From 9812f7b852cb1d0466d5b4ac7ef66247cf51e0c6 Mon Sep 17 00:00:00 2001
From: Amaury Sechet <deadalnix@gmail.com>
Date: Wed, 6 Jan 2016 09:30:39 +0000
Subject: [PATCH] Improve load/store to memcpy for aggregate

Summary: It turns out that if we don't try to do it at the store location, we can do it before any operation that alias the load, as long as no operation alias the store.

Reviewers: craig.topper, spatel, dexonsmith, Prazek, chandlerc, joker.eph

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D15903

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256923 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Transforms/Scalar/MemCpyOptimizer.cpp | 32 ++++++++++++++++-------
 test/Transforms/MemCpyOpt/fca2memcpy.ll   | 25 ++++++++++++++++++
 2 files changed, 48 insertions(+), 9 deletions(-)

diff --git a/lib/Transforms/Scalar/MemCpyOptimizer.cpp b/lib/Transforms/Scalar/MemCpyOptimizer.cpp
index 94725db56b8..7354016c212 100644
--- a/lib/Transforms/Scalar/MemCpyOptimizer.cpp
+++ b/lib/Transforms/Scalar/MemCpyOptimizer.cpp
@@ -519,19 +519,33 @@ bool MemCpyOpt::processStore(StoreInst *SI, BasicBlock::iterator &BBI) {
 
         // We use alias analysis to check if an instruction may store to
         // the memory we load from in between the load and the store. If
-        // such an instruction is found, we store it in AI.
-        Instruction *AI = nullptr;
+        // such an instruction is found, we try to promote there instead
+        // of at the store position.
+        Instruction *P = SI;
         for (BasicBlock::iterator I = ++LI->getIterator(), E = SI->getIterator();
              I != E; ++I) {
-          if (AA.getModRefInfo(&*I, LoadLoc) & MRI_Mod) {
-            AI = &*I;
-            break;
+          if (!(AA.getModRefInfo(&*I, LoadLoc) & MRI_Mod))
+            continue;
+
+          // We found an instruction that may write to the loaded memory.
+          // We can try to promote at this position instead of the store
+          // position if nothing alias the store memory after this.
+          P = &*I;
+          for (; I != E; ++I) {
+            MemoryLocation StoreLoc = MemoryLocation::get(SI);
+            if (AA.getModRefInfo(&*I, StoreLoc) != MRI_NoModRef) {
+              DEBUG(dbgs() << "Alias " << *I << "\n");
+              P = nullptr;
+              break;
+            }
           }
+
+          break;
         }
 
-        // If no aliasing instruction is found, then we can promote the
-        // load/store pair to a memcpy at the store loaction.
-        if (!AI) {
+        // If a valid insertion position is found, then we can promote
+        // the load/store pair to a memcpy.
+        if (P) {
           // If we load from memory that may alias the memory we store to,
           // memmove must be used to preserve semantic. If not, memcpy can
           // be used.
@@ -542,7 +556,7 @@ bool MemCpyOpt::processStore(StoreInst *SI, BasicBlock::iterator &BBI) {
           unsigned Align = findCommonAlignment(DL, SI, LI);
           uint64_t Size = DL.getTypeStoreSize(T);
 
-          IRBuilder<> Builder(SI);
+          IRBuilder<> Builder(P);
           Instruction *M;
           if (UseMemMove)
             M = Builder.CreateMemMove(SI->getPointerOperand(),
diff --git a/test/Transforms/MemCpyOpt/fca2memcpy.ll b/test/Transforms/MemCpyOpt/fca2memcpy.ll
index 61ec7f1d914..75a1a8f96e2 100644
--- a/test/Transforms/MemCpyOpt/fca2memcpy.ll
+++ b/test/Transforms/MemCpyOpt/fca2memcpy.ll
@@ -7,6 +7,7 @@ target triple = "x86_64-unknown-linux-gnu"
 
 define void @copy(%S* %src, %S* %dst) {
 ; CHECK-LABEL: copy
+; CHECK-NOT: load
 ; CHECK: call void @llvm.memmove.p0i8.p0i8.i64
 ; CHECK-NEXT: ret void
   %1 = load %S, %S* %src
@@ -16,6 +17,7 @@ define void @copy(%S* %src, %S* %dst) {
 
 define void @noaliassrc(%S* noalias %src, %S* %dst) {
 ; CHECK-LABEL: noaliassrc
+; CHECK-NOT: load
 ; CHECK: call void @llvm.memcpy.p0i8.p0i8.i64
 ; CHECK-NEXT: ret void
   %1 = load %S, %S* %src
@@ -25,6 +27,7 @@ define void @noaliassrc(%S* noalias %src, %S* %dst) {
 
 define void @noaliasdst(%S* %src, %S* noalias %dst) {
 ; CHECK-LABEL: noaliasdst
+; CHECK-NOT: load
 ; CHECK: call void @llvm.memcpy.p0i8.p0i8.i64
 ; CHECK-NEXT: ret void
   %1 = load %S, %S* %src
@@ -32,6 +35,28 @@ define void @noaliasdst(%S* %src, %S* noalias %dst) {
   ret void
 }
 
+define void @destroysrc(%S* %src, %S* %dst) {
+; CHECK-LABEL: destroysrc
+; CHECK-NOT: call
+; CHECK: ret void
+  %1 = load %S, %S* %src
+  store %S zeroinitializer, %S* %src
+  store %S %1, %S* %dst
+  ret void
+}
+
+define void @destroynoaliassrc(%S* noalias %src, %S* %dst) {
+; CHECK-LABEL: destroynoaliassrc
+; CHECK-NOT: load
+; CHECK: call void @llvm.memcpy.p0i8.p0i8.i64
+; CHECK-NEXT: store %S zeroinitializer, %S* %src
+; CHECK-NEXT: ret void
+  %1 = load %S, %S* %src
+  store %S zeroinitializer, %S* %src
+  store %S %1, %S* %dst
+  ret void
+}
+
 define void @copyalias(%S* %src, %S* %dst) {
 ; CHECK-LABEL: copyalias
 ; CHECK-NEXT: [[LOAD:%[a-z0-9\.]+]] = load %S, %S* %src