From: Saeed Mahameed <saeed@kernel.org>
To: "David S. Miller" <davem@davemloft.net>,
Jakub Kicinski <kuba@kernel.org>
Cc: netdev@vger.kernel.org, Emeel Hakim <ehakim@nvidia.com>,
Saeed Mahameed <saeedm@nvidia.com>
Subject: [net 5/5] net/mlx5e: IPsec: Fix work queue entry ethernet segment checksum flags
Date: Wed, 20 Oct 2021 10:56:27 -0700 [thread overview]
Message-ID: <20211020175627.269138-6-saeed@kernel.org> (raw)
In-Reply-To: <20211020175627.269138-1-saeed@kernel.org>
From: Emeel Hakim <ehakim@nvidia.com>
Current Work Queue Entry (WQE) checksum (csum) flags in the ethernet
segment (eseg) in case of IPsec crypto offload datapath are not aligned
with PRM/HW expectations.
Currently the driver always sets the l3_inner_csum flag in case of IPsec
because of the wrong usage of skb->encapsulation as indicator for inner
IPsec header since skb->encapsulation is always ON for IPsec packets
since IPsec itself is an encapsulation protocol. The above forced a
failing attempts of calculating csum of non-existing segments (like in
the IP|ESP|TCP packet case which does not have an l3_inner) which led
to lots of packet drops hence the low throughput.
Fix by using xo->inner_ipproto as indicator for inner IPsec header
instead of skb->encapsulation in addition to setting the csum flags
as following:
* Tunnel Mode:
* Pkt: MAC IP ESP IP L4
* CSUM: l3_cs | l3_inner_cs | l4_inner_cs
*
* Transport Mode:
* Pkt: MAC IP ESP L4
* CSUM: l3_cs [ | l4_cs (checksum partial case)]
*
* Tunnel(VXLAN TCP/UDP) over Transport Mode
* Pkt: MAC IP ESP UDP VXLAN IP L4
* CSUM: l3_cs | l3_inner_cs | l4_inner_cs
Fixes: f1267798c980 ("net/mlx5: Fix checksum issue of VXLAN and IPsec crypto offload")
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
---
.../net/ethernet/mellanox/mlx5/core/en_tx.c | 20 ++++++++++---------
1 file changed, 11 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
index c63d78eda606..188994d091c5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
@@ -213,19 +213,18 @@ static inline void mlx5e_insert_vlan(void *start, struct sk_buff *skb, u16 ihs)
memcpy(&vhdr->h_vlan_encapsulated_proto, skb->data + cpy1_sz, cpy2_sz);
}
-/* If packet is not IP's CHECKSUM_PARTIAL (e.g. icmd packet),
- * need to set L3 checksum flag for IPsec
- */
static void
ipsec_txwqe_build_eseg_csum(struct mlx5e_txqsq *sq, struct sk_buff *skb,
struct mlx5_wqe_eth_seg *eseg)
{
+ struct xfrm_offload *xo = xfrm_offload(skb);
+
eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM;
- if (skb->encapsulation) {
- eseg->cs_flags |= MLX5_ETH_WQE_L3_INNER_CSUM;
+ if (xo->inner_ipproto) {
+ eseg->cs_flags |= MLX5_ETH_WQE_L4_INNER_CSUM | MLX5_ETH_WQE_L3_INNER_CSUM;
+ } else if (likely(skb->ip_summed == CHECKSUM_PARTIAL)) {
+ eseg->cs_flags |= MLX5_ETH_WQE_L4_CSUM;
sq->stats->csum_partial_inner++;
- } else {
- sq->stats->csum_partial++;
}
}
@@ -234,6 +233,11 @@ mlx5e_txwqe_build_eseg_csum(struct mlx5e_txqsq *sq, struct sk_buff *skb,
struct mlx5e_accel_tx_state *accel,
struct mlx5_wqe_eth_seg *eseg)
{
+ if (unlikely(mlx5e_ipsec_eseg_meta(eseg))) {
+ ipsec_txwqe_build_eseg_csum(sq, skb, eseg);
+ return;
+ }
+
if (likely(skb->ip_summed == CHECKSUM_PARTIAL)) {
eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM;
if (skb->encapsulation) {
@@ -249,8 +253,6 @@ mlx5e_txwqe_build_eseg_csum(struct mlx5e_txqsq *sq, struct sk_buff *skb,
eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM | MLX5_ETH_WQE_L4_CSUM;
sq->stats->csum_partial++;
#endif
- } else if (unlikely(mlx5e_ipsec_eseg_meta(eseg))) {
- ipsec_txwqe_build_eseg_csum(sq, skb, eseg);
} else
sq->stats->csum_none++;
}
--
2.31.1
prev parent reply other threads:[~2021-10-20 17:56 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-10-20 17:56 [pull request][net 0/5] mlx5 fixes 2021-10-20 Saeed Mahameed
2021-10-20 17:56 ` [net 1/5] net/mlx5: Lag, change multipath and bonding to be mutually exclusive Saeed Mahameed
2021-10-21 11:20 ` patchwork-bot+netdevbpf
2021-10-20 17:56 ` [net 2/5] net/mlx5: E-switch, Return correct error code on group creation failure Saeed Mahameed
2021-10-20 17:56 ` [net 3/5] net/mlx5e: Fix vlan data lost during suspend flow Saeed Mahameed
2021-10-20 17:56 ` [net 4/5] net/mlx5e: IPsec: Fix a misuse of the software parser's fields Saeed Mahameed
2021-10-20 17:56 ` Saeed Mahameed [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20211020175627.269138-6-saeed@kernel.org \
--to=saeed@kernel.org \
--cc=davem@davemloft.net \
--cc=ehakim@nvidia.com \
--cc=kuba@kernel.org \
--cc=netdev@vger.kernel.org \
--cc=saeedm@nvidia.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).