Re: [boost] [interprocess] atomic_write32 on GCC/x86/x68_64 lacks memory barrier

Subject: Re: [boost] [interprocess] atomic_write32 on GCC/x86/x68_64 lacks memory barrier
From: Peter Dimov (lists_at_[hidden])
Date: 2014-08-11 10:17:08

Lars Hagström wrote:
> Does that "memory" statement at the end add a fence both before and after 
> the assembly statement?

Effectively. To be more precise, it turns the assembly statement into a 
compiler fence. "memory" tells the compiler to assume that the assembly code 
accesses and modifies memory in unpredictable ways, and the effect of that 
is basically a compiler fence.

You should probably use the __asm__ and __volatile__ spellings as a matter 
of style though.