From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by dpdk.org (Postfix) with ESMTP id 7428F37A4 for ; Mon, 2 Feb 2015 03:03:45 +0100 (CET) Received: from orsmga001.jf.intel.com ([10.7.209.18]) by fmsmga102.fm.intel.com with ESMTP; 01 Feb 2015 18:03:29 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.09,503,1418112000"; d="scan'208";a="645936195" Received: from shvmail01.sh.intel.com ([10.239.29.42]) by orsmga001.jf.intel.com with ESMTP; 01 Feb 2015 18:03:29 -0800 Received: from shecgisg004.sh.intel.com (shecgisg004.sh.intel.com [10.239.29.89]) by shvmail01.sh.intel.com with ESMTP id t1223Q2R014561; Mon, 2 Feb 2015 10:03:26 +0800 Received: from shecgisg004.sh.intel.com (localhost [127.0.0.1]) by shecgisg004.sh.intel.com (8.13.6/8.13.6/SuSE Linux 0.8) with ESMTP id t1223MZ3013784; Mon, 2 Feb 2015 10:03:24 +0800 Received: (from cliang18@localhost) by shecgisg004.sh.intel.com (8.13.6/8.13.6/Submit) id t1223Mg9013780; Mon, 2 Feb 2015 10:03:22 +0800 From: Cunming Liang To: dev@dpdk.org Date: Mon, 2 Feb 2015 10:02:38 +0800 Message-Id: <1422842559-13617-17-git-send-email-cunming.liang@intel.com> X-Mailer: git-send-email 1.7.4.1 In-Reply-To: <1422842559-13617-1-git-send-email-cunming.liang@intel.com> References: <1422491072-5114-1-git-send-email-cunming.liang@intel.com> <1422842559-13617-1-git-send-email-cunming.liang@intel.com> Subject: [dpdk-dev] [PATCH v4 16/17] ring: add sched_yield to avoid spin forever X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 02 Feb 2015 02:03:46 -0000 Add a sched_yield() syscall if the thread spins for too long, waiting other thread to finish its operations on the ring. That gives pre-empted thread a chance to proceed and finish with ring enqnue/dequeue operation. The purpose is to reduce contention on the ring. Signed-off-by: Cunming Liang --- lib/librte_ring/rte_ring.h | 35 +++++++++++++++++++++++++++++------ 1 file changed, 29 insertions(+), 6 deletions(-) diff --git a/lib/librte_ring/rte_ring.h b/lib/librte_ring/rte_ring.h index 39bacdd..c402c73 100644 --- a/lib/librte_ring/rte_ring.h +++ b/lib/librte_ring/rte_ring.h @@ -126,6 +126,7 @@ struct rte_ring_debug_stats { #define RTE_RING_NAMESIZE 32 /**< The maximum length of a ring name. */ #define RTE_RING_MZ_PREFIX "RG_" +#define RTE_RING_PAUSE_REP 0x100 /**< yield after num of times pause. */ /** * An RTE ring structure. @@ -410,7 +411,7 @@ __rte_ring_mp_do_enqueue(struct rte_ring *r, void * const *obj_table, uint32_t cons_tail, free_entries; const unsigned max = n; int success; - unsigned i; + unsigned i, rep; uint32_t mask = r->prod.mask; int ret; @@ -468,8 +469,19 @@ __rte_ring_mp_do_enqueue(struct rte_ring *r, void * const *obj_table, * If there are other enqueues in progress that preceded us, * we need to wait for them to complete */ - while (unlikely(r->prod.tail != prod_head)) - rte_pause(); + do { + /* avoid spin too long waiting for other thread finish */ + for (rep = RTE_RING_PAUSE_REP; + rep != 0 && r->prod.tail != prod_head; rep--) + rte_pause(); + + /* + * It gives pre-empted thread a chance to proceed and + * finish with ring enqnue operation. + */ + if (rep == 0) + sched_yield(); + } while (rep == 0); r->prod.tail = prod_next; return ret; @@ -589,7 +601,7 @@ __rte_ring_mc_do_dequeue(struct rte_ring *r, void **obj_table, uint32_t cons_next, entries; const unsigned max = n; int success; - unsigned i; + unsigned i, rep; uint32_t mask = r->prod.mask; /* move cons.head atomically */ @@ -634,8 +646,19 @@ __rte_ring_mc_do_dequeue(struct rte_ring *r, void **obj_table, * If there are other dequeues in progress that preceded us, * we need to wait for them to complete */ - while (unlikely(r->cons.tail != cons_head)) - rte_pause(); + do { + /* avoid spin too long waiting for other thread finish */ + for (rep = RTE_RING_PAUSE_REP; + rep != 0 && r->cons.tail != cons_head; rep--) + rte_pause(); + + /* + * It gives pre-empted thread a chance to proceed and + * finish with ring denqnue operation. + */ + if (rep == 0) + sched_yield(); + } while (rep == 0); __RING_STAT_ADD(r, deq_success, n); r->cons.tail = cons_next; -- 1.8.1.4