Re: [dpdk-dev] [PATCH v4] net/fm10k: add support for check descriptor status APIs

["Zhang, Qi Z" <qi.z.zhang@intel.com>]

Hi Wei:

> -----Original Message-----
> From: Zhao1, Wei
> Sent: Friday, June 22, 2018 4:20 PM
> To: dev@dpdk.org
> Cc: Lu, Wenzhuo <wenzhuo.lu@intel.com>; Zhang, Qi Z
> <qi.z.zhang@intel.com>; Zhao1, Wei <wei.zhao1@intel.com>
> Subject: [PATCH v4] net/fm10k: add support for check descriptor status APIs
> 
> Fm10k nic need to support check descriptor status APIs, they are
> rte_eth_rx_descriptor_status and rte_eth_tx_descriptor_status.
> So, this patch add ops with function pointer that enable feature.
> 
> Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
> 
> ---
> 
> v2:
> -fix DD check error in tx descriptor
> 
> v3:
> -fix DD check index error
> 
> v4:
> -fix error in RS bit list poll
> 
> ---
> ---
>  doc/guides/rel_notes/release_18_08.rst |  6 +++
>  drivers/net/fm10k/fm10k.h              |  7 +++
>  drivers/net/fm10k/fm10k_ethdev.c       |  2 +
>  drivers/net/fm10k/fm10k_rxtx.c         | 78
> ++++++++++++++++++++++++++++++++++
>  4 files changed, 93 insertions(+)
> 
> diff --git a/doc/guides/rel_notes/release_18_08.rst
> b/doc/guides/rel_notes/release_18_08.rst
> index 5bc23c5..5baa09e 100644
> --- a/doc/guides/rel_notes/release_18_08.rst
> +++ b/doc/guides/rel_notes/release_18_08.rst
> @@ -41,6 +41,12 @@ New Features
>       Also, make sure to start the actual text at the margin.
>       =========================================================
> 
> +* **Added fm10k ethernet driver to support check descriptor status
> +APIs.**
> +
> +    Fm10k nic need to support check descriptor status APIs, they are
> +    rte_eth_rx_descriptor_status and rte_eth_tx_descriptor_status.
> +    add ops pointer with new function which enable feature.
> +
> 
>  API Changes
>  -----------
> diff --git a/drivers/net/fm10k/fm10k.h b/drivers/net/fm10k/fm10k.h index
> ef30780..1bc2c18 100644
> --- a/drivers/net/fm10k/fm10k.h
> +++ b/drivers/net/fm10k/fm10k.h
> @@ -329,6 +329,13 @@ uint16_t fm10k_recv_scattered_pkts(void *rx_queue,
> int  fm10k_dev_rx_descriptor_done(void *rx_queue, uint16_t offset);
> 
> +int
> +fm10k_dev_rx_descriptor_status(void *rx_queue, uint16_t offset);
> +
> +int
> +fm10k_dev_tx_descriptor_status(void *rx_queue, uint16_t offset);
> +
> +
>  uint16_t fm10k_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
>  	uint16_t nb_pkts);
> 
> diff --git a/drivers/net/fm10k/fm10k_ethdev.c
> b/drivers/net/fm10k/fm10k_ethdev.c
> index 3ff1b0e..ea2f2bf 100644
> --- a/drivers/net/fm10k/fm10k_ethdev.c
> +++ b/drivers/net/fm10k/fm10k_ethdev.c
> @@ -2837,6 +2837,8 @@ static const struct eth_dev_ops fm10k_eth_dev_ops
> = {
>  	.tx_queue_setup		= fm10k_tx_queue_setup,
>  	.tx_queue_release	= fm10k_tx_queue_release,
>  	.rx_descriptor_done	= fm10k_dev_rx_descriptor_done,
> +	.rx_descriptor_status = fm10k_dev_rx_descriptor_status,
> +	.tx_descriptor_status = fm10k_dev_tx_descriptor_status,
>  	.rx_queue_intr_enable	= fm10k_dev_rx_queue_intr_enable,
>  	.rx_queue_intr_disable	= fm10k_dev_rx_queue_intr_disable,
>  	.reta_update		= fm10k_reta_update,
> diff --git a/drivers/net/fm10k/fm10k_rxtx.c b/drivers/net/fm10k/fm10k_rxtx.c
> index 9320748..daa2b32 100644
> --- a/drivers/net/fm10k/fm10k_rxtx.c
> +++ b/drivers/net/fm10k/fm10k_rxtx.c
> @@ -389,6 +389,84 @@ fm10k_dev_rx_descriptor_done(void *rx_queue,
> uint16_t offset)
>  	return ret;
>  }
> 
> +int
> +fm10k_dev_rx_descriptor_status(void *rx_queue, uint16_t offset) {
> +	volatile union fm10k_rx_desc *rxdp;
> +	struct fm10k_rx_queue *rxq = rx_queue;
> +	uint16_t nb_hold, trigger_last;
> +	uint16_t desc;
> +	int ret;
> +
> +	if (unlikely(offset >= rxq->nb_desc)) {
> +		PMD_DRV_LOG(ERR, "Invalid RX descriptor offset %u", offset);
> +		return 0;
> +	}
> +
> +	if (rxq->next_trigger < rxq->alloc_thresh)
> +		trigger_last = rxq->next_trigger +
> +					rxq->nb_desc - rxq->alloc_thresh;
> +	else
> +		trigger_last = rxq->next_trigger - rxq->alloc_thresh;
> +
> +	if (rxq->next_dd < trigger_last)
> +		nb_hold = rxq->next_dd + rxq->nb_desc - trigger_last;
> +	else
> +		nb_hold = rxq->next_dd - trigger_last;
> +
> +	if (offset >= rxq->nb_desc - nb_hold)
> +		return RTE_ETH_RX_DESC_UNAVAIL;
> +
> +	desc = rxq->next_dd + offset;
> +	if (desc >= rxq->nb_desc)
> +		desc -= rxq->nb_desc;
> +
> +	rxdp = &rxq->hw_ring[desc];
> +
> +	ret = !!(rxdp->w.status &
> +			rte_cpu_to_le_16(FM10K_RXD_STATUS_DD));
> +
> +	return ret;
> +}
> +
> +int
> +fm10k_dev_tx_descriptor_status(void *tx_queue, uint16_t offset) {
> +	volatile struct fm10k_tx_desc *txdp;
> +	struct fm10k_tx_queue *txq = tx_queue;
> +	uint16_t desc;
> +	uint16_t count, next_rs = txq->nb_desc;
> +	struct fifo rs_tracker = txq->rs_tracker;
> +	struct fifo *r = &rs_tracker;
> +
> +	if (unlikely(offset >= txq->nb_desc))
> +		return -EINVAL;
> +
> +	desc = txq->next_free + offset;
> +	/* go to next desc that has the RS bit */
> +	desc = (desc / txq->rs_thresh + 1) *
> +		txq->rs_thresh - 1;
> +
> +	if (desc >= txq->nb_desc) {
> +		desc -= txq->nb_desc;
> +		if (desc >= txq->nb_desc)
> +			desc -= txq->nb_desc;
> +	}
> +
> +	r->head = r->list;
> +	for (count = 0; r->head != r->endp; count++) {

I didn't see the necessary to have "count" here.

Others I didn't see issue, btw, could you rebase on dpdk-next-net-intel?

Thanks
Qi


> +		if (*r->head >= desc && *r->head < next_rs)
> +			next_rs = *r->head;
> +		++r->head;
> +	}
> +
> +	txdp = &txq->hw_ring[next_rs];
> +	if (txdp->flags & FM10K_TXD_FLAG_DONE)
> +		return RTE_ETH_TX_DESC_DONE;
> +
> +	return RTE_ETH_TX_DESC_FULL;
> +}
> +
>  /*
>   * Free multiple TX mbuf at a time if they are in the same pool
>   *
> --
> 2.7.5