[v2] net/netvsc: fix the calculation of checksums based on mbuf flag
Checks
Commit Message
From: Long Li <longli@microsoft.com>
The netvsc should use RTE_MBUF_F_TX_L4_MASK and check the masked value to
decide the correct way to calculate checksums.
Not checking for RTE_MBUF_F_TX_L4_MASK results in incorrect RNDIS packets
sent to VSP and incorrect checksums calculated by the VSP.
Fixes: 4e9c73e96e ("net/netvsc: add Hyper-V network device")
Cc: stable@dpdk.org
Signed-off-by: Long Li <longli@microsoft.com>
---
drivers/net/netvsc/hn_rxtx.c | 13 +++++++++----
1 file changed, 9 insertions(+), 4 deletions(-)
Comments
On 3/24/2022 5:46 PM, longli@linuxonhyperv.com wrote:
> From: Long Li <longli@microsoft.com>
>
> The netvsc should use RTE_MBUF_F_TX_L4_MASK and check the masked value to
> decide the correct way to calculate checksums.
>
> Not checking for RTE_MBUF_F_TX_L4_MASK results in incorrect RNDIS packets
> sent to VSP and incorrect checksums calculated by the VSP.
>
> Fixes: 4e9c73e96e ("net/netvsc: add Hyper-V network device")
> Cc: stable@dpdk.org
> Signed-off-by: Long Li <longli@microsoft.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@xilinx.com>
Moving ack from previous version:
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Applied to dpdk-next-net/main, thanks.
On 3/24/2022 5:46 PM, longli@linuxonhyperv.com wrote:
> From: Long Li <longli@microsoft.com>
>
> The netvsc should use RTE_MBUF_F_TX_L4_MASK and check the masked value to
> decide the correct way to calculate checksums.
>
> Not checking for RTE_MBUF_F_TX_L4_MASK results in incorrect RNDIS packets
> sent to VSP and incorrect checksums calculated by the VSP.
>
> Fixes: 4e9c73e96e ("net/netvsc: add Hyper-V network device")
> Cc: stable@dpdk.org
> Signed-off-by: Long Li <longli@microsoft.com>
> ---
> drivers/net/netvsc/hn_rxtx.c | 13 +++++++++----
> 1 file changed, 9 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/netvsc/hn_rxtx.c b/drivers/net/netvsc/hn_rxtx.c
> index 028f176c7e..34f40be5b8 100644
> --- a/drivers/net/netvsc/hn_rxtx.c
> +++ b/drivers/net/netvsc/hn_rxtx.c
> @@ -1348,8 +1348,11 @@ static void hn_encap(struct rndis_packet_msg *pkt,
> *pi_data = NDIS_LSO2_INFO_MAKEIPV4(hlen,
> m->tso_segsz);
> }
> - } else if (m->ol_flags &
> - (RTE_MBUF_F_TX_TCP_CKSUM | RTE_MBUF_F_TX_UDP_CKSUM | RTE_MBUF_F_TX_IP_CKSUM)) {
> + } else if ((m->ol_flags & RTE_MBUF_F_TX_L4_MASK) ==
> + RTE_MBUF_F_TX_TCP_CKSUM ||
> + (m->ol_flags & RTE_MBUF_F_TX_L4_MASK) ==
> + RTE_MBUF_F_TX_UDP_CKSUM ||
> + (m->ol_flags & RTE_MBUF_F_TX_IP_CKSUM)) {
As far as I can see following drivers also has similar issue, can
maintainers (cc'ed) of below drivers check:
bnxt
dpaa
hnic
ionic
liquidio
mlx4
mvneta
mvpp2
qede
On Tue, Apr 26, 2022 at 2:57 PM Ferruh Yigit <ferruh.yigit@xilinx.com> wrote:
>
> On 3/24/2022 5:46 PM, longli@linuxonhyperv.com wrote:
> > From: Long Li <longli@microsoft.com>
> >
> > The netvsc should use RTE_MBUF_F_TX_L4_MASK and check the masked value to
> > decide the correct way to calculate checksums.
> >
> > Not checking for RTE_MBUF_F_TX_L4_MASK results in incorrect RNDIS packets
> > sent to VSP and incorrect checksums calculated by the VSP.
> >
> > Fixes: 4e9c73e96e ("net/netvsc: add Hyper-V network device")
> > Cc: stable@dpdk.org
> > Signed-off-by: Long Li <longli@microsoft.com>
> > ---
> > drivers/net/netvsc/hn_rxtx.c | 13 +++++++++----
> > 1 file changed, 9 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/net/netvsc/hn_rxtx.c b/drivers/net/netvsc/hn_rxtx.c
> > index 028f176c7e..34f40be5b8 100644
> > --- a/drivers/net/netvsc/hn_rxtx.c
> > +++ b/drivers/net/netvsc/hn_rxtx.c
> > @@ -1348,8 +1348,11 @@ static void hn_encap(struct rndis_packet_msg *pkt,
> > *pi_data = NDIS_LSO2_INFO_MAKEIPV4(hlen,
> > m->tso_segsz);
> > }
> > - } else if (m->ol_flags &
> > - (RTE_MBUF_F_TX_TCP_CKSUM | RTE_MBUF_F_TX_UDP_CKSUM | RTE_MBUF_F_TX_IP_CKSUM)) {
> > + } else if ((m->ol_flags & RTE_MBUF_F_TX_L4_MASK) ==
> > + RTE_MBUF_F_TX_TCP_CKSUM ||
> > + (m->ol_flags & RTE_MBUF_F_TX_L4_MASK) ==
> > + RTE_MBUF_F_TX_UDP_CKSUM ||
> > + (m->ol_flags & RTE_MBUF_F_TX_IP_CKSUM)) {
>
> As far as I can see following drivers also has similar issue, can
> maintainers (cc'ed) of below drivers check:
>
> bnxt
ACK
> dpaa
> hnic
> ionic
> liquidio
> mlx4
> mvneta
> mvpp2
> qede
@@ -1348,8 +1348,11 @@ static void hn_encap(struct rndis_packet_msg *pkt,
*pi_data = NDIS_LSO2_INFO_MAKEIPV4(hlen,
m->tso_segsz);
}
- } else if (m->ol_flags &
- (RTE_MBUF_F_TX_TCP_CKSUM | RTE_MBUF_F_TX_UDP_CKSUM | RTE_MBUF_F_TX_IP_CKSUM)) {
+ } else if ((m->ol_flags & RTE_MBUF_F_TX_L4_MASK) ==
+ RTE_MBUF_F_TX_TCP_CKSUM ||
+ (m->ol_flags & RTE_MBUF_F_TX_L4_MASK) ==
+ RTE_MBUF_F_TX_UDP_CKSUM ||
+ (m->ol_flags & RTE_MBUF_F_TX_IP_CKSUM)) {
pi_data = hn_rndis_pktinfo_append(pkt, NDIS_TXCSUM_INFO_SIZE,
NDIS_PKTINFO_TYPE_CSUM);
*pi_data = 0;
@@ -1363,9 +1366,11 @@ static void hn_encap(struct rndis_packet_msg *pkt,
*pi_data |= NDIS_TXCSUM_INFO_IPCS;
}
- if (m->ol_flags & RTE_MBUF_F_TX_TCP_CKSUM)
+ if ((m->ol_flags & RTE_MBUF_F_TX_L4_MASK) ==
+ RTE_MBUF_F_TX_TCP_CKSUM)
*pi_data |= NDIS_TXCSUM_INFO_MKTCPCS(hlen);
- else if (m->ol_flags & RTE_MBUF_F_TX_UDP_CKSUM)
+ else if ((m->ol_flags & RTE_MBUF_F_TX_L4_MASK) ==
+ RTE_MBUF_F_TX_UDP_CKSUM)
*pi_data |= NDIS_TXCSUM_INFO_MKUDPCS(hlen);
}