Daily Archives: May 15, 2012
When writing lock-free code in C or C++, one must often take special care to enforce correct memory ordering. Otherwise, surprising things can happen.
Intel lists several such surprises in Volume 3, Section 8.2.3 of their x86/64 Architecture Specification. Here’s one of the simplest examples. Suppose you have two integers
Y somewhere in memory, both initially 0. Two processors, running in parallel, execute the following machine code:
Don’t be thrown off by the use of assembly language in this example. It’s really the best way to illustrate CPU ordering. Each processor stores 1 into one of the integer variables, then loads the other integer into a register. (r1 and r2 are just placeholder names for actual x86 registers, such as