* [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09
@ 2020-05-09 8:28 Saeed Mahameed
2020-05-09 8:28 ` [net-next 01/13] net/mlx5e: Return bool from TLS and IPSEC offloads Saeed Mahameed
` (13 more replies)
0 siblings, 14 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba; +Cc: netdev, Saeed Mahameed
Hi Dave,
This series includes updates to mlx5 driver and a merge commit to the
earlier bonding series which was submitted by Maor to mlx5-next branch.
For more information please see tag log below.
Please pull and let me know if there is any problem.
Please note that the series starts with a merge of mlx5-next branch,
to resolve and avoid dependency with rdma tree.
A minor conflict between Maor's bonding series [1] and Eric's [2] is
already handled in this merge commit.
[1] bonding: Add support to get xmit slave
[2] bonding: report transmit status to callers
Thanks,
Saeed.
---
The following changes since commit 76cd622fe2c2b10c1f0a7311ca797feccacc329d:
Merge branch 'mlx5-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux (2020-05-09 01:05:30 -0700)
are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux.git tags/mlx5-updates-2020-05-09
for you to fetch changes up to 28bff09518e9ef942173e41e7521b93ea7be0cf0:
net/mlx5e: Enhance ICOSQ WQE info fields (2020-05-09 01:05:42 -0700)
----------------------------------------------------------------
mlx5-updates-2020-05-09
This series includes updates to mlx5 netdev driver and bonding updates
to support getting the next active tx slave.
1) merge commit with mlx5-next that includes bonding updates from Maor
Bonding: Add support to get xmit slave
2) Maxim makes some general code improvements to TX data path
3) Tariq makes some general code improvements to kTLS and mlx5 accel layer
in preparation for mlx5 TLS RX.
----------------------------------------------------------------
Maxim Mikityanskiy (7):
net/mlx5e: Return bool from TLS and IPSEC offloads
net/mlx5e: Unify checks of TLS offloads
net/mlx5e: Return void from mlx5e_sq_xmit and mlx5i_sq_xmit
net/mlx5e: Pass only eseg to IPSEC offload
net/mlx5e: Make TLS offload independent of wqe and pi
net/mlx5e: Update UDP fields of the SKB for GSO first
net/mlx5e: Split TX acceleration offloads into two phases
Tariq Toukan (6):
net/mlx5e: kTLS, Fill work queue edge separately in TX flow
net/mlx5e: kTLS, Do not fill edge for the DUMP WQEs in TX flow
net/mlx5e: Take TX WQE info structures out of general EN header
net/mlx5e: Use struct assignment for WQE info updates
net/mlx5: Accel, Remove unnecessary header include
net/mlx5e: Enhance ICOSQ WQE info fields
.../net/ethernet/mellanox/mlx5/core/accel/accel.h | 1 -
drivers/net/ethernet/mellanox/mlx5/core/en.h | 31 +---------
drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h | 29 ++++++++-
drivers/net/ethernet/mellanox/mlx5/core/en/xdp.h | 5 ++
.../mellanox/mlx5/core/en_accel/en_accel.h | 48 ++++++++++-----
.../mellanox/mlx5/core/en_accel/ipsec_rxtx.c | 15 +++--
.../mellanox/mlx5/core/en_accel/ipsec_rxtx.h | 6 +-
.../ethernet/mellanox/mlx5/core/en_accel/ktls.h | 8 +--
.../ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c | 66 +++++++--------------
.../mellanox/mlx5/core/en_accel/tls_rxtx.c | 69 +++++++++++-----------
.../mellanox/mlx5/core/en_accel/tls_rxtx.h | 13 ++--
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 16 ++---
drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 24 +++++---
drivers/net/ethernet/mellanox/mlx5/core/en_tx.c | 35 ++++++-----
drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c | 7 ++-
.../net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c | 4 +-
.../net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h | 5 +-
17 files changed, 197 insertions(+), 185 deletions(-)
Saeed Mahameed (1):
Merge branch 'mlx5-next' of git://git.kernel.org/.../mellanox/linux
drivers/net/bonding/bond_alb.c | 39 +++++++++++++------
drivers/net/bonding/bond_main.c | 252 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--------------------------------
drivers/net/ethernet/mellanox/mlx5/core/lag.c | 66 +++++++++++++++++++++-----------
include/linux/mlx5/driver.h | 2 +
include/linux/netdevice.h | 12 ++++++
include/net/bond_alb.h | 4 ++
include/net/bonding.h | 3 +-
net/core/dev.c | 22 +++++++++++
8 files changed, 301 insertions(+), 99 deletions(-)
^ permalink raw reply [flat|nested] 15+ messages in thread
* [net-next 01/13] net/mlx5e: Return bool from TLS and IPSEC offloads
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 02/13] net/mlx5e: Unify checks of TLS offloads Saeed Mahameed
` (12 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
TLS and IPSEC offloads currently return struct sk_buff *, but the value
is either NULL or the same skb that was passed as a parameter. Return
bool instead to provide stronger guarantees to the calling code (it
won't need to support handling a different SKB that could be potentially
returned before this change) and to simplify restructuring this code in
the following commits.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../mellanox/mlx5/core/en_accel/en_accel.h | 23 ++++-----
.../mellanox/mlx5/core/en_accel/ipsec_rxtx.c | 12 ++---
.../mellanox/mlx5/core/en_accel/ipsec_rxtx.h | 6 +--
.../mellanox/mlx5/core/en_accel/ktls.h | 7 ++-
.../mellanox/mlx5/core/en_accel/ktls_tx.c | 11 ++---
.../mellanox/mlx5/core/en_accel/tls_rxtx.c | 48 ++++++++-----------
.../mellanox/mlx5/core/en_accel/tls_rxtx.h | 8 ++--
.../net/ethernet/mellanox/mlx5/core/en_tx.c | 3 +-
8 files changed, 50 insertions(+), 68 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
index a6f65d4b2f36..6249998444c0 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
@@ -102,33 +102,30 @@ mlx5e_udp_gso_handle_tx_skb(struct sk_buff *skb)
udp_hdr(skb)->len = htons(payload_len);
}
-static inline struct sk_buff *
-mlx5e_accel_handle_tx(struct sk_buff *skb,
- struct mlx5e_txqsq *sq,
- struct net_device *dev,
- struct mlx5e_tx_wqe **wqe,
- u16 *pi)
+static inline bool mlx5e_accel_handle_tx(struct sk_buff *skb,
+ struct mlx5e_txqsq *sq,
+ struct net_device *dev,
+ struct mlx5e_tx_wqe **wqe,
+ u16 *pi)
{
#ifdef CONFIG_MLX5_EN_TLS
if (test_bit(MLX5E_SQ_STATE_TLS, &sq->state)) {
- skb = mlx5e_tls_handle_tx_skb(dev, sq, skb, wqe, pi);
- if (unlikely(!skb))
- return NULL;
+ if (unlikely(!mlx5e_tls_handle_tx_skb(dev, sq, skb, wqe, pi)))
+ return false;
}
#endif
#ifdef CONFIG_MLX5_EN_IPSEC
if (test_bit(MLX5E_SQ_STATE_IPSEC, &sq->state)) {
- skb = mlx5e_ipsec_handle_tx_skb(dev, *wqe, skb);
- if (unlikely(!skb))
- return NULL;
+ if (unlikely(!mlx5e_ipsec_handle_tx_skb(dev, *wqe, skb)))
+ return false;
}
#endif
if (skb_is_gso(skb) && skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4)
mlx5e_udp_gso_handle_tx_skb(skb);
- return skb;
+ return true;
}
#endif /* __MLX5E_EN_ACCEL_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
index 0dd17514caae..f60eb6a4b57c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
@@ -233,9 +233,9 @@ static void mlx5e_ipsec_set_metadata(struct sk_buff *skb,
ntohs(mdata->content.tx.seq));
}
-struct sk_buff *mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
- struct mlx5e_tx_wqe *wqe,
- struct sk_buff *skb)
+bool mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
+ struct mlx5e_tx_wqe *wqe,
+ struct sk_buff *skb)
{
struct mlx5e_priv *priv = netdev_priv(netdev);
struct xfrm_offload *xo = xfrm_offload(skb);
@@ -245,7 +245,7 @@ struct sk_buff *mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
struct sec_path *sp;
if (!xo)
- return skb;
+ return true;
sp = skb_sec_path(skb);
if (unlikely(sp->len != 1)) {
@@ -281,11 +281,11 @@ struct sk_buff *mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
sa_entry->set_iv_op(skb, x, xo);
mlx5e_ipsec_set_metadata(skb, mdata, xo);
- return skb;
+ return true;
drop:
kfree_skb(skb);
- return NULL;
+ return false;
}
static inline struct xfrm_state *
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
index db84500b024f..64e948cc3dc5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
@@ -52,9 +52,9 @@ void mlx5e_ipsec_set_iv_esn(struct sk_buff *skb, struct xfrm_state *x,
struct xfrm_offload *xo);
void mlx5e_ipsec_set_iv(struct sk_buff *skb, struct xfrm_state *x,
struct xfrm_offload *xo);
-struct sk_buff *mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
- struct mlx5e_tx_wqe *wqe,
- struct sk_buff *skb);
+bool mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
+ struct mlx5e_tx_wqe *wqe,
+ struct sk_buff *skb);
#endif /* CONFIG_MLX5_EN_IPSEC */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
index 9daaec244385..742aca8782d6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
@@ -95,10 +95,9 @@ mlx5e_get_ktls_tx_priv_ctx(struct tls_context *tls_ctx)
void mlx5e_ktls_build_netdev(struct mlx5e_priv *priv);
void mlx5e_ktls_tx_offload_set_pending(struct mlx5e_ktls_offload_context_tx *priv_tx);
-struct sk_buff *mlx5e_ktls_handle_tx_skb(struct net_device *netdev,
- struct mlx5e_txqsq *sq,
- struct sk_buff *skb,
- struct mlx5e_tx_wqe **wqe, u16 *pi);
+bool mlx5e_ktls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
+ struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
+ u16 *pi);
void mlx5e_ktls_tx_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
struct mlx5e_tx_wqe_info *wi,
u32 *dma_fifo_cc);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
index ba973937f0b5..8fcd14803558 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
@@ -413,10 +413,9 @@ mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
return MLX5E_KTLS_SYNC_FAIL;
}
-struct sk_buff *mlx5e_ktls_handle_tx_skb(struct net_device *netdev,
- struct mlx5e_txqsq *sq,
- struct sk_buff *skb,
- struct mlx5e_tx_wqe **wqe, u16 *pi)
+bool mlx5e_ktls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
+ struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
+ u16 *pi)
{
struct mlx5e_ktls_offload_context_tx *priv_tx;
struct mlx5e_sq_stats *stats = sq->stats;
@@ -474,9 +473,9 @@ struct sk_buff *mlx5e_ktls_handle_tx_skb(struct net_device *netdev,
stats->tls_encrypted_bytes += datalen;
out:
- return skb;
+ return true;
err_out:
dev_kfree_skb_any(skb);
- return NULL;
+ return false;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
index 1d7ddeb7a46b..e8f2c214a8de 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
@@ -184,12 +184,10 @@ static void mlx5e_tls_complete_sync_skb(struct sk_buff *skb,
nskb->queue_mapping = skb->queue_mapping;
}
-static struct sk_buff *
-mlx5e_tls_handle_ooo(struct mlx5e_tls_offload_context_tx *context,
- struct mlx5e_txqsq *sq, struct sk_buff *skb,
- struct mlx5e_tx_wqe **wqe,
- u16 *pi,
- struct mlx5e_tls *tls)
+static bool mlx5e_tls_handle_ooo(struct mlx5e_tls_offload_context_tx *context,
+ struct mlx5e_txqsq *sq, struct sk_buff *skb,
+ struct mlx5e_tx_wqe **wqe, u16 *pi,
+ struct mlx5e_tls *tls)
{
u32 tcp_seq = ntohl(tcp_hdr(skb)->seq);
struct sync_info info;
@@ -217,7 +215,7 @@ mlx5e_tls_handle_ooo(struct mlx5e_tls_offload_context_tx *context,
if (likely(payload <= -info.sync_len))
/* SKB payload doesn't require offload
*/
- return skb;
+ return true;
atomic64_inc(&tls->sw_stats.tx_tls_drop_bypass_required);
goto err_out;
@@ -250,18 +248,16 @@ mlx5e_tls_handle_ooo(struct mlx5e_tls_offload_context_tx *context,
mlx5e_sq_xmit(sq, nskb, *wqe, *pi, true);
*pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
*wqe = MLX5E_TX_FETCH_WQE(sq, *pi);
- return skb;
+ return true;
err_out:
dev_kfree_skb_any(skb);
- return NULL;
+ return false;
}
-struct sk_buff *mlx5e_tls_handle_tx_skb(struct net_device *netdev,
- struct mlx5e_txqsq *sq,
- struct sk_buff *skb,
- struct mlx5e_tx_wqe **wqe,
- u16 *pi)
+bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
+ struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
+ u16 *pi)
{
struct mlx5e_priv *priv = netdev_priv(netdev);
struct mlx5e_tls_offload_context_tx *context;
@@ -270,41 +266,35 @@ struct sk_buff *mlx5e_tls_handle_tx_skb(struct net_device *netdev,
int datalen;
u32 skb_seq;
- if (MLX5_CAP_GEN(sq->channel->mdev, tls_tx)) {
- skb = mlx5e_ktls_handle_tx_skb(netdev, sq, skb, wqe, pi);
- goto out;
- }
+ if (MLX5_CAP_GEN(sq->channel->mdev, tls_tx))
+ return mlx5e_ktls_handle_tx_skb(netdev, sq, skb, wqe, pi);
if (!skb->sk || !tls_is_sk_tx_device_offloaded(skb->sk))
- goto out;
+ return true;
datalen = skb->len - (skb_transport_offset(skb) + tcp_hdrlen(skb));
if (!datalen)
- goto out;
+ return true;
tls_ctx = tls_get_ctx(skb->sk);
if (unlikely(tls_ctx->netdev != netdev))
- goto out;
+ return true;
skb_seq = ntohl(tcp_hdr(skb)->seq);
context = mlx5e_get_tls_tx_context(tls_ctx);
expected_seq = context->expected_seq;
- if (unlikely(expected_seq != skb_seq)) {
- skb = mlx5e_tls_handle_ooo(context, sq, skb, wqe, pi, priv->tls);
- goto out;
- }
+ if (unlikely(expected_seq != skb_seq))
+ return mlx5e_tls_handle_ooo(context, sq, skb, wqe, pi, priv->tls);
if (unlikely(mlx5e_tls_add_metadata(skb, context->swid))) {
atomic64_inc(&priv->tls->sw_stats.tx_tls_drop_metadata);
dev_kfree_skb_any(skb);
- skb = NULL;
- goto out;
+ return false;
}
context->expected_seq = skb_seq + datalen;
-out:
- return skb;
+ return true;
}
static int tls_update_resync_sn(struct net_device *netdev,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h
index 90bc1f2384c8..890d452bf1ae 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h
@@ -40,11 +40,9 @@
#include "en.h"
#include "en/txrx.h"
-struct sk_buff *mlx5e_tls_handle_tx_skb(struct net_device *netdev,
- struct mlx5e_txqsq *sq,
- struct sk_buff *skb,
- struct mlx5e_tx_wqe **wqe,
- u16 *pi);
+bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
+ struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
+ u16 *pi);
void mlx5e_tls_handle_rx_skb(struct net_device *netdev, struct sk_buff *skb,
u32 *cqe_bcnt);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
index 583e1b201b75..7a6ed72ae00a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
@@ -394,8 +394,7 @@ netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
wqe = MLX5E_TX_FETCH_WQE(sq, pi);
/* might send skbs and update wqe and pi */
- skb = mlx5e_accel_handle_tx(skb, sq, dev, &wqe, &pi);
- if (unlikely(!skb))
+ if (unlikely(!mlx5e_accel_handle_tx(skb, sq, dev, &wqe, &pi)))
return NETDEV_TX_OK;
return mlx5e_sq_xmit(sq, skb, wqe, pi, netdev_xmit_more());
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 02/13] net/mlx5e: Unify checks of TLS offloads
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
2020-05-09 8:28 ` [net-next 01/13] net/mlx5e: Return bool from TLS and IPSEC offloads Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 03/13] net/mlx5e: Return void from mlx5e_sq_xmit and mlx5i_sq_xmit Saeed Mahameed
` (11 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
Both INNOVA and ConnectX TLS offloads perform the same checks in the
beginning. Unify them to reduce repeating code. Do WARN_ON_ONCE on
netdev mismatch and finish with an error in both offloads, not only in
the ConnectX one.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../ethernet/mellanox/mlx5/core/en_accel/ktls.h | 4 ++--
.../mellanox/mlx5/core/en_accel/ktls_tx.c | 17 ++---------------
.../mellanox/mlx5/core/en_accel/tls_rxtx.c | 14 +++++++++-----
3 files changed, 13 insertions(+), 22 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
index 742aca8782d6..81f8b7467569 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
@@ -95,9 +95,9 @@ mlx5e_get_ktls_tx_priv_ctx(struct tls_context *tls_ctx)
void mlx5e_ktls_build_netdev(struct mlx5e_priv *priv);
void mlx5e_ktls_tx_offload_set_pending(struct mlx5e_ktls_offload_context_tx *priv_tx);
-bool mlx5e_ktls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
+bool mlx5e_ktls_handle_tx_skb(struct tls_context *tls_ctx, struct mlx5e_txqsq *sq,
struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
- u16 *pi);
+ u16 *pi, int datalen);
void mlx5e_ktls_tx_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
struct mlx5e_tx_wqe_info *wi,
u32 *dma_fifo_cc);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
index 8fcd14803558..c61604f3722c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
@@ -413,28 +413,15 @@ mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
return MLX5E_KTLS_SYNC_FAIL;
}
-bool mlx5e_ktls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
+bool mlx5e_ktls_handle_tx_skb(struct tls_context *tls_ctx, struct mlx5e_txqsq *sq,
struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
- u16 *pi)
+ u16 *pi, int datalen)
{
struct mlx5e_ktls_offload_context_tx *priv_tx;
struct mlx5e_sq_stats *stats = sq->stats;
struct mlx5_wqe_ctrl_seg *cseg;
- struct tls_context *tls_ctx;
- int datalen;
u32 seq;
- if (!skb->sk || !tls_is_sk_tx_device_offloaded(skb->sk))
- goto out;
-
- datalen = skb->len - (skb_transport_offset(skb) + tcp_hdrlen(skb));
- if (!datalen)
- goto out;
-
- tls_ctx = tls_get_ctx(skb->sk);
- if (WARN_ON_ONCE(tls_ctx->netdev != netdev))
- goto err_out;
-
priv_tx = mlx5e_get_ktls_tx_priv_ctx(tls_ctx);
if (unlikely(mlx5e_ktls_tx_offload_test_and_clear_pending(priv_tx))) {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
index e8f2c214a8de..26c59cfbec9b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
@@ -266,9 +266,6 @@ bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
int datalen;
u32 skb_seq;
- if (MLX5_CAP_GEN(sq->channel->mdev, tls_tx))
- return mlx5e_ktls_handle_tx_skb(netdev, sq, skb, wqe, pi);
-
if (!skb->sk || !tls_is_sk_tx_device_offloaded(skb->sk))
return true;
@@ -277,8 +274,11 @@ bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
return true;
tls_ctx = tls_get_ctx(skb->sk);
- if (unlikely(tls_ctx->netdev != netdev))
- return true;
+ if (WARN_ON_ONCE(tls_ctx->netdev != netdev))
+ goto err_out;
+
+ if (MLX5_CAP_GEN(sq->channel->mdev, tls_tx))
+ return mlx5e_ktls_handle_tx_skb(tls_ctx, sq, skb, wqe, pi, datalen);
skb_seq = ntohl(tcp_hdr(skb)->seq);
context = mlx5e_get_tls_tx_context(tls_ctx);
@@ -295,6 +295,10 @@ bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
context->expected_seq = skb_seq + datalen;
return true;
+
+err_out:
+ dev_kfree_skb_any(skb);
+ return false;
}
static int tls_update_resync_sn(struct net_device *netdev,
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 03/13] net/mlx5e: Return void from mlx5e_sq_xmit and mlx5i_sq_xmit
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
2020-05-09 8:28 ` [net-next 01/13] net/mlx5e: Return bool from TLS and IPSEC offloads Saeed Mahameed
2020-05-09 8:28 ` [net-next 02/13] net/mlx5e: Unify checks of TLS offloads Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 04/13] net/mlx5e: Pass only eseg to IPSEC offload Saeed Mahameed
` (10 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
mlx5e_sq_xmit and mlx5i_sq_xmit always return NETDEV_TX_OK. Drop the
return value to simplify the code.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 4 ++--
.../net/ethernet/mellanox/mlx5/core/en_tx.c | 24 +++++++++----------
.../ethernet/mellanox/mlx5/core/ipoib/ipoib.c | 4 +++-
.../ethernet/mellanox/mlx5/core/ipoib/ipoib.h | 5 ++--
4 files changed, 18 insertions(+), 19 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 0864b76ca2c0..da7fe6aafeed 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -919,8 +919,8 @@ void mlx5e_build_ptys2ethtool_map(void);
u16 mlx5e_select_queue(struct net_device *dev, struct sk_buff *skb,
struct net_device *sb_dev);
netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev);
-netdev_tx_t mlx5e_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
- struct mlx5e_tx_wqe *wqe, u16 pi, bool xmit_more);
+void mlx5e_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+ struct mlx5e_tx_wqe *wqe, u16 pi, bool xmit_more);
void mlx5e_trigger_irq(struct mlx5e_icosq *sq);
void mlx5e_completion_event(struct mlx5_core_cq *mcq, struct mlx5_eqe *eqe);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
index 7a6ed72ae00a..bb6d3774eafb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
@@ -265,8 +265,8 @@ mlx5e_txwqe_complete(struct mlx5e_txqsq *sq, struct sk_buff *skb,
mlx5e_notify_hw(wq, sq->pc, sq->uar_map, cseg);
}
-netdev_tx_t mlx5e_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
- struct mlx5e_tx_wqe *wqe, u16 pi, bool xmit_more)
+void mlx5e_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+ struct mlx5e_tx_wqe *wqe, u16 pi, bool xmit_more)
{
struct mlx5_wq_cyc *wq = &sq->wq;
struct mlx5_wqe_ctrl_seg *cseg;
@@ -373,13 +373,11 @@ netdev_tx_t mlx5e_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
mlx5e_txwqe_complete(sq, skb, opcode, ds_cnt, num_wqebbs, num_bytes,
num_dma, wi, cseg, xmit_more);
- return NETDEV_TX_OK;
+ return;
err_drop:
stats->dropped++;
dev_kfree_skb_any(skb);
-
- return NETDEV_TX_OK;
}
netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
@@ -395,9 +393,12 @@ netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
/* might send skbs and update wqe and pi */
if (unlikely(!mlx5e_accel_handle_tx(skb, sq, dev, &wqe, &pi)))
- return NETDEV_TX_OK;
+ goto out;
+
+ mlx5e_sq_xmit(sq, skb, wqe, pi, netdev_xmit_more());
- return mlx5e_sq_xmit(sq, skb, wqe, pi, netdev_xmit_more());
+out:
+ return NETDEV_TX_OK;
}
bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq, int napi_budget)
@@ -567,9 +568,8 @@ mlx5i_txwqe_build_datagram(struct mlx5_av *av, u32 dqpn, u32 dqkey,
dseg->av.key.qkey.qkey = cpu_to_be32(dqkey);
}
-netdev_tx_t mlx5i_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
- struct mlx5_av *av, u32 dqpn, u32 dqkey,
- bool xmit_more)
+void mlx5i_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+ struct mlx5_av *av, u32 dqpn, u32 dqkey, bool xmit_more)
{
struct mlx5i_tx_wqe *wqe;
@@ -647,12 +647,10 @@ netdev_tx_t mlx5i_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
mlx5e_txwqe_complete(sq, skb, opcode, ds_cnt, num_wqebbs, num_bytes,
num_dma, wi, cseg, xmit_more);
- return NETDEV_TX_OK;
+ return;
err_drop:
stats->dropped++;
dev_kfree_skb_any(skb);
-
- return NETDEV_TX_OK;
}
#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
index 068578be00f1..035bd21e5d4e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
@@ -663,7 +663,9 @@ static int mlx5i_xmit(struct net_device *dev, struct sk_buff *skb,
struct mlx5_ib_ah *mah = to_mah(address);
struct mlx5i_priv *ipriv = epriv->ppriv;
- return mlx5i_sq_xmit(sq, skb, &mah->av, dqpn, ipriv->qkey, netdev_xmit_more());
+ mlx5i_sq_xmit(sq, skb, &mah->av, dqpn, ipriv->qkey, netdev_xmit_more());
+
+ return NETDEV_TX_OK;
}
static void mlx5i_set_pkey_index(struct net_device *netdev, int id)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h
index 7844ab5d0ce7..c4aa47018c0e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.h
@@ -113,9 +113,8 @@ struct mlx5i_tx_wqe {
#define MLX5I_SQ_FETCH_WQE(sq, pi) \
((struct mlx5i_tx_wqe *)mlx5e_fetch_wqe(&(sq)->wq, pi, sizeof(struct mlx5i_tx_wqe)))
-netdev_tx_t mlx5i_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
- struct mlx5_av *av, u32 dqpn, u32 dqkey,
- bool xmit_more);
+void mlx5i_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
+ struct mlx5_av *av, u32 dqpn, u32 dqkey, bool xmit_more);
void mlx5i_handle_rx_cqe(struct mlx5e_rq *rq, struct mlx5_cqe64 *cqe);
void mlx5i_get_stats(struct net_device *dev, struct rtnl_link_stats64 *stats);
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 04/13] net/mlx5e: Pass only eseg to IPSEC offload
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (2 preceding siblings ...)
2020-05-09 8:28 ` [net-next 03/13] net/mlx5e: Return void from mlx5e_sq_xmit and mlx5i_sq_xmit Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 05/13] net/mlx5e: Make TLS offload independent of wqe and pi Saeed Mahameed
` (9 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
IPSEC offload needs to modify the eseg of the WQE that is being filled,
but it receives a pointer to the whole WQE. To make the contract
stricter, pass only the pointer to the eseg of that WQE. This commit is
preparation for the following refactoring of offloads in the TX path and
for the MPWQE support.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h | 2 +-
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c | 4 ++--
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h | 2 +-
3 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
index 6249998444c0..c658c8556863 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
@@ -117,7 +117,7 @@ static inline bool mlx5e_accel_handle_tx(struct sk_buff *skb,
#ifdef CONFIG_MLX5_EN_IPSEC
if (test_bit(MLX5E_SQ_STATE_IPSEC, &sq->state)) {
- if (unlikely(!mlx5e_ipsec_handle_tx_skb(dev, *wqe, skb)))
+ if (unlikely(!mlx5e_ipsec_handle_tx_skb(dev, &(*wqe)->eth, skb)))
return false;
}
#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
index f60eb6a4b57c..0e1ac3e68c72 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
@@ -234,7 +234,7 @@ static void mlx5e_ipsec_set_metadata(struct sk_buff *skb,
}
bool mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
- struct mlx5e_tx_wqe *wqe,
+ struct mlx5_wqe_eth_seg *eseg,
struct sk_buff *skb)
{
struct mlx5e_priv *priv = netdev_priv(netdev);
@@ -276,7 +276,7 @@ bool mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
atomic64_inc(&priv->ipsec->sw_stats.ipsec_tx_drop_metadata);
goto drop;
}
- mlx5e_ipsec_set_swp(skb, &wqe->eth, x->props.mode, xo);
+ mlx5e_ipsec_set_swp(skb, eseg, x->props.mode, xo);
sa_entry = (struct mlx5e_ipsec_sa_entry *)x->xso.offload_handle;
sa_entry->set_iv_op(skb, x, xo);
mlx5e_ipsec_set_metadata(skb, mdata, xo);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
index 64e948cc3dc5..bd6f32aee8d6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
@@ -53,7 +53,7 @@ void mlx5e_ipsec_set_iv_esn(struct sk_buff *skb, struct xfrm_state *x,
void mlx5e_ipsec_set_iv(struct sk_buff *skb, struct xfrm_state *x,
struct xfrm_offload *xo);
bool mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
- struct mlx5e_tx_wqe *wqe,
+ struct mlx5_wqe_eth_seg *eseg,
struct sk_buff *skb);
#endif /* CONFIG_MLX5_EN_IPSEC */
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 05/13] net/mlx5e: Make TLS offload independent of wqe and pi
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (3 preceding siblings ...)
2020-05-09 8:28 ` [net-next 04/13] net/mlx5e: Pass only eseg to IPSEC offload Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 06/13] net/mlx5e: Update UDP fields of the SKB for GSO first Saeed Mahameed
` (8 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
TLS offload may write a 32-bit field (tisn) to the cseg of the WQE. To
do that, it receives pi and wqe pointers. As TLS offload may also send
additional WQEs, it has to update pi and wqe, and in many cases it even
doesn't use pi calculated before and wqe zeroed before and does it
itself. Also, mlx5e_sq_xmit has to copy the whole cseg if it goes to the
mlx5e_fill_sq_frag_edge flow. This all is not efficient.
It's more efficient to do the following:
1. Just return tisn from TLS offload and make the caller fill it in a
more appropriate place.
2. Calculate pi and clear wqe after calling TLS offload.
3. If TLS offload has to send WQEs, calculate pi and clear wqe just
before that. It's already done in all places anyway, so this commit
allows to remove some redundant memsets and calls.
Copying of cseg will be eliminated in one of the following commits, and
all other stuff is done here.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../mellanox/mlx5/core/en_accel/en_accel.h | 10 +++++++++-
.../ethernet/mellanox/mlx5/core/en_accel/ktls.h | 3 +--
.../mellanox/mlx5/core/en_accel/ktls_tx.c | 13 +++----------
.../mellanox/mlx5/core/en_accel/tls_rxtx.c | 17 +++++++++--------
.../mellanox/mlx5/core/en_accel/tls_rxtx.h | 3 +--
5 files changed, 23 insertions(+), 23 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
index c658c8556863..66bfab021d6b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
@@ -109,10 +109,18 @@ static inline bool mlx5e_accel_handle_tx(struct sk_buff *skb,
u16 *pi)
{
#ifdef CONFIG_MLX5_EN_TLS
+ u32 tls_tisn = 0;
+
if (test_bit(MLX5E_SQ_STATE_TLS, &sq->state)) {
- if (unlikely(!mlx5e_tls_handle_tx_skb(dev, sq, skb, wqe, pi)))
+ /* May send SKBs and WQEs. */
+ if (unlikely(!mlx5e_tls_handle_tx_skb(dev, sq, skb, &tls_tisn)))
return false;
}
+
+ *pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
+ *wqe = MLX5E_TX_FETCH_WQE(sq, *pi);
+
+ (*wqe)->ctrl.tisn = cpu_to_be32(tls_tisn << 8);
#endif
#ifdef CONFIG_MLX5_EN_IPSEC
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
index 81f8b7467569..7d9d9420f19d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
@@ -96,8 +96,7 @@ void mlx5e_ktls_build_netdev(struct mlx5e_priv *priv);
void mlx5e_ktls_tx_offload_set_pending(struct mlx5e_ktls_offload_context_tx *priv_tx);
bool mlx5e_ktls_handle_tx_skb(struct tls_context *tls_ctx, struct mlx5e_txqsq *sq,
- struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
- u16 *pi, int datalen);
+ struct sk_buff *skb, u32 *tisn, int datalen);
void mlx5e_ktls_tx_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
struct mlx5e_tx_wqe_info *wi,
u32 *dma_fifo_cc);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
index c61604f3722c..b49d7c1e49dc 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
@@ -414,20 +414,16 @@ mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
}
bool mlx5e_ktls_handle_tx_skb(struct tls_context *tls_ctx, struct mlx5e_txqsq *sq,
- struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
- u16 *pi, int datalen)
+ struct sk_buff *skb, u32 *tisn, int datalen)
{
struct mlx5e_ktls_offload_context_tx *priv_tx;
struct mlx5e_sq_stats *stats = sq->stats;
- struct mlx5_wqe_ctrl_seg *cseg;
u32 seq;
priv_tx = mlx5e_get_ktls_tx_priv_ctx(tls_ctx);
if (unlikely(mlx5e_ktls_tx_offload_test_and_clear_pending(priv_tx))) {
mlx5e_ktls_tx_post_param_wqes(sq, priv_tx, false, false);
- *pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
- *wqe = MLX5E_TX_FETCH_WQE(sq, *pi);
stats->tls_ctx++;
}
@@ -438,23 +434,20 @@ bool mlx5e_ktls_handle_tx_skb(struct tls_context *tls_ctx, struct mlx5e_txqsq *s
switch (ret) {
case MLX5E_KTLS_SYNC_DONE:
- *pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
- *wqe = MLX5E_TX_FETCH_WQE(sq, *pi);
break;
case MLX5E_KTLS_SYNC_SKIP_NO_DATA:
if (likely(!skb->decrypted))
goto out;
WARN_ON_ONCE(1);
/* fall-through */
- default: /* MLX5E_KTLS_SYNC_FAIL */
+ case MLX5E_KTLS_SYNC_FAIL:
goto err_out;
}
}
priv_tx->expected_seq = seq + datalen;
- cseg = &(*wqe)->ctrl;
- cseg->tisn = cpu_to_be32(priv_tx->tisn << 8);
+ *tisn = priv_tx->tisn;
stats->tls_encrypted_packets += skb_is_gso(skb) ? skb_shinfo(skb)->gso_segs : 1;
stats->tls_encrypted_bytes += datalen;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
index 26c59cfbec9b..8e6b0b0ce2e4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
@@ -186,14 +186,15 @@ static void mlx5e_tls_complete_sync_skb(struct sk_buff *skb,
static bool mlx5e_tls_handle_ooo(struct mlx5e_tls_offload_context_tx *context,
struct mlx5e_txqsq *sq, struct sk_buff *skb,
- struct mlx5e_tx_wqe **wqe, u16 *pi,
struct mlx5e_tls *tls)
{
u32 tcp_seq = ntohl(tcp_hdr(skb)->seq);
+ struct mlx5e_tx_wqe *wqe;
struct sync_info info;
struct sk_buff *nskb;
int linear_len = 0;
int headln;
+ u16 pi;
int i;
sq->stats->tls_ooo++;
@@ -245,9 +246,10 @@ static bool mlx5e_tls_handle_ooo(struct mlx5e_tls_offload_context_tx *context,
sq->stats->tls_resync_bytes += nskb->len;
mlx5e_tls_complete_sync_skb(skb, nskb, tcp_seq, headln,
cpu_to_be64(info.rcd_sn));
- mlx5e_sq_xmit(sq, nskb, *wqe, *pi, true);
- *pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
- *wqe = MLX5E_TX_FETCH_WQE(sq, *pi);
+ pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
+ wqe = MLX5E_TX_FETCH_WQE(sq, pi);
+ mlx5e_sq_xmit(sq, nskb, wqe, pi, true);
+
return true;
err_out:
@@ -256,8 +258,7 @@ static bool mlx5e_tls_handle_ooo(struct mlx5e_tls_offload_context_tx *context,
}
bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
- struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
- u16 *pi)
+ struct sk_buff *skb, u32 *tisn)
{
struct mlx5e_priv *priv = netdev_priv(netdev);
struct mlx5e_tls_offload_context_tx *context;
@@ -278,14 +279,14 @@ bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
goto err_out;
if (MLX5_CAP_GEN(sq->channel->mdev, tls_tx))
- return mlx5e_ktls_handle_tx_skb(tls_ctx, sq, skb, wqe, pi, datalen);
+ return mlx5e_ktls_handle_tx_skb(tls_ctx, sq, skb, tisn, datalen);
skb_seq = ntohl(tcp_hdr(skb)->seq);
context = mlx5e_get_tls_tx_context(tls_ctx);
expected_seq = context->expected_seq;
if (unlikely(expected_seq != skb_seq))
- return mlx5e_tls_handle_ooo(context, sq, skb, wqe, pi, priv->tls);
+ return mlx5e_tls_handle_ooo(context, sq, skb, priv->tls);
if (unlikely(mlx5e_tls_add_metadata(skb, context->swid))) {
atomic64_inc(&priv->tls->sw_stats.tx_tls_drop_metadata);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h
index 890d452bf1ae..3630ed8b1206 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h
@@ -41,8 +41,7 @@
#include "en/txrx.h"
bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
- struct sk_buff *skb, struct mlx5e_tx_wqe **wqe,
- u16 *pi);
+ struct sk_buff *skb, u32 *tisn);
void mlx5e_tls_handle_rx_skb(struct net_device *netdev, struct sk_buff *skb,
u32 *cqe_bcnt);
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 06/13] net/mlx5e: Update UDP fields of the SKB for GSO first
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (4 preceding siblings ...)
2020-05-09 8:28 ` [net-next 05/13] net/mlx5e: Make TLS offload independent of wqe and pi Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 07/13] net/mlx5e: Split TX acceleration offloads into two phases Saeed Mahameed
` (7 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Raed Salem, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
mlx5e_udp_gso_handle_tx_skb updates the length field in the UDP header
in case of GSO. It doesn't interfere with other offloads, so do it first
to simplify further restructuring of the code. This way we'll make all
independent modifications to the SKB before starting to work with WQEs.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Raed Salem <raeds@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
index 66bfab021d6b..d286fb09955c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
@@ -110,7 +110,12 @@ static inline bool mlx5e_accel_handle_tx(struct sk_buff *skb,
{
#ifdef CONFIG_MLX5_EN_TLS
u32 tls_tisn = 0;
+#endif
+
+ if (skb_is_gso(skb) && skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4)
+ mlx5e_udp_gso_handle_tx_skb(skb);
+#ifdef CONFIG_MLX5_EN_TLS
if (test_bit(MLX5E_SQ_STATE_TLS, &sq->state)) {
/* May send SKBs and WQEs. */
if (unlikely(!mlx5e_tls_handle_tx_skb(dev, sq, skb, &tls_tisn)))
@@ -130,9 +135,6 @@ static inline bool mlx5e_accel_handle_tx(struct sk_buff *skb,
}
#endif
- if (skb_is_gso(skb) && skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4)
- mlx5e_udp_gso_handle_tx_skb(skb);
-
return true;
}
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 07/13] net/mlx5e: Split TX acceleration offloads into two phases
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (5 preceding siblings ...)
2020-05-09 8:28 ` [net-next 06/13] net/mlx5e: Update UDP fields of the SKB for GSO first Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 08/13] net/mlx5e: kTLS, Fill work queue edge separately in TX flow Saeed Mahameed
` (6 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
After previous modifications, the offloads are no longer called one by
one, the pi is calculated and the wqe is cleared on between of TLS and
IPSEC offloads, which doesn't quite fit mlx5e_accel_handle_tx's purpose.
This patch splits mlx5e_accel_handle_tx into two functions that
correspond to two logical phases of running offloads:
1. Before fetching a WQE. Here runs the code that can post WQEs on its
own, before the main WQE is fetched. It's the main part of TLS offload.
2. After fetching a WQE. Here runs the code that updates the WQE's
fields, but can't post other WQEs any more. It's a minor part of TLS
offload that sets the tisn field in the cseg, and eseg-based offloads
(currently IPSEC, and later patches will move GENEVE and checksum
offloads there, too).
It allows to make mlx5e_xmit take care of all actions needed to transmit
a packet in the right order, improve the structure of the code and
reduce unnecessary operations. The structure will be further improved in
the following patches (all eseg-based offloads will be moved to a single
place, and reserving space for the main WQE will happen between phase 1
and phase 2 of offloads to eliminate unneeded data movements).
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../mellanox/mlx5/core/en_accel/en_accel.h | 33 ++++++++++++-------
.../mellanox/mlx5/core/en_accel/ipsec_rxtx.c | 3 +-
.../mellanox/mlx5/core/en_accel/ipsec_rxtx.h | 2 +-
.../mellanox/mlx5/core/en_accel/ktls.h | 4 ++-
.../mellanox/mlx5/core/en_accel/ktls_tx.c | 5 +--
.../mellanox/mlx5/core/en_accel/tls_rxtx.c | 10 ++++--
.../mellanox/mlx5/core/en_accel/tls_rxtx.h | 8 ++++-
.../net/ethernet/mellanox/mlx5/core/en_tx.c | 10 ++++--
8 files changed, 52 insertions(+), 23 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
index d286fb09955c..fac145dcf2ce 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/en_accel.h
@@ -102,35 +102,44 @@ mlx5e_udp_gso_handle_tx_skb(struct sk_buff *skb)
udp_hdr(skb)->len = htons(payload_len);
}
-static inline bool mlx5e_accel_handle_tx(struct sk_buff *skb,
- struct mlx5e_txqsq *sq,
- struct net_device *dev,
- struct mlx5e_tx_wqe **wqe,
- u16 *pi)
-{
+struct mlx5e_accel_tx_state {
#ifdef CONFIG_MLX5_EN_TLS
- u32 tls_tisn = 0;
+ struct mlx5e_accel_tx_tls_state tls;
#endif
+};
+static inline bool mlx5e_accel_tx_begin(struct net_device *dev,
+ struct mlx5e_txqsq *sq,
+ struct sk_buff *skb,
+ struct mlx5e_accel_tx_state *state)
+{
if (skb_is_gso(skb) && skb_shinfo(skb)->gso_type & SKB_GSO_UDP_L4)
mlx5e_udp_gso_handle_tx_skb(skb);
#ifdef CONFIG_MLX5_EN_TLS
if (test_bit(MLX5E_SQ_STATE_TLS, &sq->state)) {
/* May send SKBs and WQEs. */
- if (unlikely(!mlx5e_tls_handle_tx_skb(dev, sq, skb, &tls_tisn)))
+ if (unlikely(!mlx5e_tls_handle_tx_skb(dev, sq, skb, &state->tls)))
return false;
}
+#endif
- *pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
- *wqe = MLX5E_TX_FETCH_WQE(sq, *pi);
+ return true;
+}
- (*wqe)->ctrl.tisn = cpu_to_be32(tls_tisn << 8);
+static inline bool mlx5e_accel_tx_finish(struct mlx5e_priv *priv,
+ struct mlx5e_txqsq *sq,
+ struct sk_buff *skb,
+ struct mlx5e_tx_wqe *wqe,
+ struct mlx5e_accel_tx_state *state)
+{
+#ifdef CONFIG_MLX5_EN_TLS
+ mlx5e_tls_handle_tx_wqe(sq, &wqe->ctrl, &state->tls);
#endif
#ifdef CONFIG_MLX5_EN_IPSEC
if (test_bit(MLX5E_SQ_STATE_IPSEC, &sq->state)) {
- if (unlikely(!mlx5e_ipsec_handle_tx_skb(dev, &(*wqe)->eth, skb)))
+ if (unlikely(!mlx5e_ipsec_handle_tx_skb(priv, &wqe->eth, skb)))
return false;
}
#endif
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
index 0e1ac3e68c72..824b87ac8f9e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.c
@@ -233,11 +233,10 @@ static void mlx5e_ipsec_set_metadata(struct sk_buff *skb,
ntohs(mdata->content.tx.seq));
}
-bool mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
+bool mlx5e_ipsec_handle_tx_skb(struct mlx5e_priv *priv,
struct mlx5_wqe_eth_seg *eseg,
struct sk_buff *skb)
{
- struct mlx5e_priv *priv = netdev_priv(netdev);
struct xfrm_offload *xo = xfrm_offload(skb);
struct mlx5e_ipsec_metadata *mdata;
struct mlx5e_ipsec_sa_entry *sa_entry;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
index bd6f32aee8d6..ba02643586a5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ipsec_rxtx.h
@@ -52,7 +52,7 @@ void mlx5e_ipsec_set_iv_esn(struct sk_buff *skb, struct xfrm_state *x,
struct xfrm_offload *xo);
void mlx5e_ipsec_set_iv(struct sk_buff *skb, struct xfrm_state *x,
struct xfrm_offload *xo);
-bool mlx5e_ipsec_handle_tx_skb(struct net_device *netdev,
+bool mlx5e_ipsec_handle_tx_skb(struct mlx5e_priv *priv,
struct mlx5_wqe_eth_seg *eseg,
struct sk_buff *skb);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
index 7d9d9420f19d..dabbc5f226ce 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
@@ -9,6 +9,7 @@
#ifdef CONFIG_MLX5_EN_TLS
#include <net/tls.h>
#include "accel/tls.h"
+#include "en_accel/tls_rxtx.h"
#define MLX5E_KTLS_STATIC_UMR_WQE_SZ \
(offsetof(struct mlx5e_umr_wqe, tls_static_params_ctx) + \
@@ -96,7 +97,8 @@ void mlx5e_ktls_build_netdev(struct mlx5e_priv *priv);
void mlx5e_ktls_tx_offload_set_pending(struct mlx5e_ktls_offload_context_tx *priv_tx);
bool mlx5e_ktls_handle_tx_skb(struct tls_context *tls_ctx, struct mlx5e_txqsq *sq,
- struct sk_buff *skb, u32 *tisn, int datalen);
+ struct sk_buff *skb, int datalen,
+ struct mlx5e_accel_tx_tls_state *state);
void mlx5e_ktls_tx_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
struct mlx5e_tx_wqe_info *wi,
u32 *dma_fifo_cc);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
index b49d7c1e49dc..352b0a3ef0ad 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
@@ -414,7 +414,8 @@ mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
}
bool mlx5e_ktls_handle_tx_skb(struct tls_context *tls_ctx, struct mlx5e_txqsq *sq,
- struct sk_buff *skb, u32 *tisn, int datalen)
+ struct sk_buff *skb, int datalen,
+ struct mlx5e_accel_tx_tls_state *state)
{
struct mlx5e_ktls_offload_context_tx *priv_tx;
struct mlx5e_sq_stats *stats = sq->stats;
@@ -447,7 +448,7 @@ bool mlx5e_ktls_handle_tx_skb(struct tls_context *tls_ctx, struct mlx5e_txqsq *s
priv_tx->expected_seq = seq + datalen;
- *tisn = priv_tx->tisn;
+ state->tls_tisn = priv_tx->tisn;
stats->tls_encrypted_packets += skb_is_gso(skb) ? skb_shinfo(skb)->gso_segs : 1;
stats->tls_encrypted_bytes += datalen;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
index 8e6b0b0ce2e4..05454a843b28 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.c
@@ -258,7 +258,7 @@ static bool mlx5e_tls_handle_ooo(struct mlx5e_tls_offload_context_tx *context,
}
bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
- struct sk_buff *skb, u32 *tisn)
+ struct sk_buff *skb, struct mlx5e_accel_tx_tls_state *state)
{
struct mlx5e_priv *priv = netdev_priv(netdev);
struct mlx5e_tls_offload_context_tx *context;
@@ -279,7 +279,7 @@ bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
goto err_out;
if (MLX5_CAP_GEN(sq->channel->mdev, tls_tx))
- return mlx5e_ktls_handle_tx_skb(tls_ctx, sq, skb, tisn, datalen);
+ return mlx5e_ktls_handle_tx_skb(tls_ctx, sq, skb, datalen, state);
skb_seq = ntohl(tcp_hdr(skb)->seq);
context = mlx5e_get_tls_tx_context(tls_ctx);
@@ -302,6 +302,12 @@ bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
return false;
}
+void mlx5e_tls_handle_tx_wqe(struct mlx5e_txqsq *sq, struct mlx5_wqe_ctrl_seg *cseg,
+ struct mlx5e_accel_tx_tls_state *state)
+{
+ cseg->tisn = cpu_to_be32(state->tls_tisn << 8);
+}
+
static int tls_update_resync_sn(struct net_device *netdev,
struct sk_buff *skb,
struct mlx5e_tls_metadata *mdata)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h
index 3630ed8b1206..a50d0394df0a 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/tls_rxtx.h
@@ -40,8 +40,14 @@
#include "en.h"
#include "en/txrx.h"
+struct mlx5e_accel_tx_tls_state {
+ u32 tls_tisn;
+};
+
bool mlx5e_tls_handle_tx_skb(struct net_device *netdev, struct mlx5e_txqsq *sq,
- struct sk_buff *skb, u32 *tisn);
+ struct sk_buff *skb, struct mlx5e_accel_tx_tls_state *state);
+void mlx5e_tls_handle_tx_wqe(struct mlx5e_txqsq *sq, struct mlx5_wqe_ctrl_seg *cseg,
+ struct mlx5e_accel_tx_tls_state *state);
void mlx5e_tls_handle_rx_skb(struct net_device *netdev, struct sk_buff *skb,
u32 *cqe_bcnt);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
index bb6d3774eafb..f79454746d0d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
@@ -383,16 +383,22 @@ void mlx5e_sq_xmit(struct mlx5e_txqsq *sq, struct sk_buff *skb,
netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
{
struct mlx5e_priv *priv = netdev_priv(dev);
+ struct mlx5e_accel_tx_state accel = {};
struct mlx5e_tx_wqe *wqe;
struct mlx5e_txqsq *sq;
u16 pi;
sq = priv->txq2sq[skb_get_queue_mapping(skb)];
+
+ /* May send SKBs and WQEs. */
+ if (unlikely(!mlx5e_accel_tx_begin(dev, sq, skb, &accel)))
+ goto out;
+
pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
wqe = MLX5E_TX_FETCH_WQE(sq, pi);
- /* might send skbs and update wqe and pi */
- if (unlikely(!mlx5e_accel_handle_tx(skb, sq, dev, &wqe, &pi)))
+ /* May update the WQE, but may not post other WQEs. */
+ if (unlikely(!mlx5e_accel_tx_finish(priv, sq, skb, wqe, &accel)))
goto out;
mlx5e_sq_xmit(sq, skb, wqe, pi, netdev_xmit_more());
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 08/13] net/mlx5e: kTLS, Fill work queue edge separately in TX flow
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (6 preceding siblings ...)
2020-05-09 8:28 ` [net-next 07/13] net/mlx5e: Split TX acceleration offloads into two phases Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 09/13] net/mlx5e: kTLS, Do not fill edge for the DUMP WQEs " Saeed Mahameed
` (5 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Tariq Toukan, Maxim Mikityanskiy, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
For the static and progress context params WQEs, do the edge
filling separately.
This improves the WQ utilization, code readability, and reduces
the chance of future bugs.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../mellanox/mlx5/core/en_accel/ktls_tx.c | 18 ++++++++----------
1 file changed, 8 insertions(+), 10 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
index 352b0a3ef0ad..efc271e24b03 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
@@ -134,14 +134,14 @@ post_static_params(struct mlx5e_txqsq *sq,
struct mlx5e_ktls_offload_context_tx *priv_tx,
bool fence)
{
+ u16 pi, num_wqebbs = MLX5E_KTLS_STATIC_WQEBBS;
struct mlx5e_umr_wqe *umr_wqe;
- u16 pi;
- pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
+ pi = mlx5e_txqsq_get_next_pi(sq, num_wqebbs);
umr_wqe = MLX5E_TLS_FETCH_UMR_WQE(sq, pi);
build_static_params(umr_wqe, sq->pc, sq->sqn, priv_tx, fence);
- tx_fill_wi(sq, pi, MLX5E_KTLS_STATIC_WQEBBS, 0, NULL);
- sq->pc += MLX5E_KTLS_STATIC_WQEBBS;
+ tx_fill_wi(sq, pi, num_wqebbs, 0, NULL);
+ sq->pc += num_wqebbs;
}
static void
@@ -149,14 +149,14 @@ post_progress_params(struct mlx5e_txqsq *sq,
struct mlx5e_ktls_offload_context_tx *priv_tx,
bool fence)
{
+ u16 pi, num_wqebbs = MLX5E_KTLS_PROGRESS_WQEBBS;
struct mlx5e_tx_wqe *wqe;
- u16 pi;
- pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
+ pi = mlx5e_txqsq_get_next_pi(sq, num_wqebbs);
wqe = MLX5E_TLS_FETCH_PROGRESS_WQE(sq, pi);
build_progress_params(wqe, sq->pc, sq->sqn, priv_tx, fence);
- tx_fill_wi(sq, pi, MLX5E_KTLS_PROGRESS_WQEBBS, 0, NULL);
- sq->pc += MLX5E_KTLS_PROGRESS_WQEBBS;
+ tx_fill_wi(sq, pi, num_wqebbs, 0, NULL);
+ sq->pc += num_wqebbs;
}
static void
@@ -166,8 +166,6 @@ mlx5e_ktls_tx_post_param_wqes(struct mlx5e_txqsq *sq,
{
bool progress_fence = skip_static_post || !fence_first_post;
- mlx5e_txqsq_get_next_pi(sq, MLX5E_KTLS_STATIC_WQEBBS + MLX5E_KTLS_PROGRESS_WQEBBS);
-
if (!skip_static_post)
post_static_params(sq, priv_tx, fence_first_post);
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 09/13] net/mlx5e: kTLS, Do not fill edge for the DUMP WQEs in TX flow
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (7 preceding siblings ...)
2020-05-09 8:28 ` [net-next 08/13] net/mlx5e: kTLS, Fill work queue edge separately in TX flow Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 10/13] net/mlx5e: Take TX WQE info structures out of general EN header Saeed Mahameed
` (4 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Tariq Toukan, Maxim Mikityanskiy, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
Every single DUMP WQE resides in a single WQEBB.
As the pi is calculated per each one separately, there is
no real need for a contiguous room for them, allow them to populate
different WQ fragments.
This reduces WQ waste and improves its utilization.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c | 5 +----
1 file changed, 1 insertion(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
index efc271e24b03..1c9d0174676d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
@@ -272,6 +272,7 @@ tx_post_resync_dump(struct mlx5e_txqsq *sq, skb_frag_t *frag, u32 tisn, bool fir
int fsz;
u16 pi;
+ BUILD_BUG_ON(MLX5E_KTLS_DUMP_WQEBBS != 1);
pi = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->pc);
wqe = MLX5E_TLS_FETCH_DUMP_WQE(sq, pi);
@@ -340,7 +341,6 @@ mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
struct mlx5e_sq_stats *stats = sq->stats;
enum mlx5e_ktls_sync_retval ret;
struct tx_sync_info info = {};
- u8 num_wqebbs;
int i = 0;
ret = tx_sync_info_get(priv_tx, seq, datalen, &info);
@@ -369,9 +369,6 @@ mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
return MLX5E_KTLS_SYNC_DONE;
}
- num_wqebbs = mlx5e_ktls_dumps_num_wqebbs(sq, info.nr_frags, info.sync_len);
- mlx5e_txqsq_get_next_pi(sq, num_wqebbs);
-
for (; i < info.nr_frags; i++) {
unsigned int orig_fsz, frag_offset = 0, n = 0;
skb_frag_t *f = &info.frags[i];
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 10/13] net/mlx5e: Take TX WQE info structures out of general EN header
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (8 preceding siblings ...)
2020-05-09 8:28 ` [net-next 09/13] net/mlx5e: kTLS, Do not fill edge for the DUMP WQEs " Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 11/13] net/mlx5e: Use struct assignment for WQE info updates Saeed Mahameed
` (3 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Tariq Toukan, Maxim Mikityanskiy, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
Into the txrx header file.
The mlx5e_sq_wqe_info structure describes WQE info for the ICOSQ,
rename it to better reflect this.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en.h | 27 -------------------
.../net/ethernet/mellanox/mlx5/core/en/txrx.h | 22 +++++++++++++++
.../net/ethernet/mellanox/mlx5/core/en/xdp.h | 5 ++++
3 files changed, 27 insertions(+), 27 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index da7fe6aafeed..3bd64c63865b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -339,16 +339,6 @@ struct mlx5e_cq_decomp {
u16 wqe_counter;
} ____cacheline_aligned_in_smp;
-struct mlx5e_tx_wqe_info {
- struct sk_buff *skb;
- u32 num_bytes;
- u8 num_wqebbs;
- u8 num_dma;
-#ifdef CONFIG_MLX5_EN_TLS
- struct page *resync_dump_frag_page;
-#endif
-};
-
enum mlx5e_dma_map_type {
MLX5E_DMA_MAP_SINGLE,
MLX5E_DMA_MAP_PAGE
@@ -370,18 +360,6 @@ enum {
MLX5E_SQ_STATE_PENDING_XSK_TX,
};
-struct mlx5e_icosq_wqe_info {
- u8 opcode;
- u8 num_wqebbs;
-
- /* Auxiliary data for different opcodes. */
- union {
- struct {
- struct mlx5e_rq *rq;
- } umr;
- };
-};
-
struct mlx5e_txqsq {
/* data path */
@@ -484,11 +462,6 @@ struct mlx5e_xdp_info_fifo {
u32 mask;
};
-struct mlx5e_xdp_wqe_info {
- u8 num_wqebbs;
- u8 num_pkts;
-};
-
struct mlx5e_xdp_mpwqe {
/* Current MPWQE session */
struct mlx5e_tx_wqe *wqe;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
index 89fe65593c16..9e150d160cde 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
@@ -81,6 +81,16 @@ mlx5e_post_nop_fence(struct mlx5_wq_cyc *wq, u32 sqn, u16 *pc)
return wqe;
}
+struct mlx5e_tx_wqe_info {
+ struct sk_buff *skb;
+ u32 num_bytes;
+ u8 num_wqebbs;
+ u8 num_dma;
+#ifdef CONFIG_MLX5_EN_TLS
+ struct page *resync_dump_frag_page;
+#endif
+};
+
static inline u16 mlx5e_txqsq_get_next_pi(struct mlx5e_txqsq *sq, u16 size)
{
struct mlx5_wq_cyc *wq = &sq->wq;
@@ -109,6 +119,18 @@ static inline u16 mlx5e_txqsq_get_next_pi(struct mlx5e_txqsq *sq, u16 size)
return pi;
}
+struct mlx5e_icosq_wqe_info {
+ u8 opcode;
+ u8 num_wqebbs;
+
+ /* Auxiliary data for different opcodes. */
+ union {
+ struct {
+ struct mlx5e_rq *rq;
+ } umr;
+ };
+};
+
static inline u16 mlx5e_icosq_get_next_pi(struct mlx5e_icosq *sq, u16 size)
{
struct mlx5_wq_cyc *wq = &sq->wq;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.h b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.h
index ed6f045febeb..e2e01f064c1e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xdp.h
@@ -137,6 +137,11 @@ mlx5e_xdp_no_room_for_inline_pkt(struct mlx5e_xdp_mpwqe *session)
session->ds_count + MLX5E_XDP_INLINE_WQE_MAX_DS_CNT > MLX5E_XDP_MPW_MAX_NUM_DS;
}
+struct mlx5e_xdp_wqe_info {
+ u8 num_wqebbs;
+ u8 num_pkts;
+};
+
static inline void
mlx5e_xdp_mpwqe_add_dseg(struct mlx5e_xdpsq *sq,
struct mlx5e_xdp_xmit_data *xdptxd,
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 11/13] net/mlx5e: Use struct assignment for WQE info updates
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (9 preceding siblings ...)
2020-05-09 8:28 ` [net-next 10/13] net/mlx5e: Take TX WQE info structures out of general EN header Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 12/13] net/mlx5: Accel, Remove unnecessary header include Saeed Mahameed
` (2 subsequent siblings)
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Tariq Toukan, Maxim Mikityanskiy, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
Struct assignment looks more clean, and implies resetting
the not assigned fields to zero, instead of holding values
from older ring cycles.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../mellanox/mlx5/core/en_accel/ktls_tx.c | 9 +++++----
.../net/ethernet/mellanox/mlx5/core/en_main.c | 16 ++++++++--------
drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 9 ++++++---
.../net/ethernet/mellanox/mlx5/core/en_txrx.c | 7 +++++--
4 files changed, 24 insertions(+), 17 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
index 1c9d0174676d..3cd78d9503c1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
@@ -108,10 +108,11 @@ static void tx_fill_wi(struct mlx5e_txqsq *sq,
{
struct mlx5e_tx_wqe_info *wi = &sq->db.wqe_info[pi];
- memset(wi, 0, sizeof(*wi));
- wi->num_wqebbs = num_wqebbs;
- wi->num_bytes = num_bytes;
- wi->resync_dump_frag_page = page;
+ *wi = (struct mlx5e_tx_wqe_info) {
+ .num_wqebbs = num_wqebbs,
+ .num_bytes = num_bytes,
+ .resync_dump_frag_page = page,
+ };
}
void mlx5e_ktls_tx_offload_set_pending(struct mlx5e_ktls_offload_context_tx *priv_tx)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 048a4f8601a8..0a9dfc31de3e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -1364,13 +1364,12 @@ static void mlx5e_deactivate_txqsq(struct mlx5e_txqsq *sq)
/* last doorbell out, godspeed .. */
if (mlx5e_wqc_has_room_for(wq, sq->cc, sq->pc, 1)) {
u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
- struct mlx5e_tx_wqe_info *wi;
struct mlx5e_tx_wqe *nop;
- wi = &sq->db.wqe_info[pi];
+ sq->db.wqe_info[pi] = (struct mlx5e_tx_wqe_info) {
+ .num_wqebbs = 1,
+ };
- memset(wi, 0, sizeof(*wi));
- wi->num_wqebbs = 1;
nop = mlx5e_post_nop(wq, sq->sqn, &sq->pc);
mlx5e_notify_hw(wq, sq->pc, sq->uar_map, &nop->ctrl);
}
@@ -1482,20 +1481,21 @@ int mlx5e_open_xdpsq(struct mlx5e_channel *c, struct mlx5e_params *params,
/* Pre initialize fixed WQE fields */
for (i = 0; i < mlx5_wq_cyc_get_size(&sq->wq); i++) {
- struct mlx5e_xdp_wqe_info *wi = &sq->db.wqe_info[i];
struct mlx5e_tx_wqe *wqe = mlx5_wq_cyc_get_wqe(&sq->wq, i);
struct mlx5_wqe_ctrl_seg *cseg = &wqe->ctrl;
struct mlx5_wqe_eth_seg *eseg = &wqe->eth;
struct mlx5_wqe_data_seg *dseg;
+ sq->db.wqe_info[i] = (struct mlx5e_xdp_wqe_info) {
+ .num_wqebbs = 1,
+ .num_pkts = 1,
+ };
+
cseg->qpn_ds = cpu_to_be32((sq->sqn << 8) | ds_cnt);
eseg->inline_hdr.sz = cpu_to_be16(inline_hdr_sz);
dseg = (struct mlx5_wqe_data_seg *)cseg + (ds_cnt - 1);
dseg->lkey = sq->mkey_be;
-
- wi->num_wqebbs = 1;
- wi->num_pkts = 1;
}
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index d9a5a669b84d..8142b6e70857 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -505,9 +505,12 @@ static int mlx5e_alloc_rx_mpwqe(struct mlx5e_rq *rq, u16 ix)
MLX5_OPCODE_UMR);
umr_wqe->uctrl.xlt_offset = cpu_to_be16(xlt_offset);
- sq->db.wqe_info[pi].opcode = MLX5_OPCODE_UMR;
- sq->db.wqe_info[pi].num_wqebbs = MLX5E_UMR_WQEBBS;
- sq->db.wqe_info[pi].umr.rq = rq;
+ sq->db.wqe_info[pi] = (struct mlx5e_icosq_wqe_info) {
+ .opcode = MLX5_OPCODE_UMR,
+ .num_wqebbs = MLX5E_UMR_WQEBBS,
+ .umr.rq = rq,
+ };
+
sq->pc += MLX5E_UMR_WQEBBS;
sq->doorbell_cseg = &umr_wqe->ctrl;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
index 869fd58a6775..73293f9c3f63 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
@@ -78,8 +78,11 @@ void mlx5e_trigger_irq(struct mlx5e_icosq *sq)
struct mlx5e_tx_wqe *nopwqe;
u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
- sq->db.wqe_info[pi].opcode = MLX5_OPCODE_NOP;
- sq->db.wqe_info[pi].num_wqebbs = 1;
+ sq->db.wqe_info[pi] = (struct mlx5e_icosq_wqe_info) {
+ .opcode = MLX5_OPCODE_NOP,
+ .num_wqebbs = 1,
+ };
+
nopwqe = mlx5e_post_nop(wq, sq->sqn, &sq->pc);
mlx5e_notify_hw(wq, sq->pc, sq->uar_map, &nopwqe->ctrl);
}
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 12/13] net/mlx5: Accel, Remove unnecessary header include
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (10 preceding siblings ...)
2020-05-09 8:28 ` [net-next 11/13] net/mlx5e: Use struct assignment for WQE info updates Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 8:28 ` [net-next 13/13] net/mlx5e: Enhance ICOSQ WQE info fields Saeed Mahameed
2020-05-09 22:13 ` [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Jakub Kicinski
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Tariq Toukan, Maxim Mikityanskiy, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
The include of Ethernet driver header in core is not needed
and actually wrong.
Remove it.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/accel/accel.h | 1 -
1 file changed, 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/accel/accel.h b/drivers/net/ethernet/mellanox/mlx5/core/accel/accel.h
index c13260467750..82b185121edb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/accel/accel.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/accel/accel.h
@@ -5,7 +5,6 @@
#include <linux/skbuff.h>
#include <linux/netdevice.h>
-#include "en.h"
static inline bool is_metadata_hdr_valid(struct sk_buff *skb)
{
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* [net-next 13/13] net/mlx5e: Enhance ICOSQ WQE info fields
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (11 preceding siblings ...)
2020-05-09 8:28 ` [net-next 12/13] net/mlx5: Accel, Remove unnecessary header include Saeed Mahameed
@ 2020-05-09 8:28 ` Saeed Mahameed
2020-05-09 22:13 ` [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Jakub Kicinski
13 siblings, 0 replies; 15+ messages in thread
From: Saeed Mahameed @ 2020-05-09 8:28 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Tariq Toukan, Maxim Mikityanskiy, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
The same WQE opcode might be used in different ICOSQ flows
and WQE types.
To have a better distinguishability, replace it with an enum that
better indicates the WQE type and flow it is used for.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../net/ethernet/mellanox/mlx5/core/en/txrx.h | 11 ++++++++---
drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 17 ++++++++++-------
.../net/ethernet/mellanox/mlx5/core/en_txrx.c | 2 +-
3 files changed, 19 insertions(+), 11 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
index 9e150d160cde..dce2bbbf9109 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
@@ -27,6 +27,11 @@
#define INL_HDR_START_SZ (sizeof(((struct mlx5_wqe_eth_seg *)NULL)->inline_hdr.start))
+enum mlx5e_icosq_wqe_type {
+ MLX5E_ICOSQ_WQE_NOP,
+ MLX5E_ICOSQ_WQE_UMR_RX,
+};
+
static inline bool
mlx5e_wqc_has_room_for(struct mlx5_wq_cyc *wq, u16 cc, u16 pc, u16 n)
{
@@ -120,10 +125,10 @@ static inline u16 mlx5e_txqsq_get_next_pi(struct mlx5e_txqsq *sq, u16 size)
}
struct mlx5e_icosq_wqe_info {
- u8 opcode;
+ u8 wqe_type;
u8 num_wqebbs;
- /* Auxiliary data for different opcodes. */
+ /* Auxiliary data for different wqe types. */
union {
struct {
struct mlx5e_rq *rq;
@@ -147,7 +152,7 @@ static inline u16 mlx5e_icosq_get_next_pi(struct mlx5e_icosq *sq, u16 size)
/* Fill SQ frag edge with NOPs to avoid WQE wrapping two pages. */
for (; wi < edge_wi; wi++) {
*wi = (struct mlx5e_icosq_wqe_info) {
- .opcode = MLX5_OPCODE_NOP,
+ .wqe_type = MLX5E_ICOSQ_WQE_NOP,
.num_wqebbs = 1,
};
mlx5e_post_nop(wq, sq->sqn, &sq->pc);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index 8142b6e70857..779600bebcca 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -506,7 +506,7 @@ static int mlx5e_alloc_rx_mpwqe(struct mlx5e_rq *rq, u16 ix)
umr_wqe->uctrl.xlt_offset = cpu_to_be16(xlt_offset);
sq->db.wqe_info[pi] = (struct mlx5e_icosq_wqe_info) {
- .opcode = MLX5_OPCODE_UMR,
+ .wqe_type = MLX5E_ICOSQ_WQE_UMR_RX,
.num_wqebbs = MLX5E_UMR_WQEBBS,
.umr.rq = rq,
};
@@ -619,15 +619,18 @@ int mlx5e_poll_ico_cq(struct mlx5e_cq *cq)
break;
}
- if (likely(wi->opcode == MLX5_OPCODE_UMR))
+ switch (wi->wqe_type) {
+ case MLX5E_ICOSQ_WQE_UMR_RX:
wi->umr.rq->mpwqe.umr_completed++;
- else if (unlikely(wi->opcode != MLX5_OPCODE_NOP))
+ break;
+ case MLX5E_ICOSQ_WQE_NOP:
+ break;
+ default:
netdev_WARN_ONCE(cq->channel->netdev,
- "Bad OPCODE in ICOSQ WQE info: 0x%x\n",
- wi->opcode);
-
+ "Bad WQE type in ICOSQ WQE info: 0x%x\n",
+ wi->wqe_type);
+ }
} while (!last_wqe);
-
} while ((++i < MLX5E_TX_CQ_POLL_BUDGET) && (cqe = mlx5_cqwq_get_cqe(&cq->wq)));
sq->cc = sqcc;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
index 73293f9c3f63..8480278f2ee2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
@@ -79,7 +79,7 @@ void mlx5e_trigger_irq(struct mlx5e_icosq *sq)
u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
sq->db.wqe_info[pi] = (struct mlx5e_icosq_wqe_info) {
- .opcode = MLX5_OPCODE_NOP,
+ .wqe_type = MLX5E_ICOSQ_WQE_NOP,
.num_wqebbs = 1,
};
--
2.25.4
^ permalink raw reply related [flat|nested] 15+ messages in thread
* Re: [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
` (12 preceding siblings ...)
2020-05-09 8:28 ` [net-next 13/13] net/mlx5e: Enhance ICOSQ WQE info fields Saeed Mahameed
@ 2020-05-09 22:13 ` Jakub Kicinski
13 siblings, 0 replies; 15+ messages in thread
From: Jakub Kicinski @ 2020-05-09 22:13 UTC (permalink / raw)
To: Saeed Mahameed; +Cc: David S. Miller, netdev
On Sat, 9 May 2020 01:28:43 -0700 Saeed Mahameed wrote:
> Hi Dave,
>
> This series includes updates to mlx5 driver and a merge commit to the
> earlier bonding series which was submitted by Maor to mlx5-next branch.
>
> For more information please see tag log below.
>
> Please pull and let me know if there is any problem.
>
> Please note that the series starts with a merge of mlx5-next branch,
> to resolve and avoid dependency with rdma tree.
>
> A minor conflict between Maor's bonding series [1] and Eric's [2] is
> already handled in this merge commit.
>
> [1] bonding: Add support to get xmit slave
> [2] bonding: report transmit status to callers
Pulled, thank you!
^ permalink raw reply [flat|nested] 15+ messages in thread
end of thread, other threads:[~2020-05-09 22:13 UTC | newest]
Thread overview: 15+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-05-09 8:28 [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Saeed Mahameed
2020-05-09 8:28 ` [net-next 01/13] net/mlx5e: Return bool from TLS and IPSEC offloads Saeed Mahameed
2020-05-09 8:28 ` [net-next 02/13] net/mlx5e: Unify checks of TLS offloads Saeed Mahameed
2020-05-09 8:28 ` [net-next 03/13] net/mlx5e: Return void from mlx5e_sq_xmit and mlx5i_sq_xmit Saeed Mahameed
2020-05-09 8:28 ` [net-next 04/13] net/mlx5e: Pass only eseg to IPSEC offload Saeed Mahameed
2020-05-09 8:28 ` [net-next 05/13] net/mlx5e: Make TLS offload independent of wqe and pi Saeed Mahameed
2020-05-09 8:28 ` [net-next 06/13] net/mlx5e: Update UDP fields of the SKB for GSO first Saeed Mahameed
2020-05-09 8:28 ` [net-next 07/13] net/mlx5e: Split TX acceleration offloads into two phases Saeed Mahameed
2020-05-09 8:28 ` [net-next 08/13] net/mlx5e: kTLS, Fill work queue edge separately in TX flow Saeed Mahameed
2020-05-09 8:28 ` [net-next 09/13] net/mlx5e: kTLS, Do not fill edge for the DUMP WQEs " Saeed Mahameed
2020-05-09 8:28 ` [net-next 10/13] net/mlx5e: Take TX WQE info structures out of general EN header Saeed Mahameed
2020-05-09 8:28 ` [net-next 11/13] net/mlx5e: Use struct assignment for WQE info updates Saeed Mahameed
2020-05-09 8:28 ` [net-next 12/13] net/mlx5: Accel, Remove unnecessary header include Saeed Mahameed
2020-05-09 8:28 ` [net-next 13/13] net/mlx5e: Enhance ICOSQ WQE info fields Saeed Mahameed
2020-05-09 22:13 ` [pull request][net-next 00/13] Mellanox, mlx5 and bonding updates 2020-05-09 Jakub Kicinski
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.