From 7cd3fe7db6c57664b47dbbaf02825caa384116ef Mon Sep 17 00:00:00 2001
From: Amaury Sechet <deadalnix@gmail.com>
Date: Wed, 2 Mar 2016 21:28:30 +0000
Subject: [PATCH] Unpack array of all sizes in InstCombine

Summary: This is another step toward improving fca support. This unpack load of array in a series of load to array's elements.

Reviewers: chandlerc, joker.eph, majnemer, reames, hfinkel

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D15890

llvm-svn: 262521
---
 .../InstCombine/InstCombineLoadStoreAlloca.cpp     | 43 +++++++++++++++++++---
 llvm/test/Transforms/InstCombine/unpack-fca.ll     | 21 +++++++++++
 2 files changed, 59 insertions(+), 5 deletions(-)

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp b/llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp
index 3a137a9..c559d93 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineLoadStoreAlloca.cpp
@@ -574,13 +574,46 @@ static Instruction *unpackLoadToAggregate(InstCombiner &IC, LoadInst &LI) {
   }
 
   if (auto *AT = dyn_cast<ArrayType>(T)) {
-    // If the array only have one element, we unpack.
-    if (AT->getNumElements() == 1) {
-      LoadInst *NewLoad = combineLoadToNewType(IC, LI, AT->getElementType(),
-                                               ".unpack");
+    auto *ET = AT->getElementType();
+    auto NumElements = AT->getNumElements();
+    if (NumElements == 1) {
+      LoadInst *NewLoad = combineLoadToNewType(IC, LI, ET, ".unpack");
       return IC.replaceInstUsesWith(LI, IC.Builder->CreateInsertValue(
-        UndefValue::get(T), NewLoad, 0, LI.getName()));
+        UndefValue::get(T), NewLoad, 0, Name));
     }
+
+    const DataLayout &DL = IC.getDataLayout();
+    auto EltSize = DL.getTypeAllocSize(ET);
+    auto Align = LI.getAlignment();
+    if (!Align)
+      Align = DL.getABITypeAlignment(T);
+
+    SmallString<16> LoadName = Name;
+    LoadName += ".unpack";
+    SmallString<16> EltName = Name;
+    EltName += ".elt";
+
+    auto *Addr = LI.getPointerOperand();
+    auto *IdxType = Type::getInt64Ty(T->getContext());
+    auto *Zero = ConstantInt::get(IdxType, 0);
+
+    Value *V = UndefValue::get(T);
+    uint64_t Offset = 0;
+    for (uint64_t i = 0; i < NumElements; i++) {
+      Value *Indices[2] = {
+        Zero,
+        ConstantInt::get(IdxType, i),
+      };
+      auto *Ptr = IC.Builder->CreateInBoundsGEP(AT, Addr, makeArrayRef(Indices),
+                                                EltName);
+      auto *L = IC.Builder->CreateAlignedLoad(Ptr, MinAlign(Align, Offset),
+                                              LoadName);
+      V = IC.Builder->CreateInsertValue(V, L, i);
+      Offset += EltSize;
+    }
+
+    V->setName(Name);
+    return IC.replaceInstUsesWith(LI, V);
   }
 
   return nullptr;
diff --git a/llvm/test/Transforms/InstCombine/unpack-fca.ll b/llvm/test/Transforms/InstCombine/unpack-fca.ll
index bed3b61..cd97ce6 100644
--- a/llvm/test/Transforms/InstCombine/unpack-fca.ll
+++ b/llvm/test/Transforms/InstCombine/unpack-fca.ll
@@ -137,6 +137,27 @@ define %B @structB(%B* %b.ptr) {
   ret %B %1
 }
 
+define [2 x %B] @loadArrayOfB([2 x %B]* %ab.ptr) {
+; CHECK-LABEL: loadArrayOfB
+; CHECK-NEXT: [[GEP1:%[a-z0-9\.]+]] = getelementptr inbounds [2 x %B], [2 x %B]* %ab.ptr, i64 0, i64 0, i32 0
+; CHECK-NEXT: [[LOAD1:%[a-z0-9\.]+]] = load i8*, i8** [[GEP1]], align 8
+; CHECK-NEXT: [[IV1:%[a-z0-9\.]+]] = insertvalue %B undef, i8* [[LOAD1]], 0
+; CHECK-NEXT: [[GEP2:%[a-z0-9\.]+]] = getelementptr inbounds [2 x %B], [2 x %B]* %ab.ptr, i64 0, i64 0, i32 1
+; CHECK-NEXT: [[LOAD2:%[a-z0-9\.]+]] = load i64, i64* [[GEP2]], align 8
+; CHECK-NEXT: [[IV2:%[a-z0-9\.]+]] = insertvalue %B [[IV1]], i64 [[LOAD2]], 1
+; CHECK-NEXT: [[IV3:%[a-z0-9\.]+]] = insertvalue [2 x %B] undef, %B [[IV2]], 0
+; CHECK-NEXT: [[GEP3:%[a-z0-9\.]+]] = getelementptr inbounds [2 x %B], [2 x %B]* %ab.ptr, i64 0, i64 1, i32 0
+; CHECK-NEXT: [[LOAD3:%[a-z0-9\.]+]] = load i8*, i8** [[GEP3]], align 8
+; CHECK-NEXT: [[IV4:%[a-z0-9\.]+]] = insertvalue %B undef, i8* [[LOAD3]], 0
+; CHECK-NEXT: [[GEP4:%[a-z0-9\.]+]] = getelementptr inbounds [2 x %B], [2 x %B]* %ab.ptr, i64 0, i64 1, i32 1
+; CHECK-NEXT: [[LOAD4:%[a-z0-9\.]+]] = load i64, i64* [[GEP4]], align 8
+; CHECK-NEXT: [[IV5:%[a-z0-9\.]+]] = insertvalue %B [[IV4]], i64 [[LOAD4]], 1
+; CHECK-NEXT: [[IV6:%[a-z0-9\.]+]] = insertvalue [2 x %B] [[IV3]], %B [[IV5]], 1
+; CHECK-NEXT: ret [2 x %B] [[IV6]]
+  %1 = load [2 x %B], [2 x %B]* %ab.ptr, align 8
+  ret [2 x %B] %1
+}
+
 %struct.S = type <{ i8, %struct.T }>
 %struct.T = type { i32, i32 }
 
-- 
2.7.4