Date: Tue, 27 May 2008 19:34:46 GMT From: Sam Leffler <sam@FreeBSD.org> To: Perforce Change Reviews <perforce@freebsd.org> Subject: PERFORCE change 142401 for review Message-ID: <200805271934.m4RJYksW026261@repoman.freebsd.org>
next in thread | raw e-mail | index | archive | help
http://perforce.freebsd.org/chv.cgi?CH=142401 Change 142401 by sam@sam_ebb on 2008/05/27 19:34:28 close a race on detach by reordering bpfdetach and taskqueue_free; add some comments to explain Affected files ... .. //depot/projects/vap/sys/dev/ath/if_ath.c#81 edit Differences ... ==== //depot/projects/vap/sys/dev/ath/if_ath.c#81 (text+ko) ==== @@ -678,13 +678,15 @@ DPRINTF(sc, ATH_DEBUG_ANY, "%s: if_flags %x\n", __func__, ifp->if_flags); - ath_stop(ifp); - bpfdetach(ifp); /* * NB: the order of these is important: + * o stop the chip so no more interrupts will fire * o call the 802.11 layer before detaching the hal to * insure callbacks into the driver to delete global * key cache entries can be handled + * o free the taskqueue which drains any pending tasks + * o reclaim the bpf tap now that we know nothing will use + * it (e.g. rx processing from the task q thread) * o reclaim the tx queue data structures after calling * the 802.11 layer as we'll get called back to reclaim * node state and potentially want to use them @@ -692,12 +694,14 @@ * it last * Other than that, it's straightforward... */ + ath_stop(ifp); ieee80211_ifdetach(ifp->if_l2com); + taskqueue_free(sc->sc_tq); + bpfdetach(ifp); #ifdef ATH_TX99_DIAG if (sc->sc_tx99 != NULL) sc->sc_tx99->detach(sc->sc_tx99); #endif - taskqueue_free(sc->sc_tq); ath_rate_detach(sc->sc_rc); ath_desc_free(sc); ath_tx_cleanup(sc);
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200805271934.m4RJYksW026261>