[dpdk-dev] [PATCH 0/2] ixgbe: Two fixes for RX scatter functions.

[dpdk-dev] [PATCH 0/2] ixgbe: Two fixes for RX scatter functions.

[Konstantin Ananyev]

Konstantin Ananyev (2):
  ixgbe: fix scalar scatter RX doesn't take into account CRC length
  ixgbe: fix vector scatter RX could produce wrong nb_segs value

 drivers/net/ixgbe/ixgbe_rxtx.c     | 19 +++++++++++++++++++
 drivers/net/ixgbe/ixgbe_rxtx_vec.c |  2 ++
 2 files changed, 21 insertions(+)

-- 
1.8.3.1

[dpdk-dev] [PATCH 1/2] ixgbe: fix scalar scatter RX doesn't take into account CRC length

[Konstantin Ananyev]

For 2.1 release, in attempt to minimize number of RX routines to support,
ixgbe scatter and ixgbe LRO RX routines were merged into one
that can handle both cases.
Though I completely missed the fact, that while LRO could only be used
when HW CRC strip is enabled, scatter RX should work for both cases
(HW CRC strip on/off).
That patch restores missed functionality.

Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 drivers/net/ixgbe/ixgbe_rxtx.c | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_rxtx.c b/drivers/net/ixgbe/ixgbe_rxtx.c
index a0c8847..a1b25aa 100644
--- a/drivers/net/ixgbe/ixgbe_rxtx.c
+++ b/drivers/net/ixgbe/ixgbe_rxtx.c
@@ -1817,6 +1817,25 @@ next_desc:
 		ixgbe_fill_cluster_head_buf(first_seg, &rxd, rxq->port_id,
 					    staterr);
 
+		/*
+		 * Deal with the case, when HW CRC srip is disabled.
+		 * That can't happen when LRO is enabled, but still could
+		 * happen for scattered RX mode.
+		 */
+		first_seg->pkt_len -= rxq->crc_len;
+		if (unlikely(rxm->data_len <= rxq->crc_len)) {
+			struct rte_mbuf *lp;
+
+			for (lp = first_seg; lp->next != rxm; lp = lp->next)
+				;
+
+			first_seg->nb_segs--;
+			lp->data_len -= rxq->crc_len - rxm->data_len;
+			lp->next = NULL;
+			rte_pktmbuf_free_seg(rxm);
+		} else
+			rxm->data_len -= rxq->crc_len;
+
 		/* Prefetch data of first segment, if configured to do so. */
 		rte_packet_prefetch((char *)first_seg->buf_addr +
 			first_seg->data_off);
-- 
1.8.3.1

[dpdk-dev] [PATCH 2/2] ixgbe: fix vector scatter RX could produce wrong nb_segs value

[Konstantin Ananyev]

Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 drivers/net/ixgbe/ixgbe_rxtx_vec.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_rxtx_vec.c b/drivers/net/ixgbe/ixgbe_rxtx_vec.c
index 6c1647e..1c16dec 100644
--- a/drivers/net/ixgbe/ixgbe_rxtx_vec.c
+++ b/drivers/net/ixgbe/ixgbe_rxtx_vec.c
@@ -497,6 +497,8 @@ reassemble_packets(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_bufs,
 				else {
 					/* free up last mbuf */
 					struct rte_mbuf *secondlast = start;
+
+					start->nb_segs--;
 					while (secondlast->next != end)
 						secondlast = secondlast->next;
 					secondlast->data_len -= (rxq->crc_len -
-- 
1.8.3.1

[dpdk-dev] [PATCHv2 0/2] ixgbe: Two fixes for RX scatter functions.

[Konstantin Ananyev]

Konstantin Ananyev (2):
  ixgbe: fix scalar scatter RX doesn't take into account CRC length
  ixgbe: fix vector scatter RX could produce wrong nb_segs value

 drivers/net/ixgbe/ixgbe_rxtx.c     | 19 +++++++++++++++++++
 drivers/net/ixgbe/ixgbe_rxtx_vec.c |  2 ++
 2 files changed, 21 insertions(+)

-- 
1.8.3.1

[dpdk-dev] [PATCHv2 1/2] ixgbe: fix scalar scatter RX doesn't take into account CRC length

[Konstantin Ananyev]

For 2.1 release, in attempt to minimize number of RX routines to support,
ixgbe scatter and ixgbe LRO RX routines were merged into one
that can handle both cases.
Though I completely missed the fact, that while LRO could only be used
when HW CRC strip is enabled, scatter RX should work for both cases
(HW CRC strip on/off).
That patch restores missed functionality.

Fixes: 9d8a92628f21 ("ixgbe: remove simple scalar scattered Rx method")

v2 changes:
- updated commit message with 'Fixes'.

Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 drivers/net/ixgbe/ixgbe_rxtx.c | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_rxtx.c b/drivers/net/ixgbe/ixgbe_rxtx.c
index a0c8847..a1b25aa 100644
--- a/drivers/net/ixgbe/ixgbe_rxtx.c
+++ b/drivers/net/ixgbe/ixgbe_rxtx.c
@@ -1817,6 +1817,25 @@ next_desc:
 		ixgbe_fill_cluster_head_buf(first_seg, &rxd, rxq->port_id,
 					    staterr);
 
+		/*
+		 * Deal with the case, when HW CRC srip is disabled.
+		 * That can't happen when LRO is enabled, but still could
+		 * happen for scattered RX mode.
+		 */
+		first_seg->pkt_len -= rxq->crc_len;
+		if (unlikely(rxm->data_len <= rxq->crc_len)) {
+			struct rte_mbuf *lp;
+
+			for (lp = first_seg; lp->next != rxm; lp = lp->next)
+				;
+
+			first_seg->nb_segs--;
+			lp->data_len -= rxq->crc_len - rxm->data_len;
+			lp->next = NULL;
+			rte_pktmbuf_free_seg(rxm);
+		} else
+			rxm->data_len -= rxq->crc_len;
+
 		/* Prefetch data of first segment, if configured to do so. */
 		rte_packet_prefetch((char *)first_seg->buf_addr +
 			first_seg->data_off);
-- 
1.8.3.1

[dpdk-dev] [PATCHv2 2/2] ixgbe: fix vector scatter RX could produce wrong nb_segs value

[Konstantin Ananyev]

Fixes: cf4b4708a88a (ixgbe: improve slow-path perf with vector scattered Rx)

v2: changes:
- updated commit message with 'Fixes'.

Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
---
 drivers/net/ixgbe/ixgbe_rxtx_vec.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/ixgbe/ixgbe_rxtx_vec.c b/drivers/net/ixgbe/ixgbe_rxtx_vec.c
index 6c1647e..1c16dec 100644
--- a/drivers/net/ixgbe/ixgbe_rxtx_vec.c
+++ b/drivers/net/ixgbe/ixgbe_rxtx_vec.c
@@ -497,6 +497,8 @@ reassemble_packets(struct ixgbe_rx_queue *rxq, struct rte_mbuf **rx_bufs,
 				else {
 					/* free up last mbuf */
 					struct rte_mbuf *secondlast = start;
+
+					start->nb_segs--;
 					while (secondlast->next != end)
 						secondlast = secondlast->next;
 					secondlast->data_len -= (rxq->crc_len -
-- 
1.8.3.1

Re: [dpdk-dev] [PATCHv2 0/2] ixgbe: Two fixes for RX scatter functions.

[Lu, Wenzhuo]

Hi,
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>

Re: [dpdk-dev] [PATCHv2 0/2] ixgbe: Two fixes for RX scatter functions.

[Thomas Monjalon]

> Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>

Applied, thanks