On Fri, 2014-10-17 at 17:38 +0100, Catalin Marinas wrote: > Commit b0c29f79ecea (futexes: Avoid taking the hb->lock if there's > nothing to wake up) changes the futex code to avoid taking a lock when > there are no waiters. This code has been subsequently fixed in commit > 11d4616bd07f (futex: revert back to the explicit waiter counting code). > Both the original commit and the fix-up rely on get_futex_key_refs() to > always imply a barrier. > > However, for private futexes, none of the cases in the switch statement > of get_futex_key_refs() would be hit and the function completes without > a memory barrier as required before checking the "waiters" in > futex_wake() -> hb_waiters_pending().
Good catch, glad I ran into this thread (my email recently changed). Private process futex (PTHREAD_PROCESS_PRIVATE) have no reference on an inode or mm so it would need the explicit barrier in those cases. > The consequence is a race with a > thread waiting on a futex on another CPU, allowing the waker thread to > read "waiters == 0" while the waiter thread to have read "futex_val == > locked" (in kernel). Yeah missing wakeups are a strong sign of a problem with the hb_waiters_pending() side. > Without this fix, the problem (user space deadlocks) can be seen with > Android bionic's mutex implementation on an arm64 multi-cluster system. > Signed-off-by: Catalin Marinas <[email protected]> > Reported-by: Matteo Franchin <[email protected]> > Fixes: b0c29f79ecea (futexes: Avoid taking the hb->lock if there's nothing to > wake up) > Cc: <[email protected]> > Cc: Davidlohr Bueso <[email protected]> > Cc: Linus Torvalds <[email protected]> > Cc: Darren Hart <[email protected]> > Cc: Thomas Gleixner <[email protected]> > Cc: Peter Zijlstra <[email protected]> > Cc: Ingo Molnar <[email protected]> > Cc: Paul E. McKenney <[email protected]> > --- > kernel/futex.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/kernel/futex.c b/kernel/futex.c > index 815d7af2ffe8..f3a3a071283c 100644 > --- a/kernel/futex.c > +++ b/kernel/futex.c > @@ -343,6 +343,8 @@ static void get_futex_key_refs(union futex_key *key) > case FUT_OFF_MMSHARED: > futex_get_mm(key); /* implies MB (B) */ > break; > + default: > + smp_mb(); /* explicit MB (B) */ > } Should we comment that this default is for the private futex case? Otherwise: Acked-by: Davidlohr Bueso <[email protected]> -- 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/
