Re: [Patch v2] mm/vmscan: fix unsequenced modification and access warning

[Michal Hocko]

On Thu 25-05-17 21:43:43, Nick Desaulniers wrote:
> On Tue, May 16, 2017 at 10:27:46AM +0200, Michal Hocko wrote:
> > I guess it is worth reporting this to clang bugzilla. Could you take
> > care of that Nick?
> 
> >From https://bugs.llvm.org//show_bug.cgi?id=33065#c5
> it seems that this is indeed a sequence bug in the previous version of
> this code and not a compiler bug.  You can read that response for the
> properly-cited wording but my TL;DR/understanding is for the given code:
> 
> struct foo bar = {
>   .a = (c = 0),
>   .b = c,
> };
> 
> That the compiler is allowed to reorder the initializations of bar.a and
> bar.b, so what the value of c here might not be what you expect.

This is interesting because what I hear from our gcc people is something
different. I am not in a possition to argue here, though.
-- 
Michal Hocko
SUSE Labs