Revert "Revert "Fix merges of non-zero vector stores""

Reapply r239539. Don't assume the collected number of
stores is the same vector size. Just take the first N
stores to fill the vector.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239825 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index 33f5f52332ae886aa2f3ca4704e7a2bda59069b9..ab72941310d4353e718657bd6056133b9e2ee0a7 100644 (file)
--- a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -388,6 +388,13 @@ namespace {
       unsigned SequenceNum;
     };
 
+    /// This is a helper function for MergeStoresOfConstantsOrVecElts. Returns a
+    /// constant build_vector of the stored constant values in Stores.
+    SDValue getMergedConstantVectorStore(SelectionDAG &DAG,
+                                         SDLoc SL,
+                                         ArrayRef<MemOpLink> Stores,
+                                         EVT Ty) const;
+
     /// This is a helper function for MergeConsecutiveStores. When the source
     /// elements of the consecutive stores are all constants or all extracted
     /// vector elements, try to merge them into one larger store.
@@ -10591,6 +10598,18 @@ struct BaseIndexOffset {
 };
 } // namespace
 
+SDValue DAGCombiner::getMergedConstantVectorStore(SelectionDAG &DAG,
+                                                  SDLoc SL,
+                                                  ArrayRef<MemOpLink> Stores,
+                                                  EVT Ty) const {
+  SmallVector<SDValue, 8> BuildVector;
+
+  for (unsigned I = 0, E = Ty.getVectorNumElements(); I != E; ++I)
+    BuildVector.push_back(cast<StoreSDNode>(Stores[I].MemNode)->getValue());
+
+  return DAG.getNode(ISD::BUILD_VECTOR, SL, Ty, BuildVector);
+}
+
 bool DAGCombiner::MergeStoresOfConstantsOrVecElts(
                   SmallVectorImpl<MemOpLink> &StoreNodes, EVT MemVT,
                   unsigned NumElem, bool IsConstantSrc, bool UseVector) {
@@ -10621,12 +10640,7 @@ bool DAGCombiner::MergeStoresOfConstantsOrVecElts(
     EVT Ty = EVT::getVectorVT(*DAG.getContext(), MemVT, NumElem);
     assert(TLI.isTypeLegal(Ty) && "Illegal vector store");
     if (IsConstantSrc) {
-      // A vector store with a constant source implies that the constant is
-      // zero; we only handle merging stores of constant zeros because the zero
-      // can be materialized without a load.
-      // It may be beneficial to loosen this restriction to allow non-zero
-      // store merging.
-      StoredVal = DAG.getConstant(0, DL, Ty);
+      StoredVal = getMergedConstantVectorStore(DAG, DL, StoreNodes, Ty);
     } else {
       SmallVector<SDValue, 8> Ops;
       for (unsigned i = 0; i < NumElem ; ++i) {

diff --git a/test/CodeGen/AMDGPU/merge-stores.ll b/test/CodeGen/AMDGPU/merge-stores.ll
index dbf9d4481ffb1bb8943d50f6fdf21955d2e8e753..34a2fc7ffa7451a49faf8667a44dc2bc717833e3 100644 (file)
--- a/test/CodeGen/AMDGPU/merge-stores.ll
+++ b/test/CodeGen/AMDGPU/merge-stores.ll
@@ -89,7 +89,11 @@ define void @merge_global_store_2_constants_i32_f32(i32 addrspace(1)* %out) #0 {
 }
 
 ; GCN-LABEL: {{^}}merge_global_store_2_constants_f32_i32:
-; GCN: buffer_store_dwordx2
+; SI-DAG: s_mov_b32 [[SLO:s[0-9]+]], 4.0
+; SI-DAG: s_movk_i32 [[SHI:s[0-9]+]], 0x7b{{$}}
+; SI-DAG: v_mov_b32_e32 v[[VLO:[0-9]+]], [[SLO]]
+; SI-DAG: v_mov_b32_e32 v[[VHI:[0-9]+]], [[SHI]]
+; GCN: buffer_store_dwordx2 v{{\[}}[[VLO]]:[[VHI]]{{\]}}
 define void @merge_global_store_2_constants_f32_i32(float addrspace(1)* %out) #0 {
   %out.gep.1 = getelementptr float, float addrspace(1)* %out, i32 1
   %out.gep.1.bc = bitcast float addrspace(1)* %out.gep.1 to i32 addrspace(1)*
@@ -99,7 +103,11 @@ define void @merge_global_store_2_constants_f32_i32(float addrspace(1)* %out) #0
 }
 
 ; GCN-LABEL: {{^}}merge_global_store_4_constants_i32:
-; GCN: buffer_store_dwordx4
+; GCN-DAG: v_mov_b32_e32 v[[HI:[0-9]+]], 0x14d{{$}}
+; GCN-DAG: v_mov_b32_e32 v{{[0-9]+}}, 0x1c8{{$}}
+; GCN-DAG: v_mov_b32_e32 v{{[0-9]+}}, 0x7b{{$}}
+; GCN-DAG: v_mov_b32_e32 v[[LO:[0-9]+]], 0x4d2{{$}}
+; GCN: buffer_store_dwordx4 v{{\[}}[[LO]]:[[HI]]{{\]}}
 define void @merge_global_store_4_constants_i32(i32 addrspace(1)* %out) #0 {
   %out.gep.1 = getelementptr i32, i32 addrspace(1)* %out, i32 1
   %out.gep.2 = getelementptr i32, i32 addrspace(1)* %out, i32 2
@@ -530,6 +538,95 @@ define void @merge_local_store_4_constants_i32(i32 addrspace(3)* %out) #0 {
   ret void
 }
 
+; GCN-LABEL: {{^}}merge_global_store_5_constants_i32:
+; GCN-DAG: v_mov_b32_e32 v[[LO:[0-9]+]], 9{{$}}
+; GCN-DAG: v_mov_b32_e32 v[[HI4:[0-9]+]], -12{{$}}
+; GCN: buffer_store_dwordx4 v{{\[}}[[LO]]:[[HI4]]{{\]}}
+; GCN: v_mov_b32_e32 v[[HI:[0-9]+]], 11{{$}}
+; GCN: buffer_store_dword v[[HI]]
+define void @merge_global_store_5_constants_i32(i32 addrspace(1)* %out) {
+  store i32 9, i32 addrspace(1)* %out, align 4
+  %idx1 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 1
+  store i32 12, i32 addrspace(1)* %idx1, align 4
+  %idx2 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 2
+  store i32 16, i32 addrspace(1)* %idx2, align 4
+  %idx3 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 3
+  store i32 -12, i32 addrspace(1)* %idx3, align 4
+  %idx4 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 4
+  store i32 11, i32 addrspace(1)* %idx4, align 4
+  ret void
+}
+
+; GCN-LABEL: {{^}}merge_global_store_6_constants_i32:
+; GCN: buffer_store_dwordx4
+; GCN: buffer_store_dwordx2
+define void @merge_global_store_6_constants_i32(i32 addrspace(1)* %out) {
+  store i32 13, i32 addrspace(1)* %out, align 4
+  %idx1 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 1
+  store i32 15, i32 addrspace(1)* %idx1, align 4
+  %idx2 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 2
+  store i32 62, i32 addrspace(1)* %idx2, align 4
+  %idx3 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 3
+  store i32 63, i32 addrspace(1)* %idx3, align 4
+  %idx4 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 4
+  store i32 11, i32 addrspace(1)* %idx4, align 4
+  %idx5 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 5
+  store i32 123, i32 addrspace(1)* %idx5, align 4
+  ret void
+}
+
+; GCN-LABEL: {{^}}merge_global_store_7_constants_i32:
+; GCN: buffer_store_dwordx4
+; GCN: buffer_store_dwordx2
+; GCN: buffer_store_dword v
+define void @merge_global_store_7_constants_i32(i32 addrspace(1)* %out) {
+  store i32 34, i32 addrspace(1)* %out, align 4
+  %idx1 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 1
+  store i32 999, i32 addrspace(1)* %idx1, align 4
+  %idx2 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 2
+  store i32 65, i32 addrspace(1)* %idx2, align 4
+  %idx3 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 3
+  store i32 33, i32 addrspace(1)* %idx3, align 4
+  %idx4 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 4
+  store i32 98, i32 addrspace(1)* %idx4, align 4
+  %idx5 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 5
+  store i32 91, i32 addrspace(1)* %idx5, align 4
+  %idx6 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 6
+  store i32 212, i32 addrspace(1)* %idx6, align 4
+  ret void
+}
+
+; GCN-LABEL: {{^}}merge_global_store_8_constants_i32:
+; XGCN: buffer_store_dwordx4
+; XGCN: buffer_store_dwordx4
+
+; GCN: buffer_store_dword v
+; GCN: buffer_store_dword v
+; GCN: buffer_store_dword v
+; GCN: buffer_store_dword v
+; GCN: buffer_store_dword v
+; GCN: buffer_store_dword v
+; GCN: buffer_store_dword v
+; GCN: buffer_store_dword v
+define void @merge_global_store_8_constants_i32(i32 addrspace(1)* %out) {
+  store i32 34, i32 addrspace(1)* %out, align 4
+  %idx1 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 1
+  store i32 999, i32 addrspace(1)* %idx1, align 4
+  %idx2 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 2
+  store i32 65, i32 addrspace(1)* %idx2, align 4
+  %idx3 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 3
+  store i32 33, i32 addrspace(1)* %idx3, align 4
+  %idx4 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 4
+  store i32 98, i32 addrspace(1)* %idx4, align 4
+  %idx5 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 5
+  store i32 91, i32 addrspace(1)* %idx5, align 4
+  %idx6 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 6
+  store i32 212, i32 addrspace(1)* %idx6, align 4
+  %idx7 = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 7
+  store i32 999, i32 addrspace(1)* %idx7, align 4
+  ret void
+}
+
 declare void @llvm.AMDGPU.barrier.local() #1
 
 attributes #0 = { nounwind }

diff --git a/test/CodeGen/X86/2012-11-28-merge-store-alias.ll b/test/CodeGen/X86/2012-11-28-merge-store-alias.ll
index ed1daadf6297c8e2408c78fb2fc3604fcc36c4a7..c16deeff3d99a01d47d1dbb0606ee63119a8e029 100644 (file)
--- a/test/CodeGen/X86/2012-11-28-merge-store-alias.ll
+++ b/test/CodeGen/X86/2012-11-28-merge-store-alias.ll
@@ -3,6 +3,7 @@
 ; CHECK: merge_stores_can
 ; CHECK: callq foo
 ; CHECK: xorps %xmm0, %xmm0
+; CHECK-NEXT: movl 36(%rsp), %ebp
 ; CHECK-NEXT: movups  %xmm0
 ; CHECK: callq foo
 ; CHECK: ret