Fix for PR18102.

Issue outcomes from DAGCombiner::MergeConsequtiveStores, more precisely from
mem-ops sequence sorting.

Consider, how MergeConsequtiveStores works for next example:

store i8 1, a[0]
store i8 2, a[1]
store i8 3, a[1]   ; a[1] again.
return   ; DAG starts here

1. Method will collect all the 3 stores.
2. It sorts them by distance from the base pointer (farthest with highest
index).
3. It takes first consecutive non-overlapping stores and (if possible) replaces
them with a single store instruction.

The point is, we can't determine here which 'store' instruction
would be the second after sorting ('store 2' or 'store 3').
It happens that 'store 3' would be the second, and 'store 2' would be the third.

So after merging we have the next result:

store i16 (1 | 3 << 8), base   ; is a[0] but bit-casted to i16
store i8 2, a[1]

So actually we swapped 'store 3' and 'store 2' and got wrong contents in a[1].

Fix: In sort routine just also take into account mem-op sequence number.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@200201 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
index b95a6c1198dbc6e507215afeb0707ef8391d288a..97c19d0e9b1869d208afb21aaaf4c6e854ee409f 100644 (file)
--- a/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -8761,7 +8761,10 @@ struct MemOpLink {
 // base ptr.
 struct ConsecutiveMemoryChainSorter {
   bool operator()(MemOpLink LHS, MemOpLink RHS) {
-    return LHS.OffsetFromBase < RHS.OffsetFromBase;
+    return
+        LHS.OffsetFromBase < RHS.OffsetFromBase ||
+        (LHS.OffsetFromBase == RHS.OffsetFromBase &&
+         LHS.SequenceNum > RHS.SequenceNum);
   }
 };
 

diff --git a/test/CodeGen/Generic/stores-merging.ll b/test/CodeGen/Generic/stores-merging.ll
new file mode 100644 (file)
index 0000000..61dea08
--- /dev/null
+++ b/test/CodeGen/Generic/stores-merging.ll
@@ -0,0 +1,23 @@
+; RUN: llc < %s | FileCheck %s
+
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+%structTy = type { i8, i32, i32 }
+
+@e = common global %structTy zeroinitializer, align 4
+
+; CHECK-LABEL: f
+define void @f() {
+entry:
+
+; CHECK:   movabsq     $528280977409, %rax
+; CHECK:   movq    %rax, e+4(%rip)
+; CHECK:   movl    $456, e+8(%rip)
+
+  store i32 1, i32* getelementptr inbounds (%structTy* @e, i64 0, i32 1), align 4
+  store i32 123, i32* getelementptr inbounds (%structTy* @e, i64 0, i32 2), align 4
+  store i32 456, i32* getelementptr inbounds (%structTy* @e, i64 0, i32 2), align 4
+  ret void
+}
+