From: Simon Pilgrim <llvm-dev@redking.me.uk>
Date: Sat, 13 Jun 2015 14:08:15 +0000 (+0000)
Subject: [DAGCombiner] Added BSWAP vector constant folding support.
X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=d8b27b1cb14a2f6ce88505372ffc1eb9f5068efe;p=oota-llvm.git
[DAGCombiner] Added BSWAP vector constant folding support.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239675 91177308-0d34-0410-b5e6-96231b3b80d8
---
diff --git a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index 35943351665..1e33f1ed10f 100644
--- a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -255,6 +255,7 @@ namespace {
SDValue visitSRA(SDNode *N);
SDValue visitSRL(SDNode *N);
SDValue visitRotate(SDNode *N);
+ SDValue visitBSWAP(SDNode *N);
SDValue visitCTLZ(SDNode *N);
SDValue visitCTLZ_ZERO_UNDEF(SDNode *N);
SDValue visitCTTZ(SDNode *N);
@@ -1335,6 +1336,7 @@ SDValue DAGCombiner::visit(SDNode *N) {
case ISD::SRL: return visitSRL(N);
case ISD::ROTR:
case ISD::ROTL: return visitRotate(N);
+ case ISD::BSWAP: return visitBSWAP(N);
case ISD::CTLZ: return visitCTLZ(N);
case ISD::CTLZ_ZERO_UNDEF: return visitCTLZ_ZERO_UNDEF(N);
case ISD::CTTZ: return visitCTTZ(N);
@@ -4764,6 +4766,16 @@ SDValue DAGCombiner::visitSRL(SDNode *N) {
return SDValue();
}
+SDValue DAGCombiner::visitBSWAP(SDNode *N) {
+ SDValue N0 = N->getOperand(0);
+ EVT VT = N->getValueType(0);
+
+ // fold (bswap c1) -> c2
+ if (isConstantIntBuildVectorOrConstantInt(N0))
+ return DAG.getNode(ISD::BSWAP, SDLoc(N), VT, N0);
+ return SDValue();
+}
+
SDValue DAGCombiner::visitCTLZ(SDNode *N) {
SDValue N0 = N->getOperand(0);
EVT VT = N->getValueType(0);
diff --git a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
index cf51e756d84..27fa3d2b1e3 100644
--- a/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
+++ b/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
@@ -2932,6 +2932,7 @@ SDValue SelectionDAG::getNode(unsigned Opcode, SDLoc DL,
case ISD::TRUNCATE:
case ISD::UINT_TO_FP:
case ISD::SINT_TO_FP:
+ case ISD::BSWAP:
case ISD::CTLZ:
case ISD::CTLZ_ZERO_UNDEF:
case ISD::CTTZ:
diff --git a/test/CodeGen/X86/bswap-vector.ll b/test/CodeGen/X86/bswap-vector.ll
index 7d5f380c1e2..16da1626f01 100644
--- a/test/CodeGen/X86/bswap-vector.ll
+++ b/test/CodeGen/X86/bswap-vector.ll
@@ -1,6 +1,6 @@
-; RUN: llc < %s -mcpu=x86-64 | FileCheck %s --check-prefix=CHECK-NOSSSE3
-; RUN: llc < %s -mcpu=core2 | FileCheck %s --check-prefix=CHECK-SSSE3
-; RUN: llc < %s -mcpu=core-avx2 | FileCheck %s --check-prefix=CHECK-AVX2
+; RUN: llc < %s -mcpu=x86-64 | FileCheck %s --check-prefix=CHECK-ALL --check-prefix=CHECK-SSE --check-prefix=CHECK-NOSSSE3
+; RUN: llc < %s -mcpu=core2 | FileCheck %s --check-prefix=CHECK-ALL --check-prefix=CHECK-SSE --check-prefix=CHECK-SSSE3
+; RUN: llc < %s -mcpu=core-avx2 | FileCheck %s --check-prefix=CHECK-AVX --check-prefix=CHECK-AVX2
; RUN: llc < %s -mcpu=core-avx2 -x86-experimental-vector-widening-legalization | FileCheck %s --check-prefix=CHECK-WIDE-AVX2
target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
@@ -285,3 +285,100 @@ entry:
%r = call <4 x i16> @llvm.bswap.v4i16(<4 x i16> %v)
ret <4 x i16> %r
}
+
+;
+; Constant Folding
+;
+
+define <8 x i16> @fold_v8i16() {
+; CHECK-SSE-LABEL: fold_v8i16:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [0,256,65535,512,65023,1024,64511,1536]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v8i16:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [0,256,65535,512,65023,1024,64511,1536]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <8 x i16> @llvm.bswap.v8i16(<8 x i16> <i16 0, i16 1, i16 -1, i16 2, i16 -3, i16 4, i16 -5, i16 6>)
+ ret <8 x i16> %r
+}
+
+define <4 x i32> @fold_v4i32() {
+; CHECK-SSE-LABEL: fold_v4i32:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [0,4294967295,33554432,4261412863]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v4i32:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [0,4294967295,33554432,4261412863]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <4 x i32> @llvm.bswap.v4i32(<4 x i32> <i32 0, i32 -1, i32 2, i32 -3>)
+ ret <4 x i32> %r
+}
+
+define <2 x i64> @fold_v2i64() {
+; CHECK-SSE-LABEL: fold_v2i64:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [18374686479671623680,18446744073709551615]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v2i64:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} xmm0 = [18374686479671623680,18446744073709551615]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <2 x i64> @llvm.bswap.v2i64(<2 x i64> <i64 255, i64 -1>)
+ ret <2 x i64> %r
+}
+
+define <16 x i16> @fold_v16i16() {
+; CHECK-SSE-LABEL: fold_v16i16:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [0,256,65535,512,65023,1024,64511,1536]
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm1 = [63999,2048,63487,2560,62975,3072,62463,3584]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v16i16:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} ymm0 = [0,256,65535,512,65023,1024,64511,1536,63999,2048,63487,2560,62975,3072,62463,3584]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <16 x i16> @llvm.bswap.v16i16(<16 x i16> <i16 0, i16 1, i16 -1, i16 2, i16 -3, i16 4, i16 -5, i16 6, i16 -7, i16 8, i16 -9, i16 10, i16 -11, i16 12, i16 -13, i16 14>)
+ ret <16 x i16> %r
+}
+
+define <8 x i32> @fold_v8i32() {
+; CHECK-SSE-LABEL: fold_v8i32:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [0,16777216,4294967295,33554432]
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm1 = [4261412863,67108864,4227858431,100663296]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v8i32:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} ymm0 = [0,16777216,4294967295,33554432,4261412863,67108864,4227858431,100663296]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <8 x i32> @llvm.bswap.v8i32(<8 x i32> <i32 0, i32 1, i32 -1, i32 2, i32 -3, i32 4, i32 -5, i32 6>)
+ ret <8 x i32> %r
+}
+
+define <4 x i64> @fold_v4i64() {
+; CHECK-SSE-LABEL: fold_v4i64:
+; CHECK-SSE: # BB#0: # %entry
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm0 = [18374686479671623680,18446744073709551615]
+; CHECK-SSE-NEXT: movaps {{.*#+}} xmm1 = [18446462598732840960,72056494526300160]
+; CHECK-SSE-NEXT: retq
+;
+; CHECK-AVX-LABEL: fold_v4i64:
+; CHECK-AVX: # BB#0: # %entry
+; CHECK-AVX-NEXT: vmovaps {{.*#+}} ymm0 = [18374686479671623680,18446744073709551615,18446462598732840960,72056494526300160]
+; CHECK-AVX-NEXT: retq
+entry:
+ %r = call <4 x i64> @llvm.bswap.v4i64(<4 x i64> <i64 255, i64 -1, i64 65535, i64 16776960>)
+ ret <4 x i64> %r
+}