* [net-next 01/16] net/mlx5e: Define one flow for TXQ selection when TCs are configured
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 02/16] net/mlx5e: Add missing LRO cap check Saeed Mahameed
` (15 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Eran Ben Elisha, Tariq Toukan, Saeed Mahameed
From: Eran Ben Elisha <eranbe@mellanox.com>
We shall always extract channel index out of the txq, regardless
of the relation between txq_ix and num channels. The extraction is
always valid, as if txq is smaller than number of channels,
txq_ix == priv->txq2sq[txq_ix]->ch_ix.
By doing so, we can remove an if clause from the select queue method,
and have one flow for all packets.
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_tx.c | 13 ++++++-------
1 file changed, 6 insertions(+), 7 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
index ee60383adc5b..fd6b2a1898c5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
@@ -72,8 +72,8 @@ u16 mlx5e_select_queue(struct net_device *dev, struct sk_buff *skb,
{
int txq_ix = netdev_pick_tx(dev, skb, NULL);
struct mlx5e_priv *priv = netdev_priv(dev);
- u16 num_channels;
int up = 0;
+ int ch_ix;
if (!netdev_get_num_tc(dev))
return txq_ix;
@@ -86,14 +86,13 @@ u16 mlx5e_select_queue(struct net_device *dev, struct sk_buff *skb,
if (skb_vlan_tag_present(skb))
up = skb_vlan_tag_get_prio(skb);
- /* txq_ix can be larger than num_channels since
- * dev->num_real_tx_queues = num_channels * num_tc
+ /* Normalize any picked txq_ix to [0, num_channels),
+ * So we can return a txq_ix that matches the channel and
+ * packet UP.
*/
- num_channels = priv->channels.params.num_channels;
- if (txq_ix >= num_channels)
- txq_ix = priv->txq2sq[txq_ix]->ch_ix;
+ ch_ix = priv->txq2sq[txq_ix]->ch_ix;
- return priv->channel_tc2realtxq[txq_ix][up];
+ return priv->channel_tc2realtxq[ch_ix][up];
}
static inline int mlx5e_skb_l2_header_offset(struct sk_buff *skb)
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 02/16] net/mlx5e: Add missing LRO cap check
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
2020-02-26 1:12 ` [net-next 01/16] net/mlx5e: Define one flow for TXQ selection when TCs are configured Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 03/16] net/mlx5e: Encapsulate updating netdev queues into a function Saeed Mahameed
` (14 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba; +Cc: netdev, Tariq Toukan, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
The LRO boolean state in params->lro_en must not be set in case
the NIC is not capable.
Enforce this check and remove the TODO comment.
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 966983674663..a4d3e1b6ab20 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -4770,9 +4770,8 @@ void mlx5e_build_nic_params(struct mlx5e_priv *priv,
mlx5e_build_rq_params(mdev, params);
/* HW LRO */
-
- /* TODO: && MLX5_CAP_ETH(mdev, lro_cap) */
- if (params->rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) {
+ if (MLX5_CAP_ETH(mdev, lro_cap) &&
+ params->rq_wq_type == MLX5_WQ_TYPE_LINKED_LIST_STRIDING_RQ) {
/* No XSK params: checking the availability of striding RQ in general. */
if (!mlx5e_rx_mpwqe_is_linear_skb(mdev, params, NULL))
params->lro_en = !slow_pci_heuristic(mdev);
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 03/16] net/mlx5e: Encapsulate updating netdev queues into a function
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
2020-02-26 1:12 ` [net-next 01/16] net/mlx5e: Define one flow for TXQ selection when TCs are configured Saeed Mahameed
2020-02-26 1:12 ` [net-next 02/16] net/mlx5e: Add missing LRO cap check Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:41 ` Jakub Kicinski
2020-02-26 1:12 ` [net-next 04/16] net/mlx5e: Rename hw_modify to preactivate Saeed Mahameed
` (13 subsequent siblings)
16 siblings, 1 reply; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
As a preparation for one of the following commits, create a function to
encapsulate the code that notifies the kernel about the new amount of
RX and TX queues. The code will be called multiple times in the next
commit.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../net/ethernet/mellanox/mlx5/core/en_main.c | 19 ++++++++++++-------
1 file changed, 12 insertions(+), 7 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index a4d3e1b6ab20..85a86ff72aac 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -2869,6 +2869,17 @@ static void mlx5e_netdev_set_tcs(struct net_device *netdev)
netdev_set_tc_queue(netdev, tc, nch, 0);
}
+static void mlx5e_update_netdev_queues(struct mlx5e_priv *priv)
+{
+ int num_txqs = priv->channels.num * priv->channels.params.num_tc;
+ int num_rxqs = priv->channels.num * priv->profile->rq_groups;
+ struct net_device *netdev = priv->netdev;
+
+ mlx5e_netdev_set_tcs(netdev);
+ netif_set_real_num_tx_queues(netdev, num_txqs);
+ netif_set_real_num_rx_queues(netdev, num_rxqs);
+}
+
static void mlx5e_build_txq_maps(struct mlx5e_priv *priv)
{
int i, ch;
@@ -2890,13 +2901,7 @@ static void mlx5e_build_txq_maps(struct mlx5e_priv *priv)
void mlx5e_activate_priv_channels(struct mlx5e_priv *priv)
{
- int num_txqs = priv->channels.num * priv->channels.params.num_tc;
- int num_rxqs = priv->channels.num * priv->profile->rq_groups;
- struct net_device *netdev = priv->netdev;
-
- mlx5e_netdev_set_tcs(netdev);
- netif_set_real_num_tx_queues(netdev, num_txqs);
- netif_set_real_num_rx_queues(netdev, num_rxqs);
+ mlx5e_update_netdev_queues(priv);
mlx5e_build_txq_maps(priv);
mlx5e_activate_channels(&priv->channels);
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* Re: [net-next 03/16] net/mlx5e: Encapsulate updating netdev queues into a function
2020-02-26 1:12 ` [net-next 03/16] net/mlx5e: Encapsulate updating netdev queues into a function Saeed Mahameed
@ 2020-02-26 1:41 ` Jakub Kicinski
2020-02-26 7:38 ` Saeed Mahameed
0 siblings, 1 reply; 20+ messages in thread
From: Jakub Kicinski @ 2020-02-26 1:41 UTC (permalink / raw)
To: Saeed Mahameed; +Cc: David S. Miller, netdev, Maxim Mikityanskiy, Tariq Toukan
On Tue, 25 Feb 2020 17:12:33 -0800 Saeed Mahameed wrote:
> From: Maxim Mikityanskiy <maximmi@mellanox.com>
>
> As a preparation for one of the following commits, create a function to
> encapsulate the code that notifies the kernel about the new amount of
> RX and TX queues. The code will be called multiple times in the next
> commit.
>
> Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
> Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
> ---
> .../net/ethernet/mellanox/mlx5/core/en_main.c | 19 ++++++++++++-------
> 1 file changed, 12 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> index a4d3e1b6ab20..85a86ff72aac 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> @@ -2869,6 +2869,17 @@ static void mlx5e_netdev_set_tcs(struct net_device *netdev)
> netdev_set_tc_queue(netdev, tc, nch, 0);
> }
>
> +static void mlx5e_update_netdev_queues(struct mlx5e_priv *priv)
> +{
> + int num_txqs = priv->channels.num * priv->channels.params.num_tc;
> + int num_rxqs = priv->channels.num * priv->profile->rq_groups;
> + struct net_device *netdev = priv->netdev;
> +
> + mlx5e_netdev_set_tcs(netdev);
> + netif_set_real_num_tx_queues(netdev, num_txqs);
> + netif_set_real_num_rx_queues(netdev, num_rxqs);
> +}
> +
> static void mlx5e_build_txq_maps(struct mlx5e_priv *priv)
> {
> int i, ch;
> @@ -2890,13 +2901,7 @@ static void mlx5e_build_txq_maps(struct mlx5e_priv *priv)
>
> void mlx5e_activate_priv_channels(struct mlx5e_priv *priv)
> {
> - int num_txqs = priv->channels.num * priv->channels.params.num_tc;
> - int num_rxqs = priv->channels.num * priv->profile->rq_groups;
> - struct net_device *netdev = priv->netdev;
> -
> - mlx5e_netdev_set_tcs(netdev);
> - netif_set_real_num_tx_queues(netdev, num_txqs);
> - netif_set_real_num_rx_queues(netdev, num_rxqs);
> + mlx5e_update_netdev_queues(priv);
Not sure where we stand on just moving bad code, but set_real_num_
_queues can fail, Dave just pointed this out to someone recently in
review.
>
> mlx5e_build_txq_maps(priv);
> mlx5e_activate_channels(&priv->channels);
^ permalink raw reply [flat|nested] 20+ messages in thread* Re: [net-next 03/16] net/mlx5e: Encapsulate updating netdev queues into a function
2020-02-26 1:41 ` Jakub Kicinski
@ 2020-02-26 7:38 ` Saeed Mahameed
0 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 7:38 UTC (permalink / raw)
To: kuba; +Cc: davem, netdev, Tariq Toukan, Maxim Mikityanskiy
On Tue, 2020-02-25 at 17:41 -0800, Jakub Kicinski wrote:
> On Tue, 25 Feb 2020 17:12:33 -0800 Saeed Mahameed wrote:
> > From: Maxim Mikityanskiy <maximmi@mellanox.com>
> >
> > As a preparation for one of the following commits, create a
> > function to
> > encapsulate the code that notifies the kernel about the new amount
> > of
> > RX and TX queues. The code will be called multiple times in the
> > next
> > commit.
> >
> > Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
> > Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
> > Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
> > ---
> > .../net/ethernet/mellanox/mlx5/core/en_main.c | 19 ++++++++++++---
> > ----
> > 1 file changed, 12 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> > b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> > index a4d3e1b6ab20..85a86ff72aac 100644
> > --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> > +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
> > @@ -2869,6 +2869,17 @@ static void mlx5e_netdev_set_tcs(struct
> > net_device *netdev)
> > netdev_set_tc_queue(netdev, tc, nch, 0);
> > }
> >
> > +static void mlx5e_update_netdev_queues(struct mlx5e_priv *priv)
> > +{
> > + int num_txqs = priv->channels.num * priv-
> > >channels.params.num_tc;
> > + int num_rxqs = priv->channels.num * priv->profile->rq_groups;
> > + struct net_device *netdev = priv->netdev;
> > +
> > + mlx5e_netdev_set_tcs(netdev);
> > + netif_set_real_num_tx_queues(netdev, num_txqs);
> > + netif_set_real_num_rx_queues(netdev, num_rxqs);
> > +}
> > +
> > static void mlx5e_build_txq_maps(struct mlx5e_priv *priv)
> > {
> > int i, ch;
> > @@ -2890,13 +2901,7 @@ static void mlx5e_build_txq_maps(struct
> > mlx5e_priv *priv)
> >
> > void mlx5e_activate_priv_channels(struct mlx5e_priv *priv)
> > {
> > - int num_txqs = priv->channels.num * priv-
> > >channels.params.num_tc;
> > - int num_rxqs = priv->channels.num * priv->profile->rq_groups;
> > - struct net_device *netdev = priv->netdev;
> > -
> > - mlx5e_netdev_set_tcs(netdev);
> > - netif_set_real_num_tx_queues(netdev, num_txqs);
> > - netif_set_real_num_rx_queues(netdev, num_rxqs);
> > + mlx5e_update_netdev_queues(priv);
>
> Not sure where we stand on just moving bad code, but set_real_num_
> _queues can fail, Dave just pointed this out to someone recently in
> review.
>
good point, but likewise not sure if this patch is the place to fix the
issue, the good thing is that now it is fixable since maxim added the
ability to allow failing mlx5_switch_channels() and roll back.
in patch 6 of this series;
net/mlx5e: Fix configuration of XPS cpumasks and netdev queues in
corner cases
you can see that Maxim is using the new function introduced in this
patch as a hook that is allowed to fail and bail out when switching
between channels:
err = mlx5e_safe_switch_channels(priv, &new_channels,
mlx5e_num_channels_changed);
so all we need is to not ignore the return value of
netif_set_real_num_tx_queues() and abort mlx5e_num_channels_changed()
in case of failure which will abort mlx5e_safe_switch_channels() and
old channels will keep working.
I guess this should be an incremental oneliner patch after this series.
Thanks,
Saeed.
> >
> > mlx5e_build_txq_maps(priv);
> > mlx5e_activate_channels(&priv->channels);
^ permalink raw reply [flat|nested] 20+ messages in thread
* [net-next 04/16] net/mlx5e: Rename hw_modify to preactivate
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (2 preceding siblings ...)
2020-02-26 1:12 ` [net-next 03/16] net/mlx5e: Encapsulate updating netdev queues into a function Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 05/16] net/mlx5e: Use preactivate hook to set the indirection table Saeed Mahameed
` (12 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
mlx5e_safe_switch_channels accepts a callback to be called before
activating new channels. It is intended to configure some hardware
parameters in cases where channels are recreated because some
configuration has changed.
Recently, this callback has started being used to update the driver's
internal MLX5E_STATE_XDP_OPEN flag, and the following patches also
intend to use this callback for software preparations. This patch
renames the hw_modify callback to preactivate, so that the name fits
better.
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 | 6 +++---
drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 14 ++++++++------
2 files changed, 11 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 220ef9f06f84..bc2c96b34de1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -1035,14 +1035,14 @@ int mlx5e_open_channels(struct mlx5e_priv *priv,
struct mlx5e_channels *chs);
void mlx5e_close_channels(struct mlx5e_channels *chs);
-/* Function pointer to be used to modify WH settings while
+/* Function pointer to be used to modify HW or kernel settings while
* switching channels
*/
-typedef int (*mlx5e_fp_hw_modify)(struct mlx5e_priv *priv);
+typedef int (*mlx5e_fp_preactivate)(struct mlx5e_priv *priv);
int mlx5e_safe_reopen_channels(struct mlx5e_priv *priv);
int mlx5e_safe_switch_channels(struct mlx5e_priv *priv,
struct mlx5e_channels *new_chs,
- mlx5e_fp_hw_modify hw_modify);
+ mlx5e_fp_preactivate preactivate);
void mlx5e_activate_priv_channels(struct mlx5e_priv *priv);
void mlx5e_deactivate_priv_channels(struct mlx5e_priv *priv);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 85a86ff72aac..152aa5d7df79 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -2937,7 +2937,7 @@ void mlx5e_deactivate_priv_channels(struct mlx5e_priv *priv)
static void mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
struct mlx5e_channels *new_chs,
- mlx5e_fp_hw_modify hw_modify)
+ mlx5e_fp_preactivate preactivate)
{
struct net_device *netdev = priv->netdev;
int new_num_txqs;
@@ -2956,9 +2956,11 @@ static void mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
priv->channels = *new_chs;
- /* New channels are ready to roll, modify HW settings if needed */
- if (hw_modify)
- hw_modify(priv);
+ /* New channels are ready to roll, call the preactivate hook if needed
+ * to modify HW settings or update kernel parameters.
+ */
+ if (preactivate)
+ preactivate(priv);
priv->profile->update_rx(priv);
mlx5e_activate_priv_channels(priv);
@@ -2970,7 +2972,7 @@ static void mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
int mlx5e_safe_switch_channels(struct mlx5e_priv *priv,
struct mlx5e_channels *new_chs,
- mlx5e_fp_hw_modify hw_modify)
+ mlx5e_fp_preactivate preactivate)
{
int err;
@@ -2978,7 +2980,7 @@ int mlx5e_safe_switch_channels(struct mlx5e_priv *priv,
if (err)
return err;
- mlx5e_switch_priv_channels(priv, new_chs, hw_modify);
+ mlx5e_switch_priv_channels(priv, new_chs, preactivate);
return 0;
}
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 05/16] net/mlx5e: Use preactivate hook to set the indirection table
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (3 preceding siblings ...)
2020-02-26 1:12 ` [net-next 04/16] net/mlx5e: Rename hw_modify to preactivate Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 06/16] net/mlx5e: Fix configuration of XPS cpumasks and netdev queues in corner cases Saeed Mahameed
` (11 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
mlx5e_ethtool_set_channels updates the indirection table before
switching to the new channels. If the switch fails, the indirection
table is new, but the channels are old, which is wrong. Fix it by using
the preactivate hook of mlx5e_safe_switch_channels to update the
indirection table at the stage when nothing can fail anymore.
As the code that updates the indirection table is now encapsulated into
a new function, use that function in the attach flow when the driver has
to reduce the number of channels, and prepare the code for the next
commit.
Fixes: 85082dba0a ("net/mlx5e: Correctly handle RSS indirection table when changing number of channels")
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 | 1 +
.../net/ethernet/mellanox/mlx5/core/en_ethtool.c | 10 ++--------
.../net/ethernet/mellanox/mlx5/core/en_main.c | 16 ++++++++++++++--
3 files changed, 17 insertions(+), 10 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index bc2c96b34de1..4ddccab02a4b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -1043,6 +1043,7 @@ int mlx5e_safe_reopen_channels(struct mlx5e_priv *priv);
int mlx5e_safe_switch_channels(struct mlx5e_priv *priv,
struct mlx5e_channels *new_chs,
mlx5e_fp_preactivate preactivate);
+int mlx5e_num_channels_changed(struct mlx5e_priv *priv);
void mlx5e_activate_priv_channels(struct mlx5e_priv *priv);
void mlx5e_deactivate_priv_channels(struct mlx5e_priv *priv);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index 68b520df07e4..ff7f5a931520 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -432,9 +432,7 @@ int mlx5e_ethtool_set_channels(struct mlx5e_priv *priv,
if (!test_bit(MLX5E_STATE_OPENED, &priv->state)) {
*cur_params = new_channels.params;
- if (!netif_is_rxfh_configured(priv->netdev))
- mlx5e_build_default_indir_rqt(priv->rss_params.indirection_rqt,
- MLX5E_INDIR_RQT_SIZE, count);
+ mlx5e_num_channels_changed(priv);
goto out;
}
@@ -442,12 +440,8 @@ int mlx5e_ethtool_set_channels(struct mlx5e_priv *priv,
if (arfs_enabled)
mlx5e_arfs_disable(priv);
- if (!netif_is_rxfh_configured(priv->netdev))
- mlx5e_build_default_indir_rqt(priv->rss_params.indirection_rqt,
- MLX5E_INDIR_RQT_SIZE, count);
-
/* Switch to new channels, set new parameters and close old ones */
- err = mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ err = mlx5e_safe_switch_channels(priv, &new_channels, mlx5e_num_channels_changed);
if (arfs_enabled) {
int err2 = mlx5e_arfs_enable(priv);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 152aa5d7df79..bbe8c32fb423 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -2880,6 +2880,17 @@ static void mlx5e_update_netdev_queues(struct mlx5e_priv *priv)
netif_set_real_num_rx_queues(netdev, num_rxqs);
}
+int mlx5e_num_channels_changed(struct mlx5e_priv *priv)
+{
+ u16 count = priv->channels.params.num_channels;
+
+ if (!netif_is_rxfh_configured(priv->netdev))
+ mlx5e_build_default_indir_rqt(priv->rss_params.indirection_rqt,
+ MLX5E_INDIR_RQT_SIZE, count);
+
+ return 0;
+}
+
static void mlx5e_build_txq_maps(struct mlx5e_priv *priv)
{
int i, ch;
@@ -5288,9 +5299,10 @@ int mlx5e_attach_netdev(struct mlx5e_priv *priv)
max_nch = mlx5e_get_max_num_channels(priv->mdev);
if (priv->channels.params.num_channels > max_nch) {
mlx5_core_warn(priv->mdev, "MLX5E: Reducing number of channels to %d\n", max_nch);
+ /* Reducing the number of channels - RXFH has to be reset. */
+ priv->netdev->priv_flags &= ~IFF_RXFH_CONFIGURED;
priv->channels.params.num_channels = max_nch;
- mlx5e_build_default_indir_rqt(priv->rss_params.indirection_rqt,
- MLX5E_INDIR_RQT_SIZE, max_nch);
+ mlx5e_num_channels_changed(priv);
}
err = profile->init_tx(priv);
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 06/16] net/mlx5e: Fix configuration of XPS cpumasks and netdev queues in corner cases
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (4 preceding siblings ...)
2020-02-26 1:12 ` [net-next 05/16] net/mlx5e: Use preactivate hook to set the indirection table Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 07/16] net/mlx5e: Remove unneeded netif_set_real_num_tx_queues Saeed Mahameed
` (10 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
Currently, mlx5e notifies the kernel about the number of queues and sets
the default XPS cpumasks when channels are activated. This
implementation has several corner cases, in which the kernel may not be
updated on time, or XPS cpumasks may be reset when not directly touched
by the user.
This commit fixes these corner cases to match the following expected
behavior:
1. The number of queues always corresponds to the number of channels
configured.
2. XPS cpumasks are set to driver's defaults on netdev attach.
3. XPS cpumasks set by user are not reset, unless the number of channels
changes. If the number of channels changes, they are reset to driver's
defaults. (In general case, when the number of channels increases or
decreases, it's not possible to guess how to convert the current XPS
cpumasks to work with the new number of channels, so we let the user
reconfigure it if they change the number of channels.)
XPS cpumasks are no longer stored per channel. Only one temporary
cpumask is used. The old stored cpumasks didn't reflect the user's
changes and were not used after applying them.
A scratchpad area is added to struct mlx5e_priv. As cpumask_var_t
requires allocation, and the preactivate hook can't fail, we need to
preallocate the temporary cpumask in advance. It's stored in the
scratchpad.
Fixes: 149e566fef81 ("net/mlx5e: Expand XPS cpumask to cover all online cpus")
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 | 11 ++-
.../net/ethernet/mellanox/mlx5/core/en_main.c | 95 +++++++++++--------
2 files changed, 65 insertions(+), 41 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 4ddccab02a4b..6d725d2acd3d 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -737,7 +737,6 @@ struct mlx5e_channel {
DECLARE_BITMAP(state, MLX5E_CHANNEL_NUM_STATES);
int ix;
int cpu;
- cpumask_var_t xps_cpumask;
};
struct mlx5e_channels {
@@ -813,6 +812,15 @@ struct mlx5e_xsk {
bool ever_used;
};
+/* Temporary storage for variables that are allocated when struct mlx5e_priv is
+ * initialized, and used where we can't allocate them because that functions
+ * must not fail. Use with care and make sure the same variable is not used
+ * simultaneously by multiple users.
+ */
+struct mlx5e_scratchpad {
+ cpumask_var_t cpumask;
+};
+
struct mlx5e_priv {
/* priv data path fields - start */
struct mlx5e_txqsq *txq2sq[MLX5E_MAX_NUM_CHANNELS * MLX5E_MAX_NUM_TC];
@@ -876,6 +884,7 @@ struct mlx5e_priv {
#if IS_ENABLED(CONFIG_PCI_HYPERV_INTERFACE)
struct mlx5e_hv_vhca_stats_agent stats_agent;
#endif
+ struct mlx5e_scratchpad scratchpad;
};
struct mlx5e_profile {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index bbe8c32fb423..4906d609aa55 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -1794,29 +1794,6 @@ static int mlx5e_set_tx_maxrate(struct net_device *dev, int index, u32 rate)
return err;
}
-static int mlx5e_alloc_xps_cpumask(struct mlx5e_channel *c,
- struct mlx5e_params *params)
-{
- int num_comp_vectors = mlx5_comp_vectors_count(c->mdev);
- int irq;
-
- if (!zalloc_cpumask_var(&c->xps_cpumask, GFP_KERNEL))
- return -ENOMEM;
-
- for (irq = c->ix; irq < num_comp_vectors; irq += params->num_channels) {
- int cpu = cpumask_first(mlx5_comp_irq_get_affinity_mask(c->mdev, irq));
-
- cpumask_set_cpu(cpu, c->xps_cpumask);
- }
-
- return 0;
-}
-
-static void mlx5e_free_xps_cpumask(struct mlx5e_channel *c)
-{
- free_cpumask_var(c->xps_cpumask);
-}
-
static int mlx5e_open_queues(struct mlx5e_channel *c,
struct mlx5e_params *params,
struct mlx5e_channel_param *cparam)
@@ -1967,10 +1944,6 @@ static int mlx5e_open_channel(struct mlx5e_priv *priv, int ix,
c->irq_desc = irq_to_desc(irq);
c->lag_port = mlx5e_enumerate_lag_port(priv->mdev, ix);
- err = mlx5e_alloc_xps_cpumask(c, params);
- if (err)
- goto err_free_channel;
-
netif_napi_add(netdev, &c->napi, mlx5e_napi_poll, 64);
err = mlx5e_open_queues(c, params, cparam);
@@ -1993,9 +1966,7 @@ static int mlx5e_open_channel(struct mlx5e_priv *priv, int ix,
err_napi_del:
netif_napi_del(&c->napi);
- mlx5e_free_xps_cpumask(c);
-err_free_channel:
kvfree(c);
return err;
@@ -2009,7 +1980,6 @@ static void mlx5e_activate_channel(struct mlx5e_channel *c)
mlx5e_activate_txqsq(&c->sq[tc]);
mlx5e_activate_icosq(&c->icosq);
mlx5e_activate_rq(&c->rq);
- netif_set_xps_queue(c->netdev, c->xps_cpumask, c->ix);
if (test_bit(MLX5E_CHANNEL_STATE_XSK, c->state))
mlx5e_activate_xsk(c);
@@ -2034,7 +2004,6 @@ static void mlx5e_close_channel(struct mlx5e_channel *c)
mlx5e_close_xsk(c);
mlx5e_close_queues(c);
netif_napi_del(&c->napi);
- mlx5e_free_xps_cpumask(c);
kvfree(c);
}
@@ -2869,10 +2838,10 @@ static void mlx5e_netdev_set_tcs(struct net_device *netdev)
netdev_set_tc_queue(netdev, tc, nch, 0);
}
-static void mlx5e_update_netdev_queues(struct mlx5e_priv *priv)
+static void mlx5e_update_netdev_queues(struct mlx5e_priv *priv, u16 count)
{
- int num_txqs = priv->channels.num * priv->channels.params.num_tc;
- int num_rxqs = priv->channels.num * priv->profile->rq_groups;
+ int num_txqs = count * priv->channels.params.num_tc;
+ int num_rxqs = count * priv->profile->rq_groups;
struct net_device *netdev = priv->netdev;
mlx5e_netdev_set_tcs(netdev);
@@ -2880,10 +2849,34 @@ static void mlx5e_update_netdev_queues(struct mlx5e_priv *priv)
netif_set_real_num_rx_queues(netdev, num_rxqs);
}
+static void mlx5e_set_default_xps_cpumasks(struct mlx5e_priv *priv,
+ struct mlx5e_params *params)
+{
+ struct mlx5_core_dev *mdev = priv->mdev;
+ int num_comp_vectors, ix, irq;
+
+ num_comp_vectors = mlx5_comp_vectors_count(mdev);
+
+ for (ix = 0; ix < params->num_channels; ix++) {
+ cpumask_clear(priv->scratchpad.cpumask);
+
+ for (irq = ix; irq < num_comp_vectors; irq += params->num_channels) {
+ int cpu = cpumask_first(mlx5_comp_irq_get_affinity_mask(mdev, irq));
+
+ cpumask_set_cpu(cpu, priv->scratchpad.cpumask);
+ }
+
+ netif_set_xps_queue(priv->netdev, priv->scratchpad.cpumask, ix);
+ }
+}
+
int mlx5e_num_channels_changed(struct mlx5e_priv *priv)
{
u16 count = priv->channels.params.num_channels;
+ mlx5e_update_netdev_queues(priv, count);
+ mlx5e_set_default_xps_cpumasks(priv, &priv->channels.params);
+
if (!netif_is_rxfh_configured(priv->netdev))
mlx5e_build_default_indir_rqt(priv->rss_params.indirection_rqt,
MLX5E_INDIR_RQT_SIZE, count);
@@ -2912,8 +2905,6 @@ static void mlx5e_build_txq_maps(struct mlx5e_priv *priv)
void mlx5e_activate_priv_channels(struct mlx5e_priv *priv)
{
- mlx5e_update_netdev_queues(priv);
-
mlx5e_build_txq_maps(priv);
mlx5e_activate_channels(&priv->channels);
mlx5e_xdp_tx_enable(priv);
@@ -3449,7 +3440,7 @@ static int mlx5e_setup_tc_mqprio(struct mlx5e_priv *priv,
goto out;
}
- err = mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ err = mlx5e_safe_switch_channels(priv, &new_channels, mlx5e_num_channels_changed);
if (err)
goto out;
@@ -5231,6 +5222,9 @@ int mlx5e_netdev_init(struct net_device *netdev,
priv->max_nch = netdev->num_rx_queues / max_t(u8, profile->rq_groups, 1);
priv->max_opened_tc = 1;
+ if (!alloc_cpumask_var(&priv->scratchpad.cpumask, GFP_KERNEL))
+ return -ENOMEM;
+
mutex_init(&priv->state_lock);
INIT_WORK(&priv->update_carrier_work, mlx5e_update_carrier_work);
INIT_WORK(&priv->set_rx_mode_work, mlx5e_set_rx_mode_work);
@@ -5239,7 +5233,7 @@ int mlx5e_netdev_init(struct net_device *netdev,
priv->wq = create_singlethread_workqueue("mlx5e");
if (!priv->wq)
- return -ENOMEM;
+ goto err_free_cpumask;
/* netdev init */
netif_carrier_off(netdev);
@@ -5249,11 +5243,17 @@ int mlx5e_netdev_init(struct net_device *netdev,
#endif
return 0;
+
+err_free_cpumask:
+ free_cpumask_var(priv->scratchpad.cpumask);
+
+ return -ENOMEM;
}
void mlx5e_netdev_cleanup(struct net_device *netdev, struct mlx5e_priv *priv)
{
destroy_workqueue(priv->wq);
+ free_cpumask_var(priv->scratchpad.cpumask);
}
struct net_device *mlx5e_create_netdev(struct mlx5_core_dev *mdev,
@@ -5288,6 +5288,7 @@ struct net_device *mlx5e_create_netdev(struct mlx5_core_dev *mdev,
int mlx5e_attach_netdev(struct mlx5e_priv *priv)
{
+ const bool take_rtnl = priv->netdev->reg_state == NETREG_REGISTERED;
const struct mlx5e_profile *profile;
int max_nch;
int err;
@@ -5299,11 +5300,25 @@ int mlx5e_attach_netdev(struct mlx5e_priv *priv)
max_nch = mlx5e_get_max_num_channels(priv->mdev);
if (priv->channels.params.num_channels > max_nch) {
mlx5_core_warn(priv->mdev, "MLX5E: Reducing number of channels to %d\n", max_nch);
- /* Reducing the number of channels - RXFH has to be reset. */
+ /* Reducing the number of channels - RXFH has to be reset, and
+ * mlx5e_num_channels_changed below will build the RQT.
+ */
priv->netdev->priv_flags &= ~IFF_RXFH_CONFIGURED;
priv->channels.params.num_channels = max_nch;
- mlx5e_num_channels_changed(priv);
}
+ /* 1. Set the real number of queues in the kernel the first time.
+ * 2. Set our default XPS cpumask.
+ * 3. Build the RQT.
+ *
+ * rtnl_lock is required by netif_set_real_num_*_queues in case the
+ * netdev has been registered by this point (if this function was called
+ * in the reload or resume flow).
+ */
+ if (take_rtnl)
+ rtnl_lock();
+ mlx5e_num_channels_changed(priv);
+ if (take_rtnl)
+ rtnl_unlock();
err = profile->init_tx(priv);
if (err)
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 07/16] net/mlx5e: Remove unneeded netif_set_real_num_tx_queues
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (5 preceding siblings ...)
2020-02-26 1:12 ` [net-next 06/16] net/mlx5e: Fix configuration of XPS cpumasks and netdev queues in corner cases Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 08/16] net/mlx5e: Allow mlx5e_switch_priv_channels to fail and recover Saeed Mahameed
` (9 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
The number of queues is now updated by mlx5e_update_netdev_queues in a
centralized way, when no channels are active. Remove an extra occurrence
of netif_set_real_num_tx_queues to prepare it for the next commit.
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_main.c | 6 ------
1 file changed, 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 4906d609aa55..ceeb9faad9ef 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -2942,17 +2942,11 @@ static void mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
mlx5e_fp_preactivate preactivate)
{
struct net_device *netdev = priv->netdev;
- int new_num_txqs;
int carrier_ok;
- new_num_txqs = new_chs->num * new_chs->params.num_tc;
-
carrier_ok = netif_carrier_ok(netdev);
netif_carrier_off(netdev);
- if (new_num_txqs < netdev->real_num_tx_queues)
- netif_set_real_num_tx_queues(netdev, new_num_txqs);
-
mlx5e_deactivate_priv_channels(priv);
mlx5e_close_channels(&priv->channels);
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 08/16] net/mlx5e: Allow mlx5e_switch_priv_channels to fail and recover
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (6 preceding siblings ...)
2020-02-26 1:12 ` [net-next 07/16] net/mlx5e: Remove unneeded netif_set_real_num_tx_queues Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 09/16] net/mlx5e: Add context to the preactivate hook Saeed Mahameed
` (8 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
Currently mlx5e_switch_priv_channels expects that the preactivate hook
doesn't fail, however, it can fail, because it may set hardware
parameters. This commit addresses this issue and provides a way to
recover from failures of the preactivate hook: the old channels are not
closed until the point where nothing can fail anymore, so in case
preactivate fails, the driver can roll back the old channels and
activate them again.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
.../net/ethernet/mellanox/mlx5/core/en_main.c | 34 +++++++++++++++----
1 file changed, 27 insertions(+), 7 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index ceeb9faad9ef..0a71fe85d21e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -2937,33 +2937,45 @@ void mlx5e_deactivate_priv_channels(struct mlx5e_priv *priv)
mlx5e_deactivate_channels(&priv->channels);
}
-static void mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
- struct mlx5e_channels *new_chs,
- mlx5e_fp_preactivate preactivate)
+static int mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
+ struct mlx5e_channels *new_chs,
+ mlx5e_fp_preactivate preactivate)
{
struct net_device *netdev = priv->netdev;
+ struct mlx5e_channels old_chs;
int carrier_ok;
+ int err = 0;
carrier_ok = netif_carrier_ok(netdev);
netif_carrier_off(netdev);
mlx5e_deactivate_priv_channels(priv);
- mlx5e_close_channels(&priv->channels);
+ old_chs = priv->channels;
priv->channels = *new_chs;
/* New channels are ready to roll, call the preactivate hook if needed
* to modify HW settings or update kernel parameters.
*/
- if (preactivate)
- preactivate(priv);
+ if (preactivate) {
+ err = preactivate(priv);
+ if (err) {
+ priv->channels = old_chs;
+ goto out;
+ }
+ }
+ mlx5e_close_channels(&old_chs);
priv->profile->update_rx(priv);
+
+out:
mlx5e_activate_priv_channels(priv);
/* return carrier back if needed */
if (carrier_ok)
netif_carrier_on(netdev);
+
+ return err;
}
int mlx5e_safe_switch_channels(struct mlx5e_priv *priv,
@@ -2976,8 +2988,16 @@ int mlx5e_safe_switch_channels(struct mlx5e_priv *priv,
if (err)
return err;
- mlx5e_switch_priv_channels(priv, new_chs, preactivate);
+ err = mlx5e_switch_priv_channels(priv, new_chs, preactivate);
+ if (err)
+ goto err_close;
+
return 0;
+
+err_close:
+ mlx5e_close_channels(new_chs);
+
+ return err;
}
int mlx5e_safe_reopen_channels(struct mlx5e_priv *priv)
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 09/16] net/mlx5e: Add context to the preactivate hook
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (7 preceding siblings ...)
2020-02-26 1:12 ` [net-next 08/16] net/mlx5e: Allow mlx5e_switch_priv_channels to fail and recover Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 10/16] net/mlx5e: Change inline mode correctly when changing trust state Saeed Mahameed
` (7 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
Sometimes the preactivate hook of mlx5e_safe_switch_channels needs more
parameters than just struct mlx5e_priv *. For such cases, a new
parameter (void *context) is added to preactivate hooks.
Some of the existing normal functions are currently used as preactivate
callbacks. To avoid adding an extra unused parameter, they are wrapped
in an automatic way using the MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX
macro.
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 | 15 +++++---
.../ethernet/mellanox/mlx5/core/en_dcbnl.c | 2 +-
.../ethernet/mellanox/mlx5/core/en_ethtool.c | 15 ++++----
.../net/ethernet/mellanox/mlx5/core/en_main.c | 36 ++++++++++++-------
.../net/ethernet/mellanox/mlx5/core/en_rep.c | 2 +-
.../ethernet/mellanox/mlx5/core/ipoib/ipoib.c | 2 +-
6 files changed, 45 insertions(+), 27 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 6d725d2acd3d..3cc439ab3253 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -1047,12 +1047,19 @@ void mlx5e_close_channels(struct mlx5e_channels *chs);
/* Function pointer to be used to modify HW or kernel settings while
* switching channels
*/
-typedef int (*mlx5e_fp_preactivate)(struct mlx5e_priv *priv);
+typedef int (*mlx5e_fp_preactivate)(struct mlx5e_priv *priv, void *context);
+#define MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX(fn) \
+int fn##_ctx(struct mlx5e_priv *priv, void *context) \
+{ \
+ return fn(priv); \
+}
int mlx5e_safe_reopen_channels(struct mlx5e_priv *priv);
int mlx5e_safe_switch_channels(struct mlx5e_priv *priv,
struct mlx5e_channels *new_chs,
- mlx5e_fp_preactivate preactivate);
+ mlx5e_fp_preactivate preactivate,
+ void *context);
int mlx5e_num_channels_changed(struct mlx5e_priv *priv);
+int mlx5e_num_channels_changed_ctx(struct mlx5e_priv *priv, void *context);
void mlx5e_activate_priv_channels(struct mlx5e_priv *priv);
void mlx5e_deactivate_priv_channels(struct mlx5e_priv *priv);
@@ -1132,10 +1139,10 @@ void mlx5e_update_ndo_stats(struct mlx5e_priv *priv);
void mlx5e_queue_update_stats(struct mlx5e_priv *priv);
int mlx5e_bits_invert(unsigned long a, int size);
-typedef int (*change_hw_mtu_cb)(struct mlx5e_priv *priv);
int mlx5e_set_dev_port_mtu(struct mlx5e_priv *priv);
+int mlx5e_set_dev_port_mtu_ctx(struct mlx5e_priv *priv, void *context);
int mlx5e_change_mtu(struct net_device *netdev, int new_mtu,
- change_hw_mtu_cb set_mtu_cb);
+ mlx5e_fp_preactivate preactivate);
/* ethtool helpers */
void mlx5e_ethtool_get_drvinfo(struct mlx5e_priv *priv,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c b/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
index 01f2918063af..1375f6483a13 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
@@ -1126,7 +1126,7 @@ static void mlx5e_trust_update_sq_inline_mode(struct mlx5e_priv *priv)
priv->channels.params.tx_min_inline_mode)
goto out;
- mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
out:
mutex_unlock(&priv->state_lock);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
index ff7f5a931520..06f6f08ff5eb 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c
@@ -357,7 +357,7 @@ int mlx5e_ethtool_set_ringparam(struct mlx5e_priv *priv,
goto unlock;
}
- err = mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ err = mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
unlock:
mutex_unlock(&priv->state_lock);
@@ -441,7 +441,8 @@ int mlx5e_ethtool_set_channels(struct mlx5e_priv *priv,
mlx5e_arfs_disable(priv);
/* Switch to new channels, set new parameters and close old ones */
- err = mlx5e_safe_switch_channels(priv, &new_channels, mlx5e_num_channels_changed);
+ err = mlx5e_safe_switch_channels(priv, &new_channels,
+ mlx5e_num_channels_changed_ctx, NULL);
if (arfs_enabled) {
int err2 = mlx5e_arfs_enable(priv);
@@ -574,7 +575,7 @@ int mlx5e_ethtool_set_coalesce(struct mlx5e_priv *priv,
goto out;
}
- err = mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ err = mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
out:
mutex_unlock(&priv->state_lock);
@@ -1742,7 +1743,7 @@ static int set_pflag_cqe_based_moder(struct net_device *netdev, bool enable,
return 0;
}
- return mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ return mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
}
static int set_pflag_tx_cqe_based_moder(struct net_device *netdev, bool enable)
@@ -1775,7 +1776,7 @@ int mlx5e_modify_rx_cqe_compression_locked(struct mlx5e_priv *priv, bool new_val
return 0;
}
- err = mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ err = mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
if (err)
return err;
@@ -1832,7 +1833,7 @@ static int set_pflag_rx_striding_rq(struct net_device *netdev, bool enable)
return 0;
}
- return mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ return mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
}
static int set_pflag_rx_no_csum_complete(struct net_device *netdev, bool enable)
@@ -1876,7 +1877,7 @@ static int set_pflag_xdp_tx_mpwqe(struct net_device *netdev, bool enable)
return 0;
}
- err = mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ err = mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
return err;
}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 0a71fe85d21e..d29e53c023f1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -2753,6 +2753,8 @@ static int mlx5e_modify_tirs_lro(struct mlx5e_priv *priv)
return err;
}
+static MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX(mlx5e_modify_tirs_lro);
+
static int mlx5e_set_mtu(struct mlx5_core_dev *mdev,
struct mlx5e_params *params, u16 mtu)
{
@@ -2802,6 +2804,8 @@ int mlx5e_set_dev_port_mtu(struct mlx5e_priv *priv)
return 0;
}
+MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX(mlx5e_set_dev_port_mtu);
+
void mlx5e_set_netdev_mtu_boundaries(struct mlx5e_priv *priv)
{
struct mlx5e_params *params = &priv->channels.params;
@@ -2884,6 +2888,8 @@ int mlx5e_num_channels_changed(struct mlx5e_priv *priv)
return 0;
}
+MLX5E_DEFINE_PREACTIVATE_WRAPPER_CTX(mlx5e_num_channels_changed);
+
static void mlx5e_build_txq_maps(struct mlx5e_priv *priv)
{
int i, ch;
@@ -2939,7 +2945,8 @@ void mlx5e_deactivate_priv_channels(struct mlx5e_priv *priv)
static int mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
struct mlx5e_channels *new_chs,
- mlx5e_fp_preactivate preactivate)
+ mlx5e_fp_preactivate preactivate,
+ void *context)
{
struct net_device *netdev = priv->netdev;
struct mlx5e_channels old_chs;
@@ -2958,7 +2965,7 @@ static int mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
* to modify HW settings or update kernel parameters.
*/
if (preactivate) {
- err = preactivate(priv);
+ err = preactivate(priv, context);
if (err) {
priv->channels = old_chs;
goto out;
@@ -2980,7 +2987,8 @@ static int mlx5e_switch_priv_channels(struct mlx5e_priv *priv,
int mlx5e_safe_switch_channels(struct mlx5e_priv *priv,
struct mlx5e_channels *new_chs,
- mlx5e_fp_preactivate preactivate)
+ mlx5e_fp_preactivate preactivate,
+ void *context)
{
int err;
@@ -2988,7 +2996,7 @@ int mlx5e_safe_switch_channels(struct mlx5e_priv *priv,
if (err)
return err;
- err = mlx5e_switch_priv_channels(priv, new_chs, preactivate);
+ err = mlx5e_switch_priv_channels(priv, new_chs, preactivate, context);
if (err)
goto err_close;
@@ -3005,7 +3013,7 @@ int mlx5e_safe_reopen_channels(struct mlx5e_priv *priv)
struct mlx5e_channels new_channels = {};
new_channels.params = priv->channels.params;
- return mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ return mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
}
void mlx5e_timestamp_init(struct mlx5e_priv *priv)
@@ -3454,7 +3462,8 @@ static int mlx5e_setup_tc_mqprio(struct mlx5e_priv *priv,
goto out;
}
- err = mlx5e_safe_switch_channels(priv, &new_channels, mlx5e_num_channels_changed);
+ err = mlx5e_safe_switch_channels(priv, &new_channels,
+ mlx5e_num_channels_changed_ctx, NULL);
if (err)
goto out;
@@ -3667,7 +3676,8 @@ static int set_feature_lro(struct net_device *netdev, bool enable)
goto out;
}
- err = mlx5e_safe_switch_channels(priv, &new_channels, mlx5e_modify_tirs_lro);
+ err = mlx5e_safe_switch_channels(priv, &new_channels,
+ mlx5e_modify_tirs_lro_ctx, NULL);
out:
mutex_unlock(&priv->state_lock);
return err;
@@ -3886,7 +3896,7 @@ static bool mlx5e_xsk_validate_mtu(struct net_device *netdev,
}
int mlx5e_change_mtu(struct net_device *netdev, int new_mtu,
- change_hw_mtu_cb set_mtu_cb)
+ mlx5e_fp_preactivate preactivate)
{
struct mlx5e_priv *priv = netdev_priv(netdev);
struct mlx5e_channels new_channels = {};
@@ -3935,13 +3945,13 @@ int mlx5e_change_mtu(struct net_device *netdev, int new_mtu,
if (!reset) {
params->sw_mtu = new_mtu;
- if (set_mtu_cb)
- set_mtu_cb(priv);
+ if (preactivate)
+ preactivate(priv, NULL);
netdev->mtu = params->sw_mtu;
goto out;
}
- err = mlx5e_safe_switch_channels(priv, &new_channels, set_mtu_cb);
+ err = mlx5e_safe_switch_channels(priv, &new_channels, preactivate, NULL);
if (err)
goto out;
@@ -3954,7 +3964,7 @@ int mlx5e_change_mtu(struct net_device *netdev, int new_mtu,
static int mlx5e_change_nic_mtu(struct net_device *netdev, int new_mtu)
{
- return mlx5e_change_mtu(netdev, new_mtu, mlx5e_set_dev_port_mtu);
+ return mlx5e_change_mtu(netdev, new_mtu, mlx5e_set_dev_port_mtu_ctx);
}
int mlx5e_hwstamp_set(struct mlx5e_priv *priv, struct ifreq *ifr)
@@ -4415,7 +4425,7 @@ static int mlx5e_xdp_set(struct net_device *netdev, struct bpf_prog *prog)
mlx5e_set_rq_type(priv->mdev, &new_channels.params);
old_prog = priv->channels.params.xdp_prog;
- err = mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ err = mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
if (err)
goto unlock;
} else {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
index 7b48ccacebe2..6be85a6b11d4 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c
@@ -1396,7 +1396,7 @@ static int mlx5e_rep_change_mtu(struct net_device *netdev, int new_mtu)
static int mlx5e_uplink_rep_change_mtu(struct net_device *netdev, int new_mtu)
{
- return mlx5e_change_mtu(netdev, new_mtu, mlx5e_set_dev_port_mtu);
+ return mlx5e_change_mtu(netdev, new_mtu, mlx5e_set_dev_port_mtu_ctx);
}
static int mlx5e_uplink_rep_set_mac(struct net_device *netdev, void *addr)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
index 56078b23f1a0..673aaa815f57 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/ipoib/ipoib.c
@@ -483,7 +483,7 @@ static int mlx5i_change_mtu(struct net_device *netdev, int new_mtu)
new_channels.params = *params;
new_channels.params.sw_mtu = new_mtu;
- err = mlx5e_safe_switch_channels(priv, &new_channels, NULL);
+ err = mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
if (err)
goto out;
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 10/16] net/mlx5e: Change inline mode correctly when changing trust state
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (8 preceding siblings ...)
2020-02-26 1:12 ` [net-next 09/16] net/mlx5e: Add context to the preactivate hook Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 11/16] net/mlx5e: RX, Use indirect calls wrapper for posting descriptors Saeed Mahameed
` (6 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Maxim Mikityanskiy, Tariq Toukan, Saeed Mahameed
From: Maxim Mikityanskiy <maximmi@mellanox.com>
The current steps that are performed when the trust state changes, if
the channels are active:
1. The trust state is changed in hardware.
2. The new inline mode is calculated.
3. If the new inline mode is different, the channels are recreated using
the new inline mode.
This approach has some issues:
1. There is a time gap between changing trust state in hardware and
starting sending enough inline headers (the latter happens after
recreation of channels). It leads to failed transmissions and error
CQEs.
2. If the new channels fail to open, we'll be left with the old ones,
but the hardware will be configured for the new trust state, so the
interval when we can see TX errors never ends.
This patch fixes the issues above by moving the trust state change into
the preactivate hook that runs during the recreation of the channels
when no channels are active, so it eliminates the gap of partially
applied configuration. If the inline mode doesn't change with the change
of the trust state, the channels won't be recreated, just like before
this patch.
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_dcbnl.c | 55 +++++++++++--------
1 file changed, 33 insertions(+), 22 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c b/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
index 1375f6483a13..47874d34156b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_dcbnl.c
@@ -1098,49 +1098,59 @@ void mlx5e_dcbnl_delete_app(struct mlx5e_priv *priv)
mlx5e_dcbnl_dscp_app(priv, DELETE);
}
-static void mlx5e_trust_update_tx_min_inline_mode(struct mlx5e_priv *priv,
- struct mlx5e_params *params)
+static void mlx5e_params_calc_trust_tx_min_inline_mode(struct mlx5_core_dev *mdev,
+ struct mlx5e_params *params,
+ u8 trust_state)
{
- mlx5_query_min_inline(priv->mdev, ¶ms->tx_min_inline_mode);
- if (priv->dcbx_dp.trust_state == MLX5_QPTS_TRUST_DSCP &&
+ mlx5_query_min_inline(mdev, ¶ms->tx_min_inline_mode);
+ if (trust_state == MLX5_QPTS_TRUST_DSCP &&
params->tx_min_inline_mode == MLX5_INLINE_MODE_L2)
params->tx_min_inline_mode = MLX5_INLINE_MODE_IP;
}
-static void mlx5e_trust_update_sq_inline_mode(struct mlx5e_priv *priv)
+static int mlx5e_update_trust_state_hw(struct mlx5e_priv *priv, void *context)
+{
+ u8 *trust_state = context;
+ int err;
+
+ err = mlx5_set_trust_state(priv->mdev, *trust_state);
+ if (err)
+ return err;
+ priv->dcbx_dp.trust_state = *trust_state;
+
+ return 0;
+}
+
+static int mlx5e_set_trust_state(struct mlx5e_priv *priv, u8 trust_state)
{
struct mlx5e_channels new_channels = {};
+ bool reset_channels = true;
+ int err = 0;
mutex_lock(&priv->state_lock);
new_channels.params = priv->channels.params;
- mlx5e_trust_update_tx_min_inline_mode(priv, &new_channels.params);
+ mlx5e_params_calc_trust_tx_min_inline_mode(priv->mdev, &new_channels.params,
+ trust_state);
if (!test_bit(MLX5E_STATE_OPENED, &priv->state)) {
priv->channels.params = new_channels.params;
- goto out;
+ reset_channels = false;
}
/* Skip if tx_min_inline is the same */
if (new_channels.params.tx_min_inline_mode ==
priv->channels.params.tx_min_inline_mode)
- goto out;
+ reset_channels = false;
- mlx5e_safe_switch_channels(priv, &new_channels, NULL, NULL);
+ if (reset_channels)
+ err = mlx5e_safe_switch_channels(priv, &new_channels,
+ mlx5e_update_trust_state_hw,
+ &trust_state);
+ else
+ err = mlx5e_update_trust_state_hw(priv, &trust_state);
-out:
mutex_unlock(&priv->state_lock);
-}
-
-static int mlx5e_set_trust_state(struct mlx5e_priv *priv, u8 trust_state)
-{
- int err;
-
- err = mlx5_set_trust_state(priv->mdev, trust_state);
- if (err)
- return err;
- priv->dcbx_dp.trust_state = trust_state;
- mlx5e_trust_update_sq_inline_mode(priv);
return err;
}
@@ -1171,7 +1181,8 @@ static int mlx5e_trust_initialize(struct mlx5e_priv *priv)
if (err)
return err;
- mlx5e_trust_update_tx_min_inline_mode(priv, &priv->channels.params);
+ mlx5e_params_calc_trust_tx_min_inline_mode(priv->mdev, &priv->channels.params,
+ priv->dcbx_dp.trust_state);
err = mlx5_query_dscp2prio(priv->mdev, priv->dcbx_dp.dscp2prio);
if (err)
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 11/16] net/mlx5e: RX, Use indirect calls wrapper for posting descriptors
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (9 preceding siblings ...)
2020-02-26 1:12 ` [net-next 10/16] net/mlx5e: Change inline mode correctly when changing trust state Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 12/16] net/mlx5e: RX, Use indirect calls wrapper for handling compressed completions Saeed Mahameed
` (5 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Tariq Toukan, Maxim Mikityanskiy, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
We can avoid an indirect call per NAPI cycle wrapping the RX descriptors
posting call with the appropriate helper.
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_txrx.c | 11 +++++++++--
1 file changed, 9 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
index 257a7c9f7a14..267f4535c36b 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_txrx.c
@@ -31,6 +31,7 @@
*/
#include <linux/irq.h>
+#include <linux/indirect_call_wrapper.h>
#include "en.h"
#include "en/xdp.h"
#include "en/xsk/rx.h"
@@ -99,7 +100,10 @@ static bool mlx5e_napi_xsk_post(struct mlx5e_xdpsq *xsksq, struct mlx5e_rq *xskr
busy_xsk |= mlx5e_xsk_tx(xsksq, MLX5E_TX_XSK_POLL_BUDGET);
mlx5e_xsk_update_tx_wakeup(xsksq);
- xsk_rx_alloc_err = xskrq->post_wqes(xskrq);
+ xsk_rx_alloc_err = INDIRECT_CALL_2(xskrq->post_wqes,
+ mlx5e_post_rx_mpwqes,
+ mlx5e_post_rx_wqes,
+ xskrq);
busy_xsk |= mlx5e_xsk_update_rx_wakeup(xskrq, xsk_rx_alloc_err);
return busy_xsk;
@@ -142,7 +146,10 @@ int mlx5e_napi_poll(struct napi_struct *napi, int budget)
mlx5e_poll_ico_cq(&c->icosq.cq);
- busy |= rq->post_wqes(rq);
+ busy |= INDIRECT_CALL_2(rq->post_wqes,
+ mlx5e_post_rx_mpwqes,
+ mlx5e_post_rx_wqes,
+ rq);
if (xsk_open) {
mlx5e_poll_ico_cq(&c->xskicosq.cq);
busy |= mlx5e_poll_xdpsq_cq(&xsksq->cq);
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 12/16] net/mlx5e: RX, Use indirect calls wrapper for handling compressed completions
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (10 preceding siblings ...)
2020-02-26 1:12 ` [net-next 11/16] net/mlx5e: RX, Use indirect calls wrapper for posting descriptors Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 13/16] Documentation: fix vxlan typo in mlx5.rst Saeed Mahameed
` (4 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Tariq Toukan, Maxim Mikityanskiy, Saeed Mahameed
From: Tariq Toukan <tariqt@mellanox.com>
We can avoid an indirect call per compressed completion wrapping the
completion handling call with the appropriate helper.
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_rx.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index 1c3ab69cbd96..065c74a2d0c5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -158,7 +158,8 @@ static inline u32 mlx5e_decompress_cqes_cont(struct mlx5e_rq *rq,
mlx5e_read_mini_arr_slot(wq, cqd, cqcc);
mlx5e_decompress_cqe_no_hash(rq, wq, cqcc);
- rq->handle_rx_cqe(rq, &cqd->title);
+ INDIRECT_CALL_2(rq->handle_rx_cqe, mlx5e_handle_rx_cqe_mpwrq,
+ mlx5e_handle_rx_cqe, rq, &cqd->title);
}
mlx5e_cqes_update_owner(wq, cqcc - wq->cc);
wq->cc = cqcc;
@@ -178,7 +179,8 @@ static inline u32 mlx5e_decompress_cqes_start(struct mlx5e_rq *rq,
mlx5e_read_title_slot(rq, wq, cc);
mlx5e_read_mini_arr_slot(wq, cqd, cc + 1);
mlx5e_decompress_cqe(rq, wq, cc);
- rq->handle_rx_cqe(rq, &cqd->title);
+ INDIRECT_CALL_2(rq->handle_rx_cqe, mlx5e_handle_rx_cqe_mpwrq,
+ mlx5e_handle_rx_cqe, rq, &cqd->title);
cqd->mini_arr_idx++;
return mlx5e_decompress_cqes_cont(rq, wq, 1, budget_rem) - 1;
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 13/16] Documentation: fix vxlan typo in mlx5.rst
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (11 preceding siblings ...)
2020-02-26 1:12 ` [net-next 12/16] net/mlx5e: RX, Use indirect calls wrapper for handling compressed completions Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 14/16] net/mlx5: Fix header guard in rsc_dump.h Saeed Mahameed
` (3 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba; +Cc: netdev, Hans Wippel, Saeed Mahameed
From: Hans Wippel <ndev@hwipl.net>
Fix a vxlan typo in the mlx5 driver documentation.
Signed-off-by: Hans Wippel <ndev@hwipl.net>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
Documentation/networking/device_drivers/mellanox/mlx5.rst | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/Documentation/networking/device_drivers/mellanox/mlx5.rst b/Documentation/networking/device_drivers/mellanox/mlx5.rst
index f575a49790e8..e9b65035cd47 100644
--- a/Documentation/networking/device_drivers/mellanox/mlx5.rst
+++ b/Documentation/networking/device_drivers/mellanox/mlx5.rst
@@ -101,7 +101,7 @@ Enabling the driver and kconfig options
**External options** ( Choose if the corresponding mlx5 feature is required )
- CONFIG_PTP_1588_CLOCK: When chosen, mlx5 ptp support will be enabled
-- CONFIG_VXLAN: When chosen, mlx5 vxaln support will be enabled.
+- CONFIG_VXLAN: When chosen, mlx5 vxlan support will be enabled.
- CONFIG_MLXFW: When chosen, mlx5 firmware flashing support will be enabled (via devlink and ethtool).
Devlink info
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 14/16] net/mlx5: Fix header guard in rsc_dump.h
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (12 preceding siblings ...)
2020-02-26 1:12 ` [net-next 13/16] Documentation: fix vxlan typo in mlx5.rst Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 15/16] net/mlx5: sparse: warning: incorrect type in assignment Saeed Mahameed
` (2 subsequent siblings)
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba
Cc: netdev, Nathan Chancellor, Jason Gunthorpe, Saeed Mahameed
From: Nathan Chancellor <natechancellor@gmail.com>
Clang warns:
In file included from
../drivers/net/ethernet/mellanox/mlx5/core/main.c:73:
../drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h:4:9: warning:
'__MLX5_RSC_DUMP_H' is used as a header guard here, followed by #define
of a different macro [-Wheader-guard]
#ifndef __MLX5_RSC_DUMP_H
^~~~~~~~~~~~~~~~~
../drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h:5:9: note:
'__MLX5_RSC_DUMP__H' is defined here; did you mean '__MLX5_RSC_DUMP_H'?
#define __MLX5_RSC_DUMP__H
^~~~~~~~~~~~~~~~~~
__MLX5_RSC_DUMP_H
1 warning generated.
Make them match to get the intended behavior and remove the warning.
Fixes: 12206b17235a ("net/mlx5: Add support for resource dump")
Link: https://github.com/ClangBuiltLinux/linux/issues/897
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h
index 3b7573461a45..148270073e71 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/rsc_dump.h
@@ -2,7 +2,7 @@
/* Copyright (c) 2019 Mellanox Technologies. */
#ifndef __MLX5_RSC_DUMP_H
-#define __MLX5_RSC_DUMP__H
+#define __MLX5_RSC_DUMP_H
#include <linux/mlx5/driver.h>
#include "mlx5_core.h"
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 15/16] net/mlx5: sparse: warning: incorrect type in assignment
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (13 preceding siblings ...)
2020-02-26 1:12 ` [net-next 14/16] net/mlx5: Fix header guard in rsc_dump.h Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-26 1:12 ` [net-next 16/16] net/mlx5: sparse: warning: Using plain integer as NULL pointer Saeed Mahameed
2020-02-27 4:31 ` [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 David Miller
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba; +Cc: netdev, Saeed Mahameed, Moshe Shemesh
drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c:191:13:
sparse: warning: incorrect type in assignment (different base types)
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
index 94d7b69a95c7..c9c9b479bda5 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c
@@ -188,7 +188,7 @@ static int mlx5_fw_tracer_create_mkey(struct mlx5_fw_tracer *tracer)
MLX5_SET(create_mkey_in, in, translations_octword_actual_size,
DIV_ROUND_UP(TRACER_BUFFER_PAGE_NUM, 2));
- mtt = (u64 *)MLX5_ADDR_OF(create_mkey_in, in, klm_pas_mtt);
+ mtt = (__be64 *)MLX5_ADDR_OF(create_mkey_in, in, klm_pas_mtt);
for (i = 0 ; i < TRACER_BUFFER_PAGE_NUM ; i++)
mtt[i] = cpu_to_be64(tracer->buff.dma + i * PAGE_SIZE);
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* [net-next 16/16] net/mlx5: sparse: warning: Using plain integer as NULL pointer
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (14 preceding siblings ...)
2020-02-26 1:12 ` [net-next 15/16] net/mlx5: sparse: warning: incorrect type in assignment Saeed Mahameed
@ 2020-02-26 1:12 ` Saeed Mahameed
2020-02-27 4:31 ` [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 David Miller
16 siblings, 0 replies; 20+ messages in thread
From: Saeed Mahameed @ 2020-02-26 1:12 UTC (permalink / raw)
To: David S. Miller, kuba; +Cc: netdev, Saeed Mahameed, Moshe Shemesh
Return NULL instead of 0.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx5/core/lib/dm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/dm.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/dm.c
index e065c2f68f5a..6cbccba56f70 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/lib/dm.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/dm.c
@@ -21,7 +21,7 @@ struct mlx5_dm *mlx5_dm_create(struct mlx5_core_dev *dev)
struct mlx5_dm *dm;
if (!(MLX5_CAP_GEN_64(dev, general_obj_types) & MLX5_GENERAL_OBJ_TYPES_CAP_SW_ICM))
- return 0;
+ return NULL;
dm = kzalloc(sizeof(*dm), GFP_KERNEL);
if (!dm)
--
2.24.1
^ permalink raw reply related [flat|nested] 20+ messages in thread* Re: [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25
2020-02-26 1:12 [pull request][net-next 00/16] Mellanox, mlx5 updates 2020-02-25 Saeed Mahameed
` (15 preceding siblings ...)
2020-02-26 1:12 ` [net-next 16/16] net/mlx5: sparse: warning: Using plain integer as NULL pointer Saeed Mahameed
@ 2020-02-27 4:31 ` David Miller
16 siblings, 0 replies; 20+ messages in thread
From: David Miller @ 2020-02-27 4:31 UTC (permalink / raw)
To: saeedm; +Cc: kuba, netdev
From: Saeed Mahameed <saeedm@mellanox.com>
Date: Tue, 25 Feb 2020 17:12:30 -0800
> This series adds misc updates to mlx5 driver.
>
> For more information please see tag log below.
>
> Please pull and let me know if there is any problem.
Pulled, thanks Saeed.
^ permalink raw reply [flat|nested] 20+ messages in thread