[PATCH net 0/6] Mellanox mlx5e fixes for 4.11-rc1

STALE3476d

10 messages, 4 authors, 2017-02-23 · open the first message on its own page

[PATCH net 0/6] Mellanox mlx5e fixes for 4.11-rc1

From: Saeed Mahameed <hidden>
Date: 2017-02-22 15:21:47

Hi Dave,

This series includes some important bug fixes for mlx5e driver.

Three misc fixes:
From Mohamad, compilation fix on s390 system
From Me, A fix for driver unload when switchdev mode is on. 
From Tariq, HW LRO frag size optimization for when build_skb is not used
(striding RQ mode).

Three CQE compression related fixes:
Two fixes from Tariq and I, to correctly setup CQE compression
parameters on driver load and on arbitrary user modifications.
Last patch, fixes a very critical issue that was originally reported
by Tom, where the driver reported csum errors or even page ref issues
for when cqe compression is enabled and rapidly active.

For your convenience this series was generated on top of net-next branch:
005c3490e9db ('Revert "ath10k: Search SMBIOS for OEM board file extension"')

for -stable:
net/mlx5e: Register/unregister vport representors on interface (for kernel >= 4.9)
net/mlx5e: Do not reduce LRO WQE size when not using build_skb (for kernel >= 4.9)
net/mlx5e: Fix broken CQE compression initialization (for kernel >= 4.9)
net/mlx5e: Update MPWQE stride size when modifying CQE compress state (for kernel >= 4.7)
net/mlx5e: Fix wrong CQE decompression (for kernel >= 4.7)

Thanks,
Saeed.

Mohamad Haj Yahia (1):
  net/mlx5e: s390 system compilation fix

Saeed Mahameed (2):
  net/mlx5e: Register/unregister vport representors on interface
    attach/detach
  net/mlx5e: Update MPWQE stride size when modifying CQE compress state

Tariq Toukan (3):
  net/mlx5e: Do not reduce LRO WQE size when not using build_skb
  net/mlx5e: Fix broken CQE compression initialization
  net/mlx5e: Fix wrong CQE decompression

 drivers/net/ethernet/mellanox/mlx5/core/en.h       |  1 +
 .../net/ethernet/mellanox/mlx5/core/en_ethtool.c   |  1 +
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c  | 40 +++++++++++++---------
 drivers/net/ethernet/mellanox/mlx5/core/en_rx.c    | 15 ++++----
 .../net/ethernet/mellanox/mlx5/core/en_selftest.c  |  1 +
 5 files changed, 35 insertions(+), 23 deletions(-)

-- 
2.11.0

[PATCH net 3/6] net/mlx5e: Do not reduce LRO WQE size when not using build_skb

From: Saeed Mahameed <hidden>
Date: 2017-02-22 15:21:47

From: Tariq Toukan <redacted>

When rq_type is Striding RQ, no room of SKB_RESERVE is needed
as SKB allocation is not done via build_skb.

Fixes: e4b85508072b ("net/mlx5e: Slightly reduce hardware LRO size")
Signed-off-by: Tariq Toukan <redacted>
Signed-off-by: Saeed Mahameed <redacted>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index c24366868b39..0d58729b42ae 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -82,6 +82,7 @@ static bool mlx5e_check_fragmented_striding_rq_cap(struct mlx5_core_dev *mdev)
 static void mlx5e_set_rq_type_params(struct mlx5e_priv *priv, u8 rq_type)
 {
 	priv->params.rq_wq_type = rq_type;
+	priv->params.lro_wqe_sz = MLX5E_PARAMS_DEFAULT_LRO_WQE_SZ;
 	switch (priv->params.rq_wq_type) {
 	case MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ:
 		priv->params.log_rq_size = is_kdump_kernel() ?
@@ -98,6 +99,10 @@ static void mlx5e_set_rq_type_params(struct mlx5e_priv *priv, u8 rq_type)
 		priv->params.log_rq_size = is_kdump_kernel() ?
 			MLX5E_PARAMS_MINIMUM_LOG_RQ_SIZE :
 			MLX5E_PARAMS_DEFAULT_LOG_RQ_SIZE;
+
+		/* Extra room needed for build_skb */
+		priv->params.lro_wqe_sz -= MLX5_RX_HEADROOM +
+			SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
 	}
 	priv->params.min_rx_wqes = mlx5_min_rx_wqes(priv->params.rq_wq_type,
 					       BIT(priv->params.log_rq_size));
@@ -3547,12 +3552,6 @@ static void mlx5e_build_nic_netdev_priv(struct mlx5_core_dev *mdev,
 	mlx5e_build_default_indir_rqt(mdev, priv->params.indirection_rqt,
 				      MLX5E_INDIR_RQT_SIZE, profile->max_nch(mdev));
 
-	priv->params.lro_wqe_sz =
-		MLX5E_PARAMS_DEFAULT_LRO_WQE_SZ -
-		/* Extra room needed for build_skb */
-		MLX5_RX_HEADROOM -
-		SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
-
 	/* Initialize pflags */
 	MLX5E_SET_PFLAG(priv, MLX5E_PFLAG_RX_CQE_BASED_MODER,
 			priv->params.rx_cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE);
-- 
2.11.0

[PATCH net 1/6] net/mlx5e: s390 system compilation fix

From: Saeed Mahameed <hidden>
Date: 2017-02-22 15:21:47

From: Mohamad Haj Yahia <redacted>

Add necessary headers include for s390 arch compilation.

Fixes: e586b3b0baee ("net/mlx5: Ethernet Datapath files")
Fixes: d605d6686dc7 ("net/mlx5e: Add support for ethtool self..")
Signed-off-by: Mohamad Haj Yahia <redacted>
Signed-off-by: Saeed Mahameed <redacted>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_rx.c       | 1 +
 drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c | 1 +
 2 files changed, 2 insertions(+)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index b039b87742a6..9fad22768aab 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -30,6 +30,7 @@
  * SOFTWARE.
  */
 
+#include <linux/prefetch.h>
 #include <linux/ip.h>
 #include <linux/ipv6.h>
 #include <linux/tcp.h>
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c b/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c
index 65442c36a6e1..31e3cb7ee5fe 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_selftest.c
@@ -30,6 +30,7 @@
  * SOFTWARE.
  */
 
+#include <linux/prefetch.h>
 #include <linux/ip.h>
 #include <linux/udp.h>
 #include <net/udp.h>
-- 
2.11.0

[PATCH net 6/6] net/mlx5e: Fix wrong CQE decompression

From: Saeed Mahameed <hidden>
Date: 2017-02-22 15:21:47

From: Tariq Toukan <redacted>

In cqe compression with striding RQ, the decompression of the CQE field
wqe_counter was done with a wrong wraparound value.
This caused handling cqes with a wrong pointer to wqe (rx descriptor)
and creating SKBs with wrong data, pointing to wrong (and already consumed)
strides/pages.

The meaning of the CQE field wqe_counter in striding RQ holds the
stride index instead of the WQE index. Hence, when decompressing
a CQE, wqe_counter should have wrapped-around the number of strides
in a single multi-packet WQE.

We dropped this wrap-around mask at all in CQE decompression of striding
RQ. It is not needed as in such cases the CQE compression session would
break because of different value of wqe_id field, starting a new
compression session.

Tested:
 ethtool -K ethxx lro off/on
 ethtool --set-priv-flags ethxx rx_cqe_compress on
 super_netperf 16 {ipv4,ipv6} -t TCP_STREAM -m 50 -D
 verified no csum errors and no page refcount issues.

Fixes: 7219ab34f184 ("net/mlx5e: CQE compression")
Signed-off-by: Tariq Toukan <redacted>
Reported-by: Tom Herbert <redacted>
Cc: kernel-team@fb.com
Signed-off-by: Saeed Mahameed <redacted>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_rx.c | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index d5ce20db3f0b..3d371688fbbb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -94,19 +94,18 @@ static inline void mlx5e_cqes_update_owner(struct mlx5e_cq *cq, u32 cqcc, int n)
 static inline void mlx5e_decompress_cqe(struct mlx5e_rq *rq,
 					struct mlx5e_cq *cq, u32 cqcc)
 {
-	u16 wqe_cnt_step;
-
 	cq->title.byte_cnt     = cq->mini_arr[cq->mini_arr_idx].byte_cnt;
 	cq->title.check_sum    = cq->mini_arr[cq->mini_arr_idx].checksum;
 	cq->title.op_own      &= 0xf0;
 	cq->title.op_own      |= 0x01 & (cqcc >> cq->wq.log_sz);
 	cq->title.wqe_counter  = cpu_to_be16(cq->decmprs_wqe_counter);
 
-	wqe_cnt_step =
-		rq->wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ ?
-		mpwrq_get_cqe_consumed_strides(&cq->title) : 1;
-	cq->decmprs_wqe_counter =
-		(cq->decmprs_wqe_counter + wqe_cnt_step) & rq->wq.sz_m1;
+	if (rq->wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ)
+		cq->decmprs_wqe_counter +=
+			mpwrq_get_cqe_consumed_strides(&cq->title);
+	else
+		cq->decmprs_wqe_counter =
+			(cq->decmprs_wqe_counter + 1) & rq->wq.sz_m1;
 }
 
 static inline void mlx5e_decompress_cqe_no_hash(struct mlx5e_rq *rq,
-- 
2.11.0

[PATCH net 5/6] net/mlx5e: Update MPWQE stride size when modifying CQE compress state

From: Saeed Mahameed <hidden>
Date: 2017-02-22 15:21:47

When the admin enables/disables cqe compression, updating
mpwqe stride size is required:
    CQE compress ON  ==> stride size = 256B
    CQE compress OFF ==> stride size = 64B

This is already done on driver load via mlx5e_set_rq_type_params, all we
need is just to call it on arbitrary admin changes of cqe compression
state via priv flags or when changing timestamping state
(as it is mutually exclusive with cqe compression).

This bug introduces no functional damage, it only makes cqe compression
occur less often, since in ConnectX4-LX CQE compression is performed
only on packets smaller than stride size.

Tested:
 ethtool --set-priv-flags ethxx rx_cqe_compress on
 pktgen with  64 < pkt size < 256 and netperf TCP_STREAM (IPv4/IPv6)
 verify `ethtool -S ethxx | grep compress` are advancing more often
 (rapidly)

Fixes: 7219ab34f184 ("net/mlx5e: CQE compression")
Signed-off-by: Saeed Mahameed <redacted>
Reviewed-by: Tariq Toukan <redacted>
Cc: kernel-team@fb.com
---
 drivers/net/ethernet/mellanox/mlx5/core/en.h         | 1 +
 drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c | 1 +
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c    | 2 +-
 drivers/net/ethernet/mellanox/mlx5/core/en_rx.c      | 1 +
 4 files changed, 4 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 95ca03c0d9f5..f6a6ded204f6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -816,6 +816,7 @@ int mlx5e_get_max_linkspeed(struct mlx5_core_dev *mdev, u32 *speed);
 
 void mlx5e_set_rx_cq_mode_params(struct mlx5e_params *params,
 				 u8 cq_period_mode);
+void mlx5e_set_rq_type_params(struct mlx5e_priv *priv, u8 rq_type);
 
 static inline void mlx5e_tx_notify_hw(struct mlx5e_sq *sq,
 				      struct mlx5_wqe_ctrl_seg *ctrl, int bf_sz)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index cc80522b5854..a004a5a1a4c2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -1487,6 +1487,7 @@ static int set_pflag_rx_cqe_compress(struct net_device *netdev,
 
 	mlx5e_modify_rx_cqe_compression_locked(priv, enable);
 	priv->params.rx_cqe_compress_def = enable;
+	mlx5e_set_rq_type_params(priv, priv->params.rq_wq_type);
 
 	return 0;
 }
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index dc621bc4e173..8ef64c4db2c2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -79,7 +79,7 @@ static bool mlx5e_check_fragmented_striding_rq_cap(struct mlx5_core_dev *mdev)
 		MLX5_CAP_ETH(mdev, reg_umr_sq);
 }
 
-static void mlx5e_set_rq_type_params(struct mlx5e_priv *priv, u8 rq_type)
+void mlx5e_set_rq_type_params(struct mlx5e_priv *priv, u8 rq_type)
 {
 	priv->params.rq_wq_type = rq_type;
 	priv->params.lro_wqe_sz = MLX5E_PARAMS_DEFAULT_LRO_WQE_SZ;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index 9fad22768aab..d5ce20db3f0b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -172,6 +172,7 @@ void mlx5e_modify_rx_cqe_compression_locked(struct mlx5e_priv *priv, bool val)
 		mlx5e_close_locked(priv->netdev);
 
 	MLX5E_SET_PFLAG(priv, MLX5E_PFLAG_RX_CQE_COMPRESS, val);
+	mlx5e_set_rq_type_params(priv, priv->params.rq_wq_type);
 
 	if (was_opened)
 		mlx5e_open_locked(priv->netdev);
-- 
2.11.0

[PATCH net 4/6] net/mlx5e: Fix broken CQE compression initialization

From: Saeed Mahameed <hidden>
Date: 2017-02-22 15:21:47

From: Tariq Toukan <redacted>

Some of RQ type parameters are derived from CQE compression state flag,
CQE compression flag was initialized only after RQ type parameters
setup. This leads to load RQ with stride size smaller than what we
want for when CQE compression is on.

This bug introduces no functional damage, it only makes CQE compression
occur less often, since in ConnectX4-LX CQE compression is performed
only on packets smaller than stride size.

Fix this by marking default status of CQE compression in PFLAG prior to
calling mlx5e_set_rq_priv_params(), as it inits some fields based on it.

Tested:
 load driver on systems where rx CQE compress will be on (MH)
 pktgen with  64 < pkt size < 256 and netperf TCP_STREAM (IPv4/IPv6)
 verify `ethtool -S ethxx | grep compress` are advancing more often
 (rapidly)

Fixes: 2fc4bfb7250d ("net/mlx5e: Dynamic RQ type infrastructure")
Signed-off-by: Tariq Toukan <redacted>
Cc: kernel-team@fb.com
Signed-off-by: Saeed Mahameed <redacted>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 0d58729b42ae..dc621bc4e173 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -3526,6 +3526,9 @@ static void mlx5e_build_nic_netdev_priv(struct mlx5_core_dev *mdev,
 			cqe_compress_heuristic(link_speed, pci_bw);
 	}
 
+	MLX5E_SET_PFLAG(priv, MLX5E_PFLAG_RX_CQE_COMPRESS,
+			priv->params.rx_cqe_compress_def);
+
 	mlx5e_set_rq_priv_params(priv);
 	if (priv->params.rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ)
 		priv->params.lro_en = true;
@@ -3555,7 +3558,6 @@ static void mlx5e_build_nic_netdev_priv(struct mlx5_core_dev *mdev,
 	/* Initialize pflags */
 	MLX5E_SET_PFLAG(priv, MLX5E_PFLAG_RX_CQE_BASED_MODER,
 			priv->params.rx_cq_period_mode == MLX5_CQ_PERIOD_MODE_START_FROM_CQE);
-	MLX5E_SET_PFLAG(priv, MLX5E_PFLAG_RX_CQE_COMPRESS, priv->params.rx_cqe_compress_def);
 
 	mutex_init(&priv->state_lock);
 
-- 
2.11.0

[PATCH net 2/6] net/mlx5e: Register/unregister vport representors on interface attach/detach

From: Saeed Mahameed <hidden>
Date: 2017-02-22 15:21:47

Currently vport representors are added only on driver load and removed on
driver unload.  Apparently we forgot to handle them when we added the
seamless reset flow feature.  This caused to leave the representors
netdevs alive and active with open HW resources on pci shutdown and on
error reset flows.

To overcome this we move their handling to interface attach/detach, so
they would be cleaned up on shutdown and recreated on reset flows.

Fixes: 26e59d8077a3 ("net/mlx5e: Implement mlx5e interface attach/detach callbacks")
Signed-off-by: Saeed Mahameed <redacted>
Reviewed-by: Hadar Hen Zion <redacted>
Reviewed-by: Roi Dayan <redacted>
---
 drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 23 +++++++++++++++--------
 1 file changed, 15 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 3cce6281e075..c24366868b39 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -3970,6 +3970,19 @@ static void mlx5e_register_vport_rep(struct mlx5_core_dev *mdev)
 	}
 }
 
+static void mlx5e_unregister_vport_rep(struct mlx5_core_dev *mdev)
+{
+	struct mlx5_eswitch *esw = mdev->priv.eswitch;
+	int total_vfs = MLX5_TOTAL_VPORTS(mdev);
+	int vport;
+
+	if (!MLX5_CAP_GEN(mdev, vport_group_manager))
+		return;
+
+	for (vport = 1; vport < total_vfs; vport++)
+		mlx5_eswitch_unregister_vport_rep(esw, vport);
+}
+
 void mlx5e_detach_netdev(struct mlx5_core_dev *mdev, struct net_device *netdev)
 {
 	struct mlx5e_priv *priv = netdev_priv(netdev);
@@ -4016,6 +4029,7 @@ static int mlx5e_attach(struct mlx5_core_dev *mdev, void *vpriv)
 		return err;
 	}
 
+	mlx5e_register_vport_rep(mdev);
 	return 0;
 }
 
@@ -4027,6 +4041,7 @@ static void mlx5e_detach(struct mlx5_core_dev *mdev, void *vpriv)
 	if (!netif_device_present(netdev))
 		return;
 
+	mlx5e_unregister_vport_rep(mdev);
 	mlx5e_detach_netdev(mdev, netdev);
 	mlx5e_destroy_mdev_resources(mdev);
 }
@@ -4045,8 +4060,6 @@ static void *mlx5e_add(struct mlx5_core_dev *mdev)
 	if (err)
 		return NULL;
 
-	mlx5e_register_vport_rep(mdev);
-
 	if (MLX5_CAP_GEN(mdev, vport_group_manager))
 		ppriv = &esw->offloads.vport_reps[0];
 
@@ -4098,13 +4111,7 @@ void mlx5e_destroy_netdev(struct mlx5_core_dev *mdev, struct mlx5e_priv *priv)
 
 static void mlx5e_remove(struct mlx5_core_dev *mdev, void *vpriv)
 {
-	struct mlx5_eswitch *esw = mdev->priv.eswitch;
-	int total_vfs = MLX5_TOTAL_VPORTS(mdev);
 	struct mlx5e_priv *priv = vpriv;
-	int vport;
-
-	for (vport = 1; vport < total_vfs; vport++)
-		mlx5_eswitch_unregister_vport_rep(esw, vport);
 
 	unregister_netdev(priv->netdev);
 	mlx5e_detach(mdev, vpriv);
-- 
2.11.0

Re: [PATCH net 3/6] net/mlx5e: Do not reduce LRO WQE size when not using build_skb

From: Alexei Starovoitov <hidden>
Date: 2017-02-22 19:46:22

On Wed, Feb 22, 2017 at 7:20 AM, Saeed Mahameed [off-list ref] wrote:
From: Tariq Toukan <redacted>

When rq_type is Striding RQ, no room of SKB_RESERVE is needed
as SKB allocation is not done via build_skb.

Fixes: e4b85508072b ("net/mlx5e: Slightly reduce hardware LRO size")
Signed-off-by: Tariq Toukan <redacted>
Signed-off-by: Saeed Mahameed <redacted>
why this one is a bug fix?
Sound like an optimization from commit log.

Re: [PATCH net 3/6] net/mlx5e: Do not reduce LRO WQE size when not using build_skb

From: Saeed Mahameed <hidden>
Date: 2017-02-23 11:39:05

On Wed, Feb 22, 2017 at 9:45 PM, Alexei Starovoitov
[off-list ref] wrote:
On Wed, Feb 22, 2017 at 7:20 AM, Saeed Mahameed [off-list ref] wrote:
quoted
From: Tariq Toukan <redacted>

When rq_type is Striding RQ, no room of SKB_RESERVE is needed
as SKB allocation is not done via build_skb.

Fixes: e4b85508072b ("net/mlx5e: Slightly reduce hardware LRO size")
Signed-off-by: Tariq Toukan <redacted>
Signed-off-by: Saeed Mahameed <redacted>
why this one is a bug fix?
Sound like an optimization from commit log.
It is a regression since  ("net/mlx5e: Slightly reduce hardware LRO size").
And we see due to this a small drop in HW LRO performance.
We just fixed the LRO size to be the same as it was before the
offending patch for striding RQ case.

Re: [PATCH net 0/6] Mellanox mlx5e fixes for 4.11-rc1

From: David Miller <davem@davemloft.net>
Date: 2017-02-23 15:44:20

From: Saeed Mahameed <redacted>
Date: Wed, 22 Feb 2017 17:20:10 +0200
This series includes some important bug fixes for mlx5e driver.

Three misc fixes:
From Mohamad, compilation fix on s390 system
From Me, A fix for driver unload when switchdev mode is on. 
From Tariq, HW LRO frag size optimization for when build_skb is not used
(striding RQ mode).

Three CQE compression related fixes:
Two fixes from Tariq and I, to correctly setup CQE compression
parameters on driver load and on arbitrary user modifications.
Last patch, fixes a very critical issue that was originally reported
by Tom, where the driver reported csum errors or even page ref issues
for when cqe compression is enabled and rapidly active.

For your convenience this series was generated on top of net-next branch:
005c3490e9db ('Revert "ath10k: Search SMBIOS for OEM board file extension"')
Series applied.
for -stable:
net/mlx5e: Register/unregister vport representors on interface (for kernel >= 4.9)
net/mlx5e: Do not reduce LRO WQE size when not using build_skb (for kernel >= 4.9)
net/mlx5e: Fix broken CQE compression initialization (for kernel >= 4.9)
net/mlx5e: Update MPWQE stride size when modifying CQE compress state (for kernel >= 4.7)
net/mlx5e: Fix wrong CQE decompression (for kernel >= 4.7)
Queued up, thanks.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help