Fix wrong allocation size in c8f621c43.
authorAndres Freund <andres@anarazel.de>
Mon, 7 Mar 2016 00:27:20 +0000 (16:27 -0800)
committerAndres Freund <andres@anarazel.de>
Mon, 7 Mar 2016 00:27:20 +0000 (16:27 -0800)
In c8f621c43 I forgot to account for MAXALIGN when allocating a new
tuplebuf in ReorderBufferGetTupleBuf(). That happens to currently not
cause active problems on a number of platforms because the affected
pointer is already aligned, but others, like ppc and hppa, trigger this
in the regression test, due to a debug memset clearing memory.

Fix that.

Backpatch: 9.4, like the previous commit.

src/backend/replication/logical/reorderbuffer.c

index a1479fe949f59724a86d5466a678f0e7a4c1fe51..8eadb4bed180c96458564148700ec1edc6ab4c6f 100644 (file)
@@ -481,7 +481,8 @@ ReorderBufferGetTupleBuf(ReorderBuffer *rb, Size tuple_len)
    {
        tuple = (ReorderBufferTupleBuf *)
            MemoryContextAlloc(rb->context,
-                              sizeof(ReorderBufferTupleBuf) + alloc_len);
+                              sizeof(ReorderBufferTupleBuf) +
+                              MAXIMUM_ALIGNOF + alloc_len);
        tuple->alloc_tuple_size = alloc_len;
        tuple->tuple.t_data = ReorderBufferTupleBufData(tuple);
    }