Refactor select to reduce contention and hide internal implementation

details from consumers.

 - Track individual selecters on a per-descriptor basis such that there
   are no longer collisions and after sleeping for events only those
   descriptors which triggered events must be rescaned.
 - Protect the selinfo (per descriptor) structure with a mtx pool mutex.
   mtx pool mutexes were chosen to preserve api compatibility with
   existing code which does nothing but bzero() to setup selinfo
   structures.
 - Use a per-thread wait channel rather than a global wait channel.
 - Hide select implementation details in a seltd structure which is
   opaque to the rest of the kernel.
 - Provide a 'selsocket' interface for those kernel consumers who wish to
   select on a socket when they have no fd so they no longer have to
   be aware of select implementation details.

Tested by:	kris
Reviewed on:	arch

1 files changed, 6 insertions, 3 deletions

diff --git a/sys/kern/kern_event.c b/sys/kern/kern_event.c
index aa446b6..4d75822 100644
--- a/sys/kern/kern_event.c
+++ b/sys/kern/kern_event.c
@@ -1400,7 +1400,8 @@ kqueue_poll(struct file *fp, int events, struct ucred *active_cred,
 			revents |= events & (POLLIN | POLLRDNORM);
 		} else {
 			selrecord(td, &kq->kq_sel);
-			kq->kq_state |= KQ_SEL;
+			if (SEL_WAITING(&kq->kq_sel))
+				kq->kq_state |= KQ_SEL;
 		}
 	}
 	kqueue_release(kq, 1);
@@ -1486,8 +1487,9 @@ kqueue_close(struct file *fp, struct thread *td)
 	}
 
 	if ((kq->kq_state & KQ_SEL) == KQ_SEL) {
-		kq->kq_state &= ~KQ_SEL;
 		selwakeuppri(&kq->kq_sel, PSOCK);
+		if (!SEL_WAITING(&kq->kq_sel))
+			kq->kq_state &= ~KQ_SEL;
 	}
 
 	KQ_UNLOCK(kq);
@@ -1522,8 +1524,9 @@ kqueue_wakeup(struct kqueue *kq)
 		wakeup(kq);
 	}
 	if ((kq->kq_state & KQ_SEL) == KQ_SEL) {
-		kq->kq_state &= ~KQ_SEL;
 		selwakeuppri(&kq->kq_sel, PSOCK);
+		if (!SEL_WAITING(&kq->kq_sel))
+			kq->kq_state &= ~KQ_SEL;
 	}
 	if (!knlist_empty(&kq->kq_sel.si_note))
 		kqueue_schedtask(kq);