Re: [PATCH] perf tests: initialize sa.sa_flags

[Arnaldo Carvalho de Melo]

Em Wed, Mar 02, 2016 at 02:21:27PM +0100, Peter Zijlstra escreveu:
> On Wed, Mar 02, 2016 at 10:03:50AM -0300, Arnaldo Carvalho de Melo wrote:
> > > Would not something like:
> > > 
> > > 	sa = (struct sigaction){
> > > 		.sa_sigaction = segfault_handler,
> > > 	};
> > > 	sigfillset(&sa.sa_mask);
> > > 
> > > Be better?
> > 
> > I thought about that, but isn't that set in stone? This would be a 4
> > liner, while his is a one' :-)
> 
> Dunno, you're right that its rather unlikely struct sigaction is going
> to grow another member, but I like the above pattern better in general,
> makes it harder to end up with uninitalized bits.
> 
> When performance matters the above pattern isn't ideal, but that should
> not be a concern here.

Right, I also always use :


	struct foo bar = {
		.baz = 1,
		.name = "whatever",
	};

Even more compact than using that cast. But didn't bother changing in
this case.

- Arnaldo