Re: 3.18 nohz + audit regression (Re: [PATCH] context_tracking: Restore previous state in schedule_user)

[Frederic Weisbecker]

On Wed, Dec 03, 2014 at 08:58:46PM -0800, Linus Torvalds wrote:
> On Wed, Dec 3, 2014 at 5:29 PM, Andy Lutomirski <luto@xxxxxxxxxxxxxx> wrote:
> >
> > So, to summarize the choices for 3.18:
> >
> > 4. This patch.
> 
> I've applied it. The alternatives look worse, and the patch doesn't
> look bad. In many ways it looks better than the old user_exit/enter
> pair, although obviously the "schedule_user()" name is kind of odd
> now. Whatever.

That doesn't look like the safest approach to me. Somehow I felt more
comfortable with exception_enter/exit on the audit function.

A few other archs use schedule_user() as well, it's likely fine but I'm
not sure how subtle their path is.

Besides, is it possible that the audit function gets called after
syscall_trace_leave() or do_notify_resume()? If so, the patch won't fix
the issue entirely.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/