[llvm] 0616e99 - [InstSimplify][SVE] Fix SimplifyGEPInst for scalable vector.

[Huihui Zhang via llvm-commits]

Author: Huihui Zhang
Date: 2020-03-16T11:46:12-07:00
New Revision: 0616e9964b9ef2e2cb12790515610e7cdd0042e3

URL: https://github.com/llvm/llvm-project/commit/0616e9964b9ef2e2cb12790515610e7cdd0042e3
DIFF: https://github.com/llvm/llvm-project/commit/0616e9964b9ef2e2cb12790515610e7cdd0042e3.diff

LOG: [InstSimplify][SVE] Fix SimplifyGEPInst for scalable vector.

Summary:
Skip folds that rely on DataLayout::getTypeAllocSize(). For scalable
vector, only minimal type alloc size is known at compile-time.

Reviewers: sdesmalen, efriedma, spatel, apazos

Reviewed By: efriedma

Subscribers: tschuett, hiraditya, rkruppe, psnobl, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D75892

Added: 
    

Modified: 
    llvm/lib/Analysis/InstructionSimplify.cpp
    llvm/test/Transforms/InstSimplify/vscale.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Analysis/InstructionSimplify.cpp b/llvm/lib/Analysis/InstructionSimplify.cpp
index bc9ae5ebffb0..6ea140f31ef4 100644
--- a/llvm/lib/Analysis/InstructionSimplify.cpp
+++ b/llvm/lib/Analysis/InstructionSimplify.cpp
@@ -4081,13 +4081,16 @@ static Value *SimplifyGEPInst(Type *SrcTy, ArrayRef<Value *> Ops,
   if (isa<UndefValue>(Ops[0]))
     return UndefValue::get(GEPTy);
 
+  bool IsScalableVec =
+      SrcTy->isVectorTy() ? SrcTy->getVectorIsScalable() : false;
+
   if (Ops.size() == 2) {
     // getelementptr P, 0 -> P.
     if (match(Ops[1], m_Zero()) && Ops[0]->getType() == GEPTy)
       return Ops[0];
 
     Type *Ty = SrcTy;
-    if (Ty->isSized()) {
+    if (!IsScalableVec && Ty->isSized()) {
       Value *P;
       uint64_t C;
       uint64_t TyAllocSize = Q.DL.getTypeAllocSize(Ty);
@@ -4135,7 +4138,7 @@ static Value *SimplifyGEPInst(Type *SrcTy, ArrayRef<Value *> Ops,
     }
   }
 
-  if (Q.DL.getTypeAllocSize(LastType) == 1 &&
+  if (!IsScalableVec && Q.DL.getTypeAllocSize(LastType) == 1 &&
       all_of(Ops.slice(1).drop_back(1),
              [](Value *Idx) { return match(Idx, m_Zero()); })) {
     unsigned IdxWidth =

diff  --git a/llvm/test/Transforms/InstSimplify/vscale.ll b/llvm/test/Transforms/InstSimplify/vscale.ll
index 5a6971c9e292..d112128042c2 100644
--- a/llvm/test/Transforms/InstSimplify/vscale.ll
+++ b/llvm/test/Transforms/InstSimplify/vscale.ll
@@ -94,3 +94,42 @@ define i32 @insert_extract_element_same_vec_idx_2(<vscale x 4 x i32> %a) {
   ret i32 %r
 }
 
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+;; Memory Access and Addressing Operations
+;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
+
+; getelementptr
+
+define <vscale x 4 x i32*> @getelementptr_constant_foldable_1() {
+; CHECK-LABEL: @getelementptr_constant_foldable_1(
+; CHECK-NEXT:    ret <vscale x 4 x i32*> zeroinitializer
+;
+  %ptr = getelementptr i32, <vscale x 4 x i32*> zeroinitializer, <vscale x 4 x i64> undef
+  ret <vscale x 4 x i32*> %ptr
+}
+
+define <vscale x 4 x <vscale x 4 x i32>*> @getelementptr_constant_foldable_2() {
+; CHECK-LABEL: @getelementptr_constant_foldable_2(
+; CHECK-NEXT:    ret <vscale x 4 x <vscale x 4 x i32>*> zeroinitializer
+;
+  %ptr = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* null, <vscale x 4 x i64> undef
+  ret <vscale x 4 x <vscale x 4 x i32>*> %ptr
+}
+
+; fold getelementptr P, 0 -> P.
+define <vscale x 4 x i32>* @getelementptr_constant_foldable_3() {
+; CHECK-LABEL: @getelementptr_constant_foldable_3(
+; CHECK-NEXT:    ret <vscale x 4 x i32>* null
+;
+  %ptr = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* null, i64 0
+  ret <vscale x 4 x i32>* %ptr
+}
+
+define <vscale x 4 x i32>* @getelementptr_not_constant_foldable(i64 %x) {
+; CHECK-LABEL: @getelementptr_not_constant_foldable(
+; CHECK-NEXT:    [[PTR:%.*]] = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* null, i64 [[X:%.*]]
+; CHECK-NEXT:    ret <vscale x 4 x i32>* [[PTR]]
+;
+  %ptr = getelementptr <vscale x 4 x i32>, <vscale x 4 x i32>* null, i64 %x
+  ret <vscale x 4 x i32>* %ptr
+}