case X86::BI__builtin_ia32_loadupd:
-  case X86::BI__builtin_ia32_loaddqu: {
-    const llvm::Type *VecTy = ConvertType(E->getType());
-    const llvm::Type *IntTy = llvm::IntegerType::get(getLLVMContext(), 128);
-
-    Value *BC = Builder.CreateBitCast(Ops[0],
-                                      llvm::PointerType::getUnqual(IntTy),
-                                      "cast");
-    LoadInst *LI = Builder.CreateLoad(BC);
-    LI->setAlignment(1); // Unaligned load.
-    return Builder.CreateBitCast(LI, VecTy, "loadu.cast");
-  }
   case X86::BI__builtin_ia32_movntps:
   case X86::BI__builtin_ia32_movntpd:
   case X86::BI__builtin_ia32_movntdq:

diff --git a/test/CodeGen/builtins-x86.c b/test/CodeGen/builtins-x86.c
index 190fa55c5756e448a814c30a3d9cfe84a3ab29cd..728ade34c10b0ed5422a053375af74bc3de30eff 100644 (file)
--- a/test/CodeGen/builtins-x86.c
+++ b/test/CodeGen/builtins-x86.c
@@ -273,7 +273,6 @@ void f0() {
 #endif
   tmp_V2i = __builtin_ia32_cvttps2pi(tmp_V4f);
   (void) __builtin_ia32_maskmovq(tmp_V8c, tmp_V8c, tmp_cp);
-  tmp_V4f = __builtin_ia32_loadups(tmp_fCp);
   (void) __builtin_ia32_storeups(tmp_fp, tmp_V4f);
   (void) __builtin_ia32_storehps(tmp_V2ip, tmp_V4f);
   (void) __builtin_ia32_storelps(tmp_V2ip, tmp_V4f);
@@ -291,7 +290,6 @@ void f0() {
   tmp_V4f = __builtin_ia32_sqrtps(tmp_V4f);
   tmp_V4f = __builtin_ia32_sqrtss(tmp_V4f);
   (void) __builtin_ia32_maskmovdqu(tmp_V16c, tmp_V16c, tmp_cp);
-  tmp_V2d = __builtin_ia32_loadupd(tmp_dCp);
   (void) __builtin_ia32_storeupd(tmp_dp, tmp_V2d);
   tmp_i = __builtin_ia32_movmskpd(tmp_V2d);
   tmp_i = __builtin_ia32_pmovmskb128(tmp_V16c);
@@ -319,7 +317,6 @@ void f0() {
   (void) __builtin_ia32_clflush(tmp_vCp);
   (void) __builtin_ia32_lfence();
   (void) __builtin_ia32_mfence();
-  tmp_V16c = __builtin_ia32_loaddqu(tmp_cCp);
   (void) __builtin_ia32_storedqu(tmp_cp, tmp_V16c);
   tmp_V4s = __builtin_ia32_psllwi(tmp_V4s, tmp_i);
   tmp_V2i = __builtin_ia32_pslldi(tmp_V2i, tmp_i);