[PATCH v2] net/ice: write rx timestamp to the first mbuf segment in scattered rx

[PATCH v2] net/ice: write rx timestamp to the first mbuf segment in scattered rx

[Martin Weiser]

Previously, the rx timestamp was written to the last segment of the mbuf
chain, which was unexpected.

Signed-off-by: Martin Weiser <martin.weiser@allegro-packets.com>
---
 drivers/net/ice/ice_rxtx.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ice/ice_rxtx.c b/drivers/net/ice/ice_rxtx.c
index 64c4486b4b..e07c6d1f15 100644
--- a/drivers/net/ice/ice_rxtx.c
+++ b/drivers/net/ice/ice_rxtx.c
@@ -2150,7 +2150,7 @@ ice_recv_scattered_pkts(void *rx_queue,
 			}
 			rxq->hw_time_update = rte_get_timer_cycles() /
 					     (rte_get_timer_hz() / 1000);
-			*RTE_MBUF_DYNFIELD(rxm,
+			*RTE_MBUF_DYNFIELD(first_seg,
 					   (ice_timestamp_dynfield_offset),
 					   rte_mbuf_timestamp_t *) = ts_ns;
 			pkt_flags |= ice_timestamp_dynflag;
-- 
2.34.1

Re: [PATCH v2] net/ice: write rx timestamp to the first mbuf segment in scattered rx

[Martin Weiser]

Hi,

sorry for the first patch. I did not manage to get git send-mail to work 
with office365 OAuth2 and Thunderbird messed up the format even in text 
mode.

By the way, are there any plans to support hardware timestamping in the 
vectorized codepath?

Best regards,
Martin


Am 08.08.23 um 13:39 schrieb Martin Weiser:
> Previously, the rx timestamp was written to the last segment of the mbuf
> chain, which was unexpected.
>
> Signed-off-by: Martin Weiser <martin.weiser@allegro-packets.com>
> ---
>   drivers/net/ice/ice_rxtx.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/ice/ice_rxtx.c b/drivers/net/ice/ice_rxtx.c
> index 64c4486b4b..e07c6d1f15 100644
> --- a/drivers/net/ice/ice_rxtx.c
> +++ b/drivers/net/ice/ice_rxtx.c
> @@ -2150,7 +2150,7 @@ ice_recv_scattered_pkts(void *rx_queue,
>   			}
>   			rxq->hw_time_update = rte_get_timer_cycles() /
>   					     (rte_get_timer_hz() / 1000);
> -			*RTE_MBUF_DYNFIELD(rxm,
> +			*RTE_MBUF_DYNFIELD(first_seg,
>   					   (ice_timestamp_dynfield_offset),
>   					   rte_mbuf_timestamp_t *) = ts_ns;
>   			pkt_flags |= ice_timestamp_dynflag;

RE: [PATCH v2] net/ice: write rx timestamp to the first mbuf segment in scattered rx

[Zhang, Qi Z]

> -----Original Message-----
> From: Martin Weiser <martin.weiser@allegro-packets.com>
> Sent: Tuesday, August 8, 2023 7:40 PM
> To: dev@dpdk.org
> Cc: Martin Weiser <martin.weiser@allegro-packets.com>; Yang, Qiming
> <qiming.yang@intel.com>; Zhang, Qi Z <qi.z.zhang@intel.com>
> Subject: [PATCH v2] net/ice: write rx timestamp to the first mbuf segment in
> scattered rx
> 
> Previously, the rx timestamp was written to the last segment of the mbuf chain,
> which was unexpected.

Added:

Fixes: 5543827fc6df ("net/ice: improve performance of Rx timestamp offload")
Cc: stable@dpdk.org> 

> Signed-off-by: Martin Weiser <martin.weiser@allegro-packets.com>
> ---
>  drivers/net/ice/ice_rxtx.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ice/ice_rxtx.c b/drivers/net/ice/ice_rxtx.c index
> 64c4486b4b..e07c6d1f15 100644
> --- a/drivers/net/ice/ice_rxtx.c
> +++ b/drivers/net/ice/ice_rxtx.c
> @@ -2150,7 +2150,7 @@ ice_recv_scattered_pkts(void *rx_queue,
>  			}
>  			rxq->hw_time_update = rte_get_timer_cycles() /
>  					     (rte_get_timer_hz() / 1000);
> -			*RTE_MBUF_DYNFIELD(rxm,
> +			*RTE_MBUF_DYNFIELD(first_seg,
>  					   (ice_timestamp_dynfield_offset),
>  					   rte_mbuf_timestamp_t *) = ts_ns;
>  			pkt_flags |= ice_timestamp_dynflag;
> --
> 2.34.1

Thanks for capture this.

Acked-by: Qi Zhang <qi.z.zhang@intel.com>

Applied to dpdk-next-net-intel.