[dpdk-dev] net/enic: set L4 checksum flags for IPv6 packets
Checks
Commit Message
From: Hyong Youb Kim <hyonkim@cisco.com>
enic_cq_rx_to_pkt_flags() currently sets checksum good/bad flags only
for IPv4. The hardware actually validates the TCP/UDP checksum of
IPv6 packets too. Set PKT_RX_L4_CKSUM_{GOOD,BAD} accordingly.
Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
Reviewed-by: John Daley <johndale@cisco.com>
---
drivers/net/enic/enic_rxtx.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
Comments
On 1/23/2018 1:05 AM, John Daley wrote:
> From: Hyong Youb Kim <hyonkim@cisco.com>
>
> enic_cq_rx_to_pkt_flags() currently sets checksum good/bad flags only
> for IPv4. The hardware actually validates the TCP/UDP checksum of
> IPv6 packets too. Set PKT_RX_L4_CKSUM_{GOOD,BAD} accordingly.
>
> Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
> Reviewed-by: John Daley <johndale@cisco.com>
> ---
> drivers/net/enic/enic_rxtx.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/net/enic/enic_rxtx.c b/drivers/net/enic/enic_rxtx.c
> index 98902caa0..8157697a0 100644
> --- a/drivers/net/enic/enic_rxtx.c
> +++ b/drivers/net/enic/enic_rxtx.c
> @@ -185,14 +185,14 @@ enic_cq_rx_to_pkt_flags(struct cq_desc *cqd, struct rte_mbuf *mbuf)
> }
>
> /* checksum flags */
> - if (mbuf->packet_type & RTE_PTYPE_L3_IPV4) {
> + if (mbuf->packet_type & (RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L3_IPV6)) {
> if (!enic_cq_rx_desc_csum_not_calc(cqrd)) {
> uint32_t l4_flags;
> l4_flags = mbuf->packet_type & RTE_PTYPE_L4_MASK;
>
> if (enic_cq_rx_desc_ipv4_csum_ok(cqrd))
> pkt_flags |= PKT_RX_IP_CKSUM_GOOD;
> - else
> + else if (mbuf->packet_type & RTE_PTYPE_L3_IPV4)
This looks like conflicting with commit log.
Is pkt_flags intentionally not set for this case?
If so can you update commit log to say only PKT_RX_IP_CKSUM_GOOD set for ipv6?
> pkt_flags |= PKT_RX_IP_CKSUM_BAD;
>
> if (l4_flags == RTE_PTYPE_L4_UDP ||
>
On Wed, Jan 24, 2018 at 05:18:37PM +0000, Ferruh Yigit wrote:
> On 1/23/2018 1:05 AM, John Daley wrote:
> > From: Hyong Youb Kim <hyonkim@cisco.com>
> >
> > enic_cq_rx_to_pkt_flags() currently sets checksum good/bad flags only
> > for IPv4. The hardware actually validates the TCP/UDP checksum of
> > IPv6 packets too. Set PKT_RX_L4_CKSUM_{GOOD,BAD} accordingly.
[...]
> > /* checksum flags */
> > - if (mbuf->packet_type & RTE_PTYPE_L3_IPV4) {
> > + if (mbuf->packet_type & (RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L3_IPV6)) {
> > if (!enic_cq_rx_desc_csum_not_calc(cqrd)) {
> > uint32_t l4_flags;
> > l4_flags = mbuf->packet_type & RTE_PTYPE_L4_MASK;
> >
> > if (enic_cq_rx_desc_ipv4_csum_ok(cqrd))
> > pkt_flags |= PKT_RX_IP_CKSUM_GOOD;
> > - else
> > + else if (mbuf->packet_type & RTE_PTYPE_L3_IPV4)
>
> This looks like conflicting with commit log.
> Is pkt_flags intentionally not set for this case?
> If so can you update commit log to say only PKT_RX_IP_CKSUM_GOOD set for ipv6?
>
> > pkt_flags |= PKT_RX_IP_CKSUM_BAD;
Yes, it is intentional. IPv6 has no IP header checksum, and
PKT_RX_IP_CKSUM_{GOOD,BAD} does not apply. So, the code does not set
PKT_RX_IP_CKSUM for IPv6. As the commit log says, for IPv6, we are
only setting L4 checksum flags.
-Hyong
On 1/24/2018 5:30 PM, Hyong Youb Kim wrote:
> On Wed, Jan 24, 2018 at 05:18:37PM +0000, Ferruh Yigit wrote:
>> On 1/23/2018 1:05 AM, John Daley wrote:
>>> From: Hyong Youb Kim <hyonkim@cisco.com>
>>>
>>> enic_cq_rx_to_pkt_flags() currently sets checksum good/bad flags only
>>> for IPv4. The hardware actually validates the TCP/UDP checksum of
>>> IPv6 packets too. Set PKT_RX_L4_CKSUM_{GOOD,BAD} accordingly.
> [...]
>>> /* checksum flags */
>>> - if (mbuf->packet_type & RTE_PTYPE_L3_IPV4) {
>>> + if (mbuf->packet_type & (RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L3_IPV6)) {
>>> if (!enic_cq_rx_desc_csum_not_calc(cqrd)) {
>>> uint32_t l4_flags;
>>> l4_flags = mbuf->packet_type & RTE_PTYPE_L4_MASK;
>>>
>>> if (enic_cq_rx_desc_ipv4_csum_ok(cqrd))
>>> pkt_flags |= PKT_RX_IP_CKSUM_GOOD;
>>> - else
>>> + else if (mbuf->packet_type & RTE_PTYPE_L3_IPV4)
>>
>> This looks like conflicting with commit log.
>> Is pkt_flags intentionally not set for this case?
>> If so can you update commit log to say only PKT_RX_IP_CKSUM_GOOD set for ipv6?
>>
>>> pkt_flags |= PKT_RX_IP_CKSUM_BAD;
>
> Yes, it is intentional. IPv6 has no IP header checksum, and
> PKT_RX_IP_CKSUM_{GOOD,BAD} does not apply. So, the code does not set
> PKT_RX_IP_CKSUM for IPv6. As the commit log says, for IPv6, we are
> only setting L4 checksum flags.
Ahh, I see.
And I assume enic_cq_rx_desc_ipv4_csum_ok() always false for IPv6.
>
> -Hyong
>
On Wed, Jan 24, 2018 at 05:45:48PM +0000, Ferruh Yigit wrote:
> On 1/24/2018 5:30 PM, Hyong Youb Kim wrote:
> > On Wed, Jan 24, 2018 at 05:18:37PM +0000, Ferruh Yigit wrote:
> >> On 1/23/2018 1:05 AM, John Daley wrote:
> >>> From: Hyong Youb Kim <hyonkim@cisco.com>
> >>>
> >>> enic_cq_rx_to_pkt_flags() currently sets checksum good/bad flags only
> >>> for IPv4. The hardware actually validates the TCP/UDP checksum of
> >>> IPv6 packets too. Set PKT_RX_L4_CKSUM_{GOOD,BAD} accordingly.
> > [...]
> >>> /* checksum flags */
> >>> - if (mbuf->packet_type & RTE_PTYPE_L3_IPV4) {
> >>> + if (mbuf->packet_type & (RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L3_IPV6)) {
> >>> if (!enic_cq_rx_desc_csum_not_calc(cqrd)) {
> >>> uint32_t l4_flags;
> >>> l4_flags = mbuf->packet_type & RTE_PTYPE_L4_MASK;
> >>>
> >>> if (enic_cq_rx_desc_ipv4_csum_ok(cqrd))
> >>> pkt_flags |= PKT_RX_IP_CKSUM_GOOD;
> >>> - else
> >>> + else if (mbuf->packet_type & RTE_PTYPE_L3_IPV4)
> >>
> >> This looks like conflicting with commit log.
> >> Is pkt_flags intentionally not set for this case?
> >> If so can you update commit log to say only PKT_RX_IP_CKSUM_GOOD set for ipv6?
> >>
> >>> pkt_flags |= PKT_RX_IP_CKSUM_BAD;
> >
> > Yes, it is intentional. IPv6 has no IP header checksum, and
> > PKT_RX_IP_CKSUM_{GOOD,BAD} does not apply. So, the code does not set
> > PKT_RX_IP_CKSUM for IPv6. As the commit log says, for IPv6, we are
> > only setting L4 checksum flags.
>
> Ahh, I see.
> And I assume enic_cq_rx_desc_ipv4_csum_ok() always false for IPv6.
>
Yes, that is correct. enic_cq_rx_desc_ipv4_csum_ok() returns true only
for IPv4. It is a bit confusing..
-Hyong
On 1/23/2018 1:05 AM, John Daley wrote:
> From: Hyong Youb Kim <hyonkim@cisco.com>
>
> enic_cq_rx_to_pkt_flags() currently sets checksum good/bad flags only
> for IPv4. The hardware actually validates the TCP/UDP checksum of
> IPv6 packets too. Set PKT_RX_L4_CKSUM_{GOOD,BAD} accordingly.
>
> Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
> Reviewed-by: John Daley <johndale@cisco.com>
Applied to dpdk-next-net/master, thanks.
@@ -185,14 +185,14 @@ enic_cq_rx_to_pkt_flags(struct cq_desc *cqd, struct rte_mbuf *mbuf)
}
/* checksum flags */
- if (mbuf->packet_type & RTE_PTYPE_L3_IPV4) {
+ if (mbuf->packet_type & (RTE_PTYPE_L3_IPV4 | RTE_PTYPE_L3_IPV6)) {
if (!enic_cq_rx_desc_csum_not_calc(cqrd)) {
uint32_t l4_flags;
l4_flags = mbuf->packet_type & RTE_PTYPE_L4_MASK;
if (enic_cq_rx_desc_ipv4_csum_ok(cqrd))
pkt_flags |= PKT_RX_IP_CKSUM_GOOD;
- else
+ else if (mbuf->packet_type & RTE_PTYPE_L3_IPV4)
pkt_flags |= PKT_RX_IP_CKSUM_BAD;
if (l4_flags == RTE_PTYPE_L4_UDP ||