Tune rosalloc buffer sizes.

We now use one-page buffers for size brackets 4-7, instead of two-page
buffers, and the first 8 size brackets for thread-local allocations,
instead of 11.

No slowdown observed with MemAllocTest, EvaluateAndApplyChanges, and
BinaryTrees.

(cherrypick commit c4cd95fa37b7138a0fa26d07c235aa409542aecd)

Bug: 18377775
Change-Id: I311f3adf9cab660d258833b17df7e6d905f73c72
diff --git a/runtime/gc/allocator/rosalloc.cc b/runtime/gc/allocator/rosalloc.cc
index 991b956..7c2474f 100644
--- a/runtime/gc/allocator/rosalloc.cc
+++ b/runtime/gc/allocator/rosalloc.cc
@@ -1772,7 +1772,7 @@
     if (i < 4) {
       numOfPages[i] = 1;
     } else if (i < 8) {
-      numOfPages[i] = 2;
+      numOfPages[i] = 1;
     } else if (i < 16) {
       numOfPages[i] = 4;
     } else if (i < 32) {
diff --git a/runtime/gc/allocator/rosalloc.h b/runtime/gc/allocator/rosalloc.h
index 431686a..3269e10 100644
--- a/runtime/gc/allocator/rosalloc.h
+++ b/runtime/gc/allocator/rosalloc.h
@@ -414,8 +414,7 @@
 
   // We use thread-local runs for the size Brackets whose indexes
   // are less than this index. We use shared (current) runs for the rest.
-
-  static const size_t kNumThreadLocalSizeBrackets = 11;
+  static const size_t kNumThreadLocalSizeBrackets = 8;
 
  private:
   // The base address of the memory region that's managed by this allocator.