From: Daniel Colascione <dancol@fb.com>
Date: Sat, 12 Mar 2016 21:31:46 +0000 (-0800)
Subject: Explain union safety in MicroLock comment
X-Git-Tag: 2016.07.26~447
X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=b29f38a428c896be68d3c6ddc4f2f2f31de9d3aa;p=folly.git

Explain union safety in MicroLock comment

Reviewed By: ot

Differential Revision: D3045850

fb-gh-sync-id: ee19b146c43410d8d8804c9bfe79c3811394b10e
shipit-source-id: ee19b146c43410d8d8804c9bfe79c3811394b10e
---

diff --git a/folly/MicroLock.h b/folly/MicroLock.h
index 07aa40b8..159e3341 100644
--- a/folly/MicroLock.h
+++ b/folly/MicroLock.h
@@ -43,6 +43,10 @@ namespace folly {
  * limit yourself to MicroLock slot zero, which will use the two
  * least-significant bits in the bottom byte.
  *
+ * (Note that such a union is safe only because MicroLock is based on
+ * a character type, and even under a strict interpretation of C++'s
+ * aliasing rules, character types may alias anything.)
+ *
  * MicroLock uses a dirty trick: it actually operates on the full
  * word-size, word-aligned bit of memory into which it is embedded.
  * It never modifies bits outside the ones it's defined to modify, but