On Mon, Jul 20, 2015 at 05:18:31PM +0200, Joerg Roedel wrote:
> On Mon, Jul 20, 2015 at 05:10:00PM +0200, Borislav Petkov wrote:
> > On Mon, Jul 20, 2015 at 05:02:40PM +0200, Joerg Roedel wrote:
> > > I have seen a report where this happens on bare metal, when the change
> > > to the cpu_active bit becomes visible on the other CPU significantly
> > > later than the the cpu_online bit. This happened on a pretty big machine
> > > with 88 cores.
> >
> > So how about what I proposed at the end of my previous mail?
>
> Oh sorry, I missed that. Setting cpu_active first should work on x86,
> where writes become visible in the order they are executed. But this
> function is in generic code and I have no idea what this change might
> break on other architectures.
>
> In the end cpu_active means that the scheduler can push tasks to the
> CPU, no idea if some arch code breaks when the scheduler is already
> working on a CPU before it becomes visibly online.
Hmm...
Let's run it by Peter.
@Peter: see the first patch in the mail for the problem of which cpumask
bit to test wrt scheduler and migrating tasks to newly appearing
cores...
Thanks.
--
Regards/Gruss,
Boris.
ECO tip #101: Trim your mails when you reply.
--
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
