diff mbox

[dpdk-dev,4/4] net/ena: calculate partial checksum if DF bit is disabled

Message ID 1491562138-2178-5-git-send-email-mw@semihalf.com (mailing list archive)
State Superseded, archived
Delegated to: Ferruh Yigit
Headers

Checks

Context Check Description
ci/checkpatch success coding style OK
ci/Intel-compilation success Compilation OK

Commit Message

Marcin Wojtas April 7, 2017, 10:48 a.m. UTC 
  From: Michal Krawczyk <mk@semihalf.com>

When TSO is disabled we still have to calculate partial checksum if DF bit
if turned off. This is caused by firmware bug.

If application will not set m2_len field, we assume we that it was Ethernet
frame because we have to look inside the packet to check for the DF flag.
To make it work properly, PMD is assuming that before sending
packet application called function rte_eth_tx_prepare().

Signed-off-by: Michal Krawczyk <mk@semihalf.com>
---
 drivers/net/ena/ena_ethdev.c | 30 ++++++++++++++++++++----------
 1 file changed, 20 insertions(+), 10 deletions(-)

Comments

Jan Mędala April 10, 2017, 11:05 a.m. UTC | #1 
I would recommend to check if packet type is IPv4 before processing IPv4 header for DF flag.
This patch can break logic and go to unknown state when mbuf will contain IPv6 packet. I believe that in case of IPv6 pkt the loop should be skipped to next mbuf, if exists.

Best regards,
Jan
diff mbox

Patch

diff --git a/drivers/net/ena/ena_ethdev.c b/drivers/net/ena/ena_ethdev.c
index 54ba5c1..bb2a8ba 100644
--- a/drivers/net/ena/ena_ethdev.c
+++ b/drivers/net/ena/ena_ethdev.c
@@ -1599,14 +1599,30 @@  static uint16_t eth_ena_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
 	uint64_t ol_flags;
 	uint16_t frag_field;
 
-	/* ENA needs partial checksum for TSO packets only, skip early */
-	if (!tx_ring->adapter->tso4_supported)
-		return nb_pkts;
-
 	for (i = 0; i != nb_pkts; i++) {
 		m = tx_pkts[i];
 		ol_flags = m->ol_flags;
 
+		/* If there was not L2 header length specified, assume it is
+		 * length of the ethernet header.
+		 */
+		if (unlikely(m->l2_len == 0))
+			m->l2_len = sizeof(struct ether_hdr);
+
+		ip_hdr = rte_pktmbuf_mtod_offset(m, struct ipv4_hdr *,
+						 m->l2_len);
+		frag_field = rte_be_to_cpu_16(ip_hdr->fragment_offset);
+
+		if ((frag_field & IPV4_HDR_DF_FLAG) != 0) {
+			m->packet_type |= RTE_PTYPE_L4_NONFRAG;
+
+			/* If IPv4 header has DF flag enabled and TSO support is
+			 * disabled, partial chcecksum should not be calculated.
+			 */
+			if (!tx_ring->adapter->tso4_supported)
+				continue;
+		}
+
 		if ((ol_flags & ENA_TX_OFFLOAD_NOTSUP_MASK) != 0 ||
 				(ol_flags & PKT_TX_L4_MASK) ==
 				PKT_TX_SCTP_CKSUM) {
@@ -1625,12 +1641,6 @@  static uint16_t eth_ena_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
 		if (!(m->ol_flags & PKT_TX_IPV4))
 			continue;
 
-		ip_hdr = rte_pktmbuf_mtod_offset(m, struct ipv4_hdr *,
-						 m->l2_len);
-		frag_field = rte_be_to_cpu_16(ip_hdr->fragment_offset);
-		if (frag_field & IPV4_HDR_DF_FLAG)
-			continue;
-
 		/* In case we are supposed to TSO and have DF not set (DF=0)
 		 * hardware must be provided with partial checksum, otherwise
 		 * it will take care of necessary calculations.