http://tech.puredanger.com/2008/05/08/javaone-cliff-click-on-data-races/ Alex Miller's technical blog Sun, 27 Jul 2008 01:12:02 +0000 http://wordpress.org/?v=2.0.2 http://tech.puredanger.com/2008/05/08/javaone-cliff-click-on-data-races/#comment-46485 Fri, 09 May 2008 05:36:34 +0000 http://tech.puredanger.com/2008/05/08/javaone-cliff-click-on-data-races/#comment-46485 Double-checked locking is safe if you make the instance a volatile. I'm not sure why you think otherwise - Cliff said the same during the talk. Double-checked locking is safe if you make the instance a volatile. I’m not sure why you think otherwise - Cliff said the same during the talk.

]]> http://tech.puredanger.com/2008/05/08/javaone-cliff-click-on-data-races/#comment-46401 Fri, 09 May 2008 00:11:50 +0000 http://tech.puredanger.com/2008/05/08/javaone-cliff-click-on-data-races/#comment-46401 I could have sworn that I read somewhere that the changes to the memory model in Java 5 meant that double checked locking is now safe if you use volatile. The JVM now has to ensure that no memory access is reordered relative to accesses to volatiles, even absent monitor ownership. I could have sworn that I read somewhere that the changes to the memory model in Java 5 meant that double checked locking is now safe if you use volatile. The JVM now has to ensure that no memory access is reordered relative to accesses to volatiles, even absent monitor ownership.

]]>