[llvm] cd12e79 - [x86] Propagate memory operands during ISel DAG postprocessing
Jean-Michel Gorius via llvm-commits
llvm-commits at lists.llvm.org
Mon May 18 12:35:38 PDT 2020
Author: Jean-Michel Gorius
Date: 2020-05-18T21:35:31+02:00
New Revision: cd12e79e6ddd235739716fff5c6a748915f664b9
URL: https://github.com/llvm/llvm-project/commit/cd12e79e6ddd235739716fff5c6a748915f664b9
DIFF: https://github.com/llvm/llvm-project/commit/cd12e79e6ddd235739716fff5c6a748915f664b9.diff
LOG: [x86] Propagate memory operands during ISel DAG postprocessing
Summary:
Propagate memory operands when folding test instructions.
This was split from D80062.
Reviewers: craig.topper, rnk, lebedev.ri
Reviewed By: craig.topper
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D80140
Added:
llvm/test/CodeGen/X86/isel-postprocessing-test-fold-memop.ll
Modified:
llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
Removed:
################################################################################
diff --git a/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp b/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
index ab816ecc559d..ffe9d432ccdf 100644
--- a/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
+++ b/llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
@@ -1388,6 +1388,8 @@ void X86DAGToDAGISel::PostprocessISelDAG() {
And.getOperand(6) /* Chain */ };
MachineSDNode *Test = CurDAG->getMachineNode(NewOpc, SDLoc(N),
MVT::i32, MVT::Other, Ops);
+ CurDAG->setNodeMemRefs(
+ Test, cast<MachineSDNode>(And.getNode())->memoperands());
ReplaceUses(N, Test);
MadeChange = true;
continue;
diff --git a/llvm/test/CodeGen/X86/isel-postprocessing-test-fold-memop.ll b/llvm/test/CodeGen/X86/isel-postprocessing-test-fold-memop.ll
new file mode 100644
index 000000000000..99a80fae441b
--- /dev/null
+++ b/llvm/test/CodeGen/X86/isel-postprocessing-test-fold-memop.ll
@@ -0,0 +1,14 @@
+; RUN: llc -O2 -mtriple=x86_64-- -stop-after=finalize-isel < %s | FileCheck %s
+
+define i1 @fold_test(i64* %x, i64 %l) {
+entry:
+ %0 = load i64, i64* %x, align 8
+ %and = and i64 %0, %l
+ %tobool = icmp ne i64 %and, 0
+ ret i1 %tobool
+
+ ; Folding the load+and+icmp instructions into a TEST64mr instruction
+ ; should preserve memory operands.
+ ; CHECK: TEST64mr {{.*}} :: (load 8 from {{%.*}})
+}
+
More information about the llvm-commits
mailing list