Re: process 'stuck' at exit.

[Dave Jones <davej@redhat.com>]

On Tue, Dec 10, 2013 at 02:58:19PM -0800, Darren Hart wrote:
 > On Tue, 2013-12-10 at 14:48 -0800, Linus Torvalds wrote:
 > > On Tue, Dec 10, 2013 at 2:42 PM, Thomas Gleixner <tglx@linutronix.de> wrote:
 > > >
 > > > And yes, I remember that we do not do an extra check for the fshared
 > > > case, because get_user_pages_fast() should do it for us already. If
 > > > not we are fubared not only in the futex code.
 > > 
 > > Yeah. It turns out we do do the access check indirectly - by looking
 > > at the PAGE_USER bit, even if we don't necessarily check the actual
 > > limits. So get_user_pages_fast() is fine.
 > > 
 > > > But there is a subtle detail:
 > > 
 > > Yup, see my email from ten minutes ago, we found the same thing. And
 > > that would seem to explain the endless loop, and also the timing
 > > (since Dave mentions he started doing large-pages lately).
 > > 
 > > So I think the "__get_user_pages_fast(address, 1, !ro, &page)" thing
 > > should work.
 > > 
 > > Dave, can you re-create that trinity run and test that patch? I think
 > > we've got this, but it might be nice to leave the hung machine up and
 > > running until it's verified.. Although I don't really see what else we
 > > could need or get out of it, so..
 > 
 > Would it be possible to limit the options to only pass FUTEX_CMP_REQUEUE
 > and a read-only uaddr? That should improve confidence when it doesn't
 > fail :-)

easy enough to hack into the code yeah. A bit complicated to come up with
a sensible grammar for a command line parser for such cases sadly.

I'll give the patch a try after dinner.

	Dave