DAGCombiner: Don't stop finding better chain on 2 aliases

The comment says this was stopped because it was unlikely to be
profitable. This is not true if you want to combine vector loads
with multiple components.

For a simple case that looks like

t0 = load t0 ...
t1 = load t0 ...
t2 = load t0 ...
t3 = load t0 ...

t4 = store t0:1, t0:1
  t5 = store t4, t1:0
    t6 = store t5, t2:0
	  t7 = store t6, t3:0

We want to get all of these stores onto a chain
that is a TokenFactor of these N loads. This mostly
solves the AMDGPU merge-stores.ll regressions
with -combiner-alias-analysis for merging vector
stores of vector loads.

llvm-svn: 250138

1 files changed, 2 insertions, 4 deletions

diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index a26f378b49a..b9b37fd0be8 100644
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -14298,14 +14298,12 @@ void DAGCombiner::GatherAllAliases(SDNode *N, SDValue OriginalChain,
     SDValue Chain = Chains.pop_back_val();
 
     // For TokenFactor nodes, look at each operand and only continue up the
-    // chain until we find two aliases.  If we've seen two aliases, assume we'll
-    // find more and revert to original chain since the xform is unlikely to be
-    // profitable.
+    // chain until we reach the depth limit.
     //
     // FIXME: The depth check could be made to return the last non-aliasing
     // chain we found before we hit a tokenfactor rather than the original
     // chain.
-    if (Depth > 6 || Aliases.size() == 2) {
+    if (Depth > 6) {
       Aliases.clear();
       Aliases.push_back(OriginalChain);
       return;