[llvm] r308833 - [DAG] Fix typo preventing some stores merges to truncated stores.

[Nirav Dave via llvm-commits]

Author: niravd
Date: Sat Jul 22 19:06:28 2017
New Revision: 308833

URL: http://llvm.org/viewvc/llvm-project?rev=308833&view=rev
Log:
[DAG] Fix typo preventing some stores merges to truncated stores.

Check the actual memory type stored and not the extended value size
when considering if truncated store merge is worthwhile.

Reviewers: efriedma, RKSimon, spatel, jyknight

Reviewed By: efriedma

Subscribers: llvm-commits, nhaehnle

Differential Revision: https://reviews.llvm.org/D35623

Modified:
    llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
    llvm/trunk/test/CodeGen/AMDGPU/merge-stores.ll
    llvm/trunk/test/CodeGen/BPF/undef.ll

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp?rev=308833&r1=308832&r2=308833&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp Sat Jul 22 19:06:28 2017
@@ -12866,8 +12866,8 @@ bool DAGCombiner::MergeConsecutiveStores
               TLI.getTypeToTransformTo(Context, StoredVal.getValueType());
           if (TLI.isTruncStoreLegal(LegalizedStoredValueTy, StoreTy) &&
               TLI.canMergeStoresTo(FirstStoreAS, LegalizedStoredValueTy, DAG) &&
-              TLI.allowsMemoryAccess(Context, DL, LegalizedStoredValueTy,
-                                     FirstStoreAS, FirstStoreAlign, &IsFast) &&
+              TLI.allowsMemoryAccess(Context, DL, StoreTy, FirstStoreAS,
+                                     FirstStoreAlign, &IsFast) &&
               IsFast) {
             LastIntegerTrunc = true;
             LastLegalType = i + 1;
@@ -13098,8 +13098,8 @@ bool DAGCombiner::MergeConsecutiveStores
             TLI.isLoadExtLegal(ISD::SEXTLOAD, LegalizedStoredValueTy,
                                StoreTy) &&
             TLI.isLoadExtLegal(ISD::EXTLOAD, LegalizedStoredValueTy, StoreTy) &&
-            TLI.allowsMemoryAccess(Context, DL, LegalizedStoredValueTy,
-                                   FirstStoreAS, FirstStoreAlign, &IsFastSt) &&
+            TLI.allowsMemoryAccess(Context, DL, StoreTy, FirstStoreAS,
+                                   FirstStoreAlign, &IsFastSt) &&
             IsFastSt &&
             TLI.allowsMemoryAccess(Context, DL, StoreTy, FirstLoadAS,
                                    FirstLoadAlign, &IsFastLd) &&

Modified: llvm/trunk/test/CodeGen/AMDGPU/merge-stores.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/AMDGPU/merge-stores.ll?rev=308833&r1=308832&r2=308833&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/AMDGPU/merge-stores.ll (original)
+++ llvm/trunk/test/CodeGen/AMDGPU/merge-stores.ll Sat Jul 22 19:06:28 2017
@@ -10,8 +10,7 @@
 
 
 ; GCN-LABEL: {{^}}merge_global_store_2_constants_i8:
-; GCN: buffer_store_byte
-; GCN: buffer_store_byte
+; GCN: buffer_store_short
 ; GCN: s_endpgm
 define amdgpu_kernel void @merge_global_store_2_constants_i8(i8 addrspace(1)* %out) #0 {
   %out.gep.1 = getelementptr i8, i8 addrspace(1)* %out, i32 1
@@ -489,8 +488,7 @@ define amdgpu_kernel void @merge_global_
 }
 
 ; GCN-LABEL: {{^}}merge_local_store_2_constants_i8:
-; GCN: ds_write_b8
-; GCN: ds_write_b8
+; GCN: ds_write_b16
 ; GCN: s_endpgm
 define amdgpu_kernel void @merge_local_store_2_constants_i8(i8 addrspace(3)* %out) #0 {
   %out.gep.1 = getelementptr i8, i8 addrspace(3)* %out, i32 1

Modified: llvm/trunk/test/CodeGen/BPF/undef.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/BPF/undef.ll?rev=308833&r1=308832&r2=308833&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/BPF/undef.ll (original)
+++ llvm/trunk/test/CodeGen/BPF/undef.ll Sat Jul 22 19:06:28 2017
@@ -23,10 +23,9 @@ define i32 @ebpf_filter(%struct.__sk_buf
 ; EL: r1 = 134678021
 ; EB: r1 = 84281096
 ; CHECK: *(u32 *)(r10 - 8) = r1
-; CHECK: r1 = 9
-; CHECK: *(u8 *)(r10 - 4) = r1
-; CHECK: r1 = 10
-; CHECK: *(u8 *)(r10 - 3) = r1
+; EL: r1 = 2569
+; EB: r1 = 2314
+; CHECK: *(u16 *)(r10 - 4) = r1
 ; CHECK: *(u16 *)(r10 + 24) = r2
 ; CHECK: *(u16 *)(r10 + 22) = r2
 ; CHECK: *(u16 *)(r10 + 20) = r2