[dpdk-dev] [PATCH] timer: fix break list when timer_cb reset running timer

Sanford, Robert rsanford at akamai.com
Sat Jul 23 00:14:23 CEST 2016



On 7/17/16 2:08 PM, "Hiroyuki Mikita" <h.mikita89 at gmail.com> wrote:

>When timer_cb resets another running timer on the same lcore,
>the list of expired timers is chained to the pending-list.
>This commit prevents a running timer from being reset
>by not its own timer_cb.
>
>Signed-off-by: Hiroyuki Mikita <h.mikita89 at gmail.com>
>---
> lib/librte_timer/rte_timer.c | 12 ++++++++++--
> 1 file changed, 10 insertions(+), 2 deletions(-)
>
>diff --git a/lib/librte_timer/rte_timer.c b/lib/librte_timer/rte_timer.c
>index 3dcdab5..00e80c9 100644
>--- a/lib/librte_timer/rte_timer.c
>+++ b/lib/librte_timer/rte_timer.c
>@@ -69,6 +69,9 @@ struct priv_timer {
> 
> 	unsigned prev_lcore;              /**< used for lcore round robin */
> 
>+	/** running timer on this lcore now */
>+	struct rte_timer *running_tim;
>+
> #ifdef RTE_LIBRTE_TIMER_DEBUG
> 	/** per-lcore statistics */
> 	struct rte_timer_debug_stats stats;
>@@ -135,9 +138,12 @@ timer_set_config_state(struct rte_timer *tim,
> 	while (success == 0) {
> 		prev_status.u32 = tim->status.u32;
> 
>-		/* timer is running on another core, exit */
>+		/* timer is running on another core
>+		 * or ready to run on local core, exit
>+		 */
> 		if (prev_status.state == RTE_TIMER_RUNNING &&
>-		    prev_status.owner != (uint16_t)lcore_id)
>+		    (prev_status.owner != (uint16_t)lcore_id ||
>+		     tim != priv_timer[lcore_id].running_tim))
> 			return -1;
> 
> 		/* timer is being configured on another core */
>@@ -580,6 +586,7 @@ void rte_timer_manage(void)
> 	for (tim = run_first_tim; tim != NULL; tim = next_tim) {
> 		next_tim = tim->sl_next[0];
> 		priv_timer[lcore_id].updated = 0;
>+		priv_timer[lcore_id].running_tim = tim;
> 
> 		/* execute callback function with list unlocked */
> 		tim->f(tim, tim->arg);
>@@ -610,6 +617,7 @@ void rte_timer_manage(void)
> 			rte_spinlock_unlock(&priv_timer[lcore_id].list_lock);
> 		}
> 	}
>+	priv_timer[lcore_id].running_tim = NULL;
> }
> 
> /* dump statistics about timers */
>-- 
>2.7.4
>

Acked-by: Robert Sanford <rsanford at akamai.com>


I tested the three timer patches with app/test timer_autotest and
timer_racecond_autotest, and additional private tests.



More information about the dev mailing list