summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorNirav Dave <niravd@google.com>2017-04-06 19:05:41 +0000
committerNirav Dave <niravd@google.com>2017-04-06 19:05:41 +0000
commit974f7c23ae084f830f94f5749f4f24b48e98f77d (patch)
tree784a5768de5e70570599a3db0b10c47deb94b6bd
parent8ef20ea7c24c96670e8ef3d3b77772d17ada581b (diff)
downloadbcm5719-llvm-974f7c23ae084f830f94f5749f4f24b48e98f77d.tar.gz
bcm5719-llvm-974f7c23ae084f830f94f5749f4f24b48e98f77d.zip
[SDAG] Fix visitAND optimization to deal with vector extract case again.
Summary: Fix case elided by rL298920. Fixes PR32545. Reviewers: eli.friedman, RKSimon Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D31759 llvm-svn: 299688
-rw-r--r--llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp2
-rw-r--r--llvm/test/CodeGen/ARM/pr32545.ll22
2 files changed, 23 insertions, 1 deletions
diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index 8a7a8904539..8f05c61a957 100644
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -3589,7 +3589,7 @@ SDValue DAGCombiner::visitAND(SDNode *N) {
SDValue NewLoad(Load, 0);
// Fold the AND away. NewLoad may get replaced immediately.
- CombineTo(N, NewLoad);
+ CombineTo(N, (N0.getNode() == Load) ? NewLoad : N0);
if (Load->getExtensionType() == ISD::EXTLOAD) {
NewLoad = DAG.getLoad(Load->getAddressingMode(), ISD::ZEXTLOAD,
diff --git a/llvm/test/CodeGen/ARM/pr32545.ll b/llvm/test/CodeGen/ARM/pr32545.ll
new file mode 100644
index 00000000000..5bfb01b4598
--- /dev/null
+++ b/llvm/test/CodeGen/ARM/pr32545.ll
@@ -0,0 +1,22 @@
+; RUN: llc %s -o - | FileCheck %s
+
+target datalayout = "e-m:e-p:32:32-i64:64-v128:64:128-a:0:32-n32-S64"
+target triple = "armv7--linux-gnueabi"
+
+; CHECK: vld1.16 {[[DREG:d[0-9]+]][0]}, {{.*}}
+; CHECK: vmovl.u8 [[QREG:q[0-9]+]], [[DREG]]
+; CHECK: vmovl.u16 [[QREG]], [[DREG]]
+
+define void @f(i32 %dstStride, i8* %indvars.iv, <2 x i8>* %zz) {
+entry:
+ br label %for.body
+
+for.body:
+ %tmp = load <2 x i8>, <2 x i8>* %zz, align 1
+ %tmp1 = extractelement <2 x i8> %tmp, i32 0
+ %.lhs.rhs = zext i8 %tmp1 to i32
+ call void @g(i32 %.lhs.rhs)
+ br label %for.body
+}
+
+declare void @g(i32)
OpenPOWER on IntegriCloud