ACCESS_ONCE() and compiler bugs

The ACCESS_ONCE() macro is used throughout the kernel to ensure that code generated by the compiler will access the indicated variable once (and only once); see this article for details on how it works and when its use is necessary. When that article was written (2012), there were 200 invocations of ACCESS_ONCE() in the kernel; now there are over 700 of them. Like many low-level techniques for concurrency management, ACCESS_ONCE() relies on trickery that is best hidden from view. And, like such techniques, it may break if the compiler changes behavior or, as has been seen recently, contains a bug.

ACCESS_ONCE() 宏在整个内核中广泛使用，用于确保编译器生成的代码对指定变量的访问只发生一次（且仅一次）；关于它的工作原理以及何时需要使用，可以参考这篇文章。撰写该文时（2012 年），内核中约有 200 处使用 ACCESS_ONCE()，而现在已经超过 700 处。像许多底层并发管理技术一样，ACCESS_ONCE() 依赖一些最好被隐藏的技巧。而且，和这类技术一样，如果编译器行为发生变化，或者像最近发现的情况那样出现了 bug，它也可能出问题。

Back in November, Christian Borntraeger posted a message regarding the interactions between ACCESS_ONCE() and an obscure GCC bug. To understand the problem, it is worth looking at the macro, which is defined simply in current kernels (in <linux/compiler.h>):

去年十一月，Christian Borntraeger 发布了一则消息，讨论了 ACCESS_ONCE() 与一个隐蔽的 GCC bug 之间的交互问题。为了理解这个问题，有必要看看当前内核中对该宏的定义（位于 <linux/compiler.h>）：

#define ACCESS_ONCE(x) (*(volatile typeof(x) *)&(x))

In short, ACCESS_ONCE() forces the variable to be treated as being a volatile type, even though it (like almost all variables in the kernel) is not declared that way. The problem reported by Christian