[PATCH] D80062: [x86] Propagate memory operands during call frame optimization & ISel DAG postprocessing

[Jean-Michel Gorius via Phabricator via llvm-commits]

Kayjukh created this revision.
Kayjukh added a reviewer: craig.topper.
Herald added subscribers: llvm-commits, hiraditya.
Herald added a project: LLVM.

This propagates memory operands when folding loads instructions into instructions that directly operate on memory.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D80062

Files:
  llvm/lib/Target/X86/X86CallFrameOptimization.cpp
  llvm/lib/Target/X86/X86ISelDAGToDAG.cpp


Index: llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
===================================================================
--- llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
+++ llvm/lib/Target/X86/X86ISelDAGToDAG.cpp
@@ -1393,6 +1393,8 @@
                           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;
Index: llvm/lib/Target/X86/X86CallFrameOptimization.cpp
===================================================================
--- llvm/lib/Target/X86/X86CallFrameOptimization.cpp
+++ llvm/lib/Target/X86/X86CallFrameOptimization.cpp
@@ -562,6 +562,7 @@
         unsigned NumOps = DefMov->getDesc().getNumOperands();
         for (unsigned i = NumOps - X86::AddrNumOperands; i != NumOps; ++i)
           Push->addOperand(DefMov->getOperand(i));
+        Push->setMemRefs(MF, DefMov->memoperands());
 
         DefMov->eraseFromParent();
       } else {


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D80062.264425.patch
Type: text/x-patch
Size: 1171 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20200516/fba20bef/attachment.bin>