Do no reprobe for a supported protocol when enabling a pointing

device of type PMS_STANDARD, probing during autoconf(9) is enough!.

In such case, the device can be a simple PS/2 mouse, a unsupported
touchpad or a downgraded touchpad if something bad happened.  But
it is very unlikely that reprobing will improve the situation and
transform a standard mouse into a multitouch-aware touchpad ;)

This fixes the 12 seconds delay seen on various Dell laptops (E4310
and Latitude D630) when starting Xorg, reported by Kārlis Miķelsons
on bugs@.

ok stsp@, dcoppa@, shadchin@

diff --git a/sys/dev/pckbc/pms.c b/sys/dev/pckbc/pms.c
index af1b908..4a6de17 100644 (file)
--- a/sys/dev/pckbc/pms.c
+++ b/sys/dev/pckbc/pms.c
@@ -1,4 +1,4 @@
-/* $OpenBSD: pms.c,v 1.49 2013/10/30 18:00:56 shadchin Exp $ */
+/* $OpenBSD: pms.c,v 1.50 2014/04/25 10:33:36 mpi Exp $ */
 /* $NetBSD: psm.c,v 1.11 2000/06/05 22:20:57 sommerfeld Exp $ */
 
 /*-
@@ -735,7 +735,7 @@ pms_change_state(struct pms_softc *sc, int newstate, int dev)
                        pckbc_flush(sc->sc_kbctag, PCKBC_AUX_SLOT);
 
                pms_reset(sc);
-               if (sc->protocol->type == PMS_STANDARD ||
+               if (sc->protocol->enable != NULL &&
                    sc->protocol->enable(sc) == 0)
                        pms_protocol_lookup(sc);