From 5e249b4a14a1553039de338e5e50a1656605a7ea Mon Sep 17 00:00:00 2001 From: Bill Wendling Date: Thu, 2 Oct 2008 05:56:52 +0000 Subject: [PATCH] "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 --- include/llvm/IntrinsicsX86.td | 6 ++++++ lib/Target/X86/X86InstrSSE.td | 4 ++++ 2 files changed, 10 insertions(+) diff --git a/include/llvm/IntrinsicsX86.td b/include/llvm/IntrinsicsX86.td index cc263e49d62..682eb51bcdf 100644 --- 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 8b4579517e2..93f722332bd 100644 --- 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)))>; -- 2.34.1