From c5490e5a295bff69e24845d540d7fe8622a72a38 Mon Sep 17 00:00:00 2001
From: Ayman Musa <ayman.musa@intel.com>
Date: Mon, 15 May 2017 11:30:54 +0000
Subject: [X86] Relocate code of replacement of subtarget unsupported masked
 memory intrinsics to run also on -O0 option.

Currently, when masked load, store, gather or scatter intrinsics are used, we check in CodeGenPrepare pass if the subtarget support these intrinsics, if not we replace them with scalar code - this is a functional transformation not an optimization (not optional).

CodeGenPrepare pass does not run when the optimization level is set to CodeGenOpt::None (-O0).

Functional transformation should run with all optimization levels, so here I created a new pass which runs on all optimization levels and does no more than this transformation.

Differential Revision: https://reviews.llvm.org/D32487

llvm-svn: 303050
---
 llvm/lib/CodeGen/CodeGen.cpp | 1 +
 1 file changed, 1 insertion(+)

(limited to 'llvm/lib/CodeGen/CodeGen.cpp')

diff --git a/llvm/lib/CodeGen/CodeGen.cpp b/llvm/lib/CodeGen/CodeGen.cpp
index 26a98d76c13..4d30c6574b1 100644
--- a/llvm/lib/CodeGen/CodeGen.cpp
+++ b/llvm/lib/CodeGen/CodeGen.cpp
@@ -81,6 +81,7 @@ void llvm::initializeCodeGen(PassRegistry &Registry) {
   initializeRegisterCoalescerPass(Registry);
   initializeRenameIndependentSubregsPass(Registry);
   initializeSafeStackLegacyPassPass(Registry);
+  initializeScalarizeMaskedMemIntrinPass(Registry);
   initializeShrinkWrapPass(Registry);
   initializeSlotIndexesPass(Registry);
   initializeStackColoringPass(Registry);
-- 
cgit v1.2.3