[dpdk-dev] [PATCH v1 1/1] power: fix multi-queue scale mode for pmd mgmt

[David Hunt]

Hi Anatoly,

On 21/7/2021 3:26 PM, Anatoly Burakov wrote:
> Currently in scale mode, multi-queue initialization will attempt to
> initialize and de-initialize the per-lcore power library structures
> multiple times. Fix it to only do this whenever we either enabling
> first queue or disabling last queue.
>
> Fixes: 5dff9a72b0ef ("power: support callbacks for multiple Rx queues")
>
> Signed-off-by: Anatoly Burakov <anatoly.burakov at intel.com>
> ---
>   lib/power/rte_power_pmd_mgmt.c | 19 +++++++++++++------
>   1 file changed, 13 insertions(+), 6 deletions(-)
>
> diff --git a/lib/power/rte_power_pmd_mgmt.c b/lib/power/rte_power_pmd_mgmt.c
> index 2586204b93..0ce40f0875 100644
> --- a/lib/power/rte_power_pmd_mgmt.c
> +++ b/lib/power/rte_power_pmd_mgmt.c
> @@ -534,11 +534,15 @@ rte_power_ethdev_pmgmt_queue_enable(unsigned int lcore_id, uint16_t port_id,
>   		clb = get_monitor_callback();
>   		break;
>   	case RTE_POWER_MGMT_TYPE_SCALE:
> -		/* check if we can add a new queue */
> -		ret = check_scale(lcore_id);
> -		if (ret < 0)
> -			goto end;
>   		clb = clb_scale_freq;
> +
> +		/* we only have to check this when enabling first queue */
> +		if (lcore_cfg->pwr_mgmt_state != PMD_MGMT_DISABLED)
> +			break;
> +		/* check if we can add a new queue */
> +		ret = check_scale(lcore_id);
> +		if (ret < 0)
> +			goto end;
>   		break;
>   	case RTE_POWER_MGMT_TYPE_PAUSE:
>   		/* figure out various time-to-tsc conversions */
> @@ -633,9 +637,12 @@ rte_power_ethdev_pmgmt_queue_disable(unsigned int lcore_id,
>   		rte_eth_remove_rx_callback(port_id, queue_id, queue_cfg->cb);
>   		break;
>   	case RTE_POWER_MGMT_TYPE_SCALE:
> -		rte_power_freq_max(lcore_id);
>   		rte_eth_remove_rx_callback(port_id, queue_id, queue_cfg->cb);
> -		rte_power_exit(lcore_id);
> +		/* disable power library on this lcore if this was last queue */
> +		if (lcore_cfg->pwr_mgmt_state == PMD_MGMT_DISABLED) {
> +			rte_power_freq_max(lcore_id);
> +			rte_power_exit(lcore_id);
> +		}
>   		break;
>   	}
>   	/*


Fix looks good. Previous to this patch, was failing on adding second 
queue to a core, now with this patch, succeeds.

Tested-by: David Hunt <david.hunt at intel.com>