"The original bug was a complaint that _mm_srli_si128 mis-compiled when passed

a constant vector ("{0x123, 0x456}" syntax).  The fix is to simplify the
_mm_srli_si128 macro, and  move the "* 8" from the macro into the compiler
back-end.  I can't change the existing __builtins because so many people are
using them :-(."
Patch by Stuart Hastings!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@56944 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/llvm/IntrinsicsX86.td b/include/llvm/IntrinsicsX86.td
index cc263e49d620eea3ae7d5fdb8e3955538f7ba20e..682eb51bcdfd1b5d43a84d76c24a8324c474a125 100644 (file)
--- a/include/llvm/IntrinsicsX86.td
+++ b/include/llvm/IntrinsicsX86.td
@@ -371,6 +371,12 @@ let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
   def int_x86_sse2_psrl_dq : GCCBuiltin<"__builtin_ia32_psrldqi128">,
               Intrinsic<[llvm_v2i64_ty, llvm_v2i64_ty,
                          llvm_i32_ty], [IntrNoMem]>;
+  def int_x86_sse2_psll_dq_bs : GCCBuiltin<"__builtin_ia32_pslldqi128_byteshift">,
+              Intrinsic<[llvm_v2i64_ty, llvm_v2i64_ty,
+                         llvm_i32_ty], [IntrNoMem]>;
+  def int_x86_sse2_psrl_dq_bs : GCCBuiltin<"__builtin_ia32_psrldqi128_byteshift">,
+              Intrinsic<[llvm_v2i64_ty, llvm_v2i64_ty,
+                         llvm_i32_ty], [IntrNoMem]>;
 }
 
 // Integer comparison ops

diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index 8b4579517e24d99fb88d01d252c16cc4d63171f7..93f722332bdd9f0d60b3e805aedf08d2b0f1f203 100644 (file)
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -1969,6 +1969,10 @@ let Predicates = [HasSSE2] in {
             (v2i64 (PSLLDQri VR128:$src1, (PSxLDQ_imm imm:$src2)))>;
   def : Pat<(int_x86_sse2_psrl_dq VR128:$src1, imm:$src2),
             (v2i64 (PSRLDQri VR128:$src1, (PSxLDQ_imm imm:$src2)))>;
+  def : Pat<(int_x86_sse2_psll_dq_bs VR128:$src1, imm:$src2),
+            (v2i64 (PSLLDQri VR128:$src1, imm:$src2))>;
+  def : Pat<(int_x86_sse2_psrl_dq_bs VR128:$src1, imm:$src2),
+            (v2i64 (PSRLDQri VR128:$src1, imm:$src2))>;
   def : Pat<(v2f64 (X86fsrl VR128:$src1, i32immSExt8:$src2)),
             (v2f64 (PSRLDQri VR128:$src1, (PSxLDQ_imm imm:$src2)))>;