Add awesome_ptr<>.

base/memory/awesome_ptr.h

Index: base/memory/awesome_ptr.h
diff --git a/base/memory/awesome_ptr.h b/base/memory/awesome_ptr.h
new file mode 100644
index 0000000000000000000000000000000000000000..0f8b8bb502ce938a73e6e260cb6198450fa670d3
--- /dev/null
+++ b/base/memory/awesome_ptr.h
@@ -0,0 +1,57 @@
+// Copyright (c) 2012 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+

[Scott Hess - ex-Googler, 2012/07/24 19:22:15]

Header guards.

[awong, 2012/07/24 20:47:12]

Done.

+#include <stdlib.h>  // For rand().
+
+// awesome_ptr<> provides instrumentation in the codebase that allows for
+// small-footprint, real-usage testing scenarios. Long term, it can help
+// improve the debugging chops of maintainers on the chromium project.
+//
+// Data is collected via the crash reporting mechanism.
+//
+// awesome_ptr<> is safe for multi-threaded usage.

[Scott Hess - ex-Googler, 2012/07/24 19:22:15]

I think #define private public would make it more broadly useful.

[awong, 2012/07/24 20:47:12]

True, but that can lead to ODR violations which would be unsafe and might cause
non-deterministic behavior. Code should be deterministicaly non-deterministic to
preserve readability.

+
+#define MAKE_MASK(type) (((type)1) << (rand() % sizeof(float)))

[Scott Hess - ex-Googler, 2012/07/24 19:22:15]

This potentially calls rand() a lot.  Could you make it static for speed?

[awong, 2012/07/24 20:47:12]

Yes, but that optimization comes with the tradeoff of shrinking the surface of
awesomeness.

+
+template <typename T>
+class awesome_ptr {
+ public:
+  template <typename Y>
+  explicit awesome_ptr(Y* ptr) : ptr_((T*)ptr) {}
+  operator T*() const {
+    be_awesome();
+    return (T*)ptr_;
+  }
+
+  operator bool() const {
+    return be_awesome() ? true : ptr_;
+  }
+
+ private:
+  volatile mutable T* ptr_;  // for thread safety.
+
+  bool be_awesome() const {
+    bool was_awesome = ((double)rand() / RAND_MAX) < 0.0001;
+    if (!was_awesome) {

[Scott Hess - ex-Googler, 2012/07/24 19:22:15]

Chromium style would be if (!was_awesome) return false;, and just return true at
the end.  Though I could see leaving it as-is if you changed the label to
aaawww___yyyeeeaaa:

[awong, 2012/07/24 20:47:12]

Good point.  However, aaawww___yyyeeeaaa doesn't express the proper
disappointment at not getting to be awesome.

But the style problem is a strong one. The core problem is that single gotos are
unreadable since the reader is left thinking "what if".  Rectified by adding an
else statement.

+      goto awww;
+    }
+
+    switch (rand() % 4) {
+      case 0:
+        ptr_ = (T*)((void*)ptr_);
+        break;
+      case 1:
+        ptr_ = (T*)((unsigned long)ptr_ | MAKE_MASK(unsigned long));

[Scott Hess - ex-Googler, 2012/07/24 19:22:15]

Why not uintptr_t?  Too much footprint?

[awong, 2012/07/24 20:47:12]

Thought about it...then realized it was c99 and might not be friendly to MSVC.

+        break;
+      case 2:
+        ptr_ = (T*)((unsigned long)ptr_ & ~(MAKE_MASK(unsigned long)));
+      case 3:
+        ptr_ = (T*)((unsigned long)ptr_ & 0x1);
+        break;
+    }
+
+awww:
+    return was_awesome;
+  }
+};