Date: Fri, 8 Jul 2016 02:52:39 +0000 (UTC) From: Don Lewis <truckman@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-10@freebsd.org Subject: svn commit: r302422 - stable/10/sys/netpfil/ipfw Message-ID: <201607080252.u682qdx7059411@repo.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: truckman Date: Fri Jul 8 02:52:39 2016 New Revision: 302422 URL: https://svnweb.freebsd.org/changeset/base/302422 Log: MFC r302338 Fix a race condition between the main thread in aqm_pie_cleanup() and the callout thread that can cause a kernel panic. Always do the final cleanup in the callout thread by passing a separate callout function for that task to callout_reset_sbt(). Protect the ref_count decrement in the callout with DN_BH_WLOCK(). All other ref_count manipulation is protected with this lock. There is still a tiny window between ref_count reaching zero and the end of the callout function where it is unsafe to unload the module. Fixing this would require the use of callout_drain(), but this can't be done because dummynet holds a mutex and callout_drain() might sleep. Remove the callout_pending(), callout_active(), and callout_deactivate() calls from calculate_drop_prob(). They are not needed because this callout uses callout_init_mtx(). Submitted by: Rasool Al-Saadi <ralsaadi@swin.edu.au> Differential Revision: https://reviews.freebsd.org/D6928 Modified: stable/10/sys/netpfil/ipfw/dn_aqm_pie.c Directory Properties: stable/10/ (props changed) Modified: stable/10/sys/netpfil/ipfw/dn_aqm_pie.c ============================================================================== --- stable/10/sys/netpfil/ipfw/dn_aqm_pie.c Fri Jul 8 02:34:04 2016 (r302421) +++ stable/10/sys/netpfil/ipfw/dn_aqm_pie.c Fri Jul 8 02:52:39 2016 (r302422) @@ -207,24 +207,6 @@ calculate_drop_prob(void *x) struct dn_aqm_pie_parms *pprms; struct pie_status *pst = (struct pie_status *) x; - /* dealing with race condition */ - if (callout_pending(&pst->aqm_pie_callout)) { - /* callout was reset */ - mtx_unlock(&pst->lock_mtx); - return; - } - - if (!callout_active(&pst->aqm_pie_callout)) { - /* callout was stopped */ - mtx_unlock(&pst->lock_mtx); - mtx_destroy(&pst->lock_mtx); - free(x, M_DUMMYNET); - //pst->pq->aqm_status = NULL; - pie_desc.ref_count--; - return; - } - callout_deactivate(&pst->aqm_pie_callout); - pprms = pst->parms; prob = pst->drop_prob; @@ -576,7 +558,7 @@ aqm_pie_init(struct dn_queue *q) do { /* exit with break when error occurs*/ if (!pprms){ - D("AQM_PIE is not configured"); + DX(2, "AQM_PIE is not configured"); err = EINVAL; break; } @@ -615,6 +597,22 @@ aqm_pie_init(struct dn_queue *q) } /* + * Callout function to destroy pie mtx and free PIE status memory + */ +static void +pie_callout_cleanup(void *x) +{ + struct pie_status *pst = (struct pie_status *) x; + + mtx_unlock(&pst->lock_mtx); + mtx_destroy(&pst->lock_mtx); + free(x, M_DUMMYNET); + DN_BH_WLOCK(); + pie_desc.ref_count--; + DN_BH_WUNLOCK(); +} + +/* * Clean up PIE status for queue 'q' * Destroy memory allocated for PIE status. */ @@ -640,22 +638,19 @@ aqm_pie_cleanup(struct dn_queue *q) return 1; } + /* + * Free PIE status allocated memory using pie_callout_cleanup() callout + * function to avoid any potential race. + * We reset aqm_pie_callout to call pie_callout_cleanup() in next 1um. This + * stops the scheduled calculate_drop_prob() callout and call pie_callout_cleanup() + * which does memory freeing. + */ mtx_lock(&pst->lock_mtx); + callout_reset_sbt(&pst->aqm_pie_callout, + SBT_1US, 0, pie_callout_cleanup, pst, 0); + q->aqm_status = NULL; + mtx_unlock(&pst->lock_mtx); - /* stop callout timer */ - if (callout_stop(&pst->aqm_pie_callout) || !(pst->sflags & PIE_ACTIVE)) { - mtx_unlock(&pst->lock_mtx); - mtx_destroy(&pst->lock_mtx); - free(q->aqm_status, M_DUMMYNET); - q->aqm_status = NULL; - pie_desc.ref_count--; - return 0; - } else { - q->aqm_status = NULL; - mtx_unlock(&pst->lock_mtx); - DX(2, "PIE callout has not been stoped from cleanup!"); - return EBUSY; - } return 0; }
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201607080252.u682qdx7059411>