mlx5 fixes 2020-05-28
Hi Dave/Jakub,
This series introduces some fixes to mlx5 driver.
v1->v2:
- Fix bad sha1, Jakub.
- Added one more patch by Pablo.
net/mlx5e: replace EINVAL in mlx5e_flower_parse_meta()
Nothing major, the only patch worth mentioning is the suspend/resume crash fix by adding the missing pci device handlers, the fix is very straight forward and as Dexuan already expressed, the patch is important for Azure users to avoid crash on VM hibernation, patch is marked for -stable v4.6 below.
Conflict note:
('net/mlx5e: Fix MLX5_TC_CT dependencies') has a trivial one line conflict with current net-next, which can be resolved by simply using the line from net-next.
Please pull and let me know if there is any problem.
For -stable v4.6
('net/mlx5: Fix crash upon suspend/resume')
For -stable v5.6
('net/mlx5e: replace EINVAL in mlx5e_flower_parse_meta()')
Thanks,
Saeed.
---
The following changes since commit 7c6d2ecbda83150b2036a2b36b21381ad4667762:
net: be more gentle about silly gso requests coming from user (2020-05-28 16:31:30 -0700)
are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux.git tags/mlx5-fixes-2020-05-28
for you to fetch changes up to a683012a8e77675a1947cc8f11f97cdc1d5bb769:
net/mlx5e: replace EINVAL in mlx5e_flower_parse_meta() (2020-05-29 13:07:54 -0700)
----------------------------------------------------------------
mlx5-fixes-2020-05-28
----------------------------------------------------------------
Aya Levin (1):
net/mlx5e: Fix arch depending casting issue in FEC
Maor Dickman (1):
net/mlx5e: Remove warning "devices are not on same switch HW"
Mark Bloch (1):
net/mlx5: Fix crash upon suspend/resume
Pablo Neira Ayuso (1):
net/mlx5e: replace EINVAL in mlx5e_flower_parse_meta()
Roi Dayan (1):
net/mlx5e: Fix stats update for matchall classifier
Tal Gilboa (1):
net/mlx5e: Properly set default values when disabling adaptive moderation
Vlad Buslov (1):
net/mlx5e: Fix MLX5_TC_CT dependencies
drivers/net/ethernet/mellanox/mlx5/core/Kconfig | 2 +- drivers/net/ethernet/mellanox/mlx5/core/en.h | 10 +++--- drivers/net/ethernet/mellanox/mlx5/core/en/port.c | 24 +++++++------ .../net/ethernet/mellanox/mlx5/core/en_ethtool.c | 41 +++++++++++++++------- drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 20 +++++++---- drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 12 +++---- drivers/net/ethernet/mellanox/mlx5/core/main.c | 18 ++++++++++ 7 files changed, 84 insertions(+), 43 deletions(-)
PATCH v2 1/7 net/mlx5: Fix crash upon suspend/resume
From: Mark Bloch <[email protected]>
Currently a Linux system with the mlx5 NIC always crashes upon hibernation - suspend/resume.
Add basic callbacks so the NIC could be suspended and resumed.
--- drivers/net/ethernet/mellanox/mlx5/core/main.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+)
drivers/net/ethernet/mellanox/mlx5/core/main.c +18 -0
@@ -1549,6 +1549,22 @@ static void shutdown(struct pci_dev *pdev)mlx5_pci_disable_device (dev); } +static int mlx5_suspend (struct pci_dev *pdev,pm_message_t state) +{ +struct mlx5_core_dev *dev =pci_get_drvdata (pdev); + +mlx5_unload_one (dev, false); + +return 0 ; +} + +static int mlx5_resume (struct pci_dev *pdev) +{ +struct mlx5_core_dev *dev =pci_get_drvdata (pdev); + +return mlx5_load_one (dev, false); +} +static const struct pci_device_id mlx5_core_pci_table[] = { {PCI_VDEVICE (MELLANOX ,PCI_DEVICE_ID_MELLANOX_CONNECTIB ) }, {PCI_VDEVICE (MELLANOX ,0x1012 ),MLX5_PCI_DEV_IS_VF },/* Connect-IB VF */
@@ -1592,6 +1608,8 @@ static struct pci_driver mlx5_core_driver = { .id_table = mlx5_core_pci_table, .probe = init_one, .remove = remove_one, + .suspend = mlx5_suspend, + .resume = mlx5_resume, .shutdown = shutdown, .err_handler = &mlx5_err_handler, .sriov_configure = mlx5_core_sriov_configure,
Signature · 1 line
2.26.2
PATCH v2 2/7 net/mlx5e: Fix stats update for matchall classifier
In reply to Saeed MahameedFrom: Roi Dayan <[email protected]>
It's bytes, packets, lastused.
--- drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +1 -1
@@ -4614,7 +4614,7 @@ void mlx5e_tc_stats_matchall(struct mlx5e_priv *priv, dpkts = cur_stats.rx_packets - rpriv->prev_vf_vport_stats.rx_packets; dbytes = cur_stats.rx_bytes - rpriv->prev_vf_vport_stats.rx_bytes; rpriv->prev_vf_vport_stats = cur_stats; -flow_stats_update (&ma->stats, dpkts, dbytes, jiffies, +flow_stats_update (&ma->stats, dbytes, dpkts, jiffies,FLOW_ACTION_HW_STATS_DELAYED ); }
Signature · 1 line
2.26.2
PATCH v2 3/7 net/mlx5e: Remove warning "devices are not on same switch HW"
In reply to Saeed MahameedFrom: Maor Dickman <[email protected]>
On tunnel decap rule insertion, the indirect mechanism will attempt to offload the rule on all uplink representors which will trigger the "devices are not on same switch HW, can't offload forwarding" message for the uplink which isn't on the same switch HW as the VF representor.
The above flow is valid and shouldn't cause warning message, fix by removing the warning and only report this flow using extack.
--- drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 4 ---- 1 file changed, 4 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +0 -4
@@ -3849,10 +3849,6 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv,if (!mlx5e_is_valid_eswitch_fwd_dev (priv, out_dev)) {NL_SET_ERR_MSG_MOD (extack,"devices are not on same switch HW, can't offload forwarding" ); -netdev_warn (priv->netdev, -"devices %s %s not on same switch HW, can't offload forwarding\n" , - priv->netdev->name, - out_dev->name);return -EOPNOTSUPP ; }
Signature · 1 line
2.26.2
PATCH v2 4/7 net/mlx5e: Fix arch depending casting issue in FEC
In reply to Saeed MahameedFrom: Aya Levin <[email protected]>
Change type of active_fec to u32 to match the type expected by mlx5e_get_fec_mode. Copy active_fec and configured_fec values to unsigned long before preforming bitwise manipulations. Take the same approach when configuring FEC over 50G link modes: copy the policy into an unsigned long and only than preform bitwise operations.
--- .../net/ethernet/mellanox/mlx5/core/en/port.c | 24 ++++++++++--------- .../ethernet/mellanox/mlx5/core/en_ethtool.c | 20 +++++++++------- 2 files changed, 24 insertions(+), 20 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en/port.c +13 -11
@@ -369,17 +369,19 @@ enum mlx5e_fec_supported_link_mode { *_policy = MLX5_GET(pplm_reg, _buf, fec_override_admin_##link); \ } while (0) -#define MLX5E_FEC_OVERRIDE_ADMIN_50G_POLICY (buf, policy, write, link) \ - do { \ - u16 *__policy = &(policy); \ - bool _write = (write); \ - \ - if (_write && *__policy) \ - *__policy = find_first_bit((u_long *)__policy, \ - sizeof(u16) * BITS_PER_BYTE);\ - MLX5E_FEC_OVERRIDE_ADMIN_POLICY(buf, *__policy, _write, link); \ - if (!_write && *__policy) \ - *__policy = 1 << *__policy; \ +#define MLX5E_FEC_OVERRIDE_ADMIN_50G_POLICY (buf, policy, write, link) \ + do { \ + unsigned long policy_long; \ + u16 *__policy = &(policy); \ + bool _write = (write); \ + \ + policy_long = *__policy; \ + if (_write && *__policy) \ + *__policy = find_first_bit(&policy_long, \ + sizeof(policy_long) * BITS_PER_BYTE);\ + MLX5E_FEC_OVERRIDE_ADMIN_POLICY(buf, *__policy, _write, link); \ + if (!_write && *__policy) \ + *__policy = 1 << *__policy; \ } while (0)/* get/set FEC admin field for a given speed */
drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c +11 -9
@@ -665,11 +665,12 @@ static const u32 pplm_fec_2_ethtool_linkmodes[] = {static int get_fec_supported_advertised (struct mlx5_core_dev *dev,struct ethtool_link_ksettings *link_ksettings) { -u_long active_fec =0 ; +unsigned long active_fec_long; +u32 active_fec;u32 bitn;int err; - err =mlx5e_get_fec_mode (dev, (u32 *)&active_fec,NULL ); + err =mlx5e_get_fec_mode (dev, &active_fec,NULL );if (err)return (err == -EOPNOTSUPP ) ?0 : err;
@@ -682,10 +683,11 @@ static int get_fec_supported_advertised(struct mlx5_core_dev *dev,MLX5E_ADVERTISE_SUPPORTED_FEC (MLX5E_FEC_LLRS_272_257_1 ,ETHTOOL_LINK_MODE_FEC_LLRS_BIT ); + active_fec_long = active_fec;/* active fec is a bit set, find out which bit is set and * advertise the corresponding ethtool bit */ - bitn =find_first_bit (&active_fec,sizeof (u32) *BITS_PER_BYTE ); + bitn =find_first_bit (&active_fec_long,sizeof (active_fec_long) *BITS_PER_BYTE );if (bitn <ARRAY_SIZE (pplm_fec_2_ethtool_linkmodes))__set_bit (pplm_fec_2_ethtool_linkmodes[bitn], link_ksettings->link_modes.advertising);
@@ -1517,8 +1519,8 @@ static int mlx5e_get_fecparam(struct net_device *netdev, {struct mlx5e_priv *priv =netdev_priv (netdev);struct mlx5_core_dev *mdev = priv->mdev; -u16 fec_configured =0 ; -u32 fec_active =0 ; +u16 fec_configured; +u32 fec_active;int err; err =mlx5e_get_fec_mode (mdev, &fec_active, &fec_configured);
@@ -1526,14 +1528,14 @@ static int mlx5e_get_fecparam(struct net_device *netdev,if (err)return err; - fecparam->active_fec =pplm2ethtool_fec ((u_long )fec_active, -sizeof (u32) *BITS_PER_BYTE ); + fecparam->active_fec =pplm2ethtool_fec ((unsigned long )fec_active, +sizeof (unsigned long ) *BITS_PER_BYTE );if (!fecparam->active_fec)return -EOPNOTSUPP ; - fecparam->fec =pplm2ethtool_fec ((u_long )fec_configured, -sizeof (u16) *BITS_PER_BYTE ); + fecparam->fec =pplm2ethtool_fec ((unsigned long )fec_configured, +sizeof (unsigned long ) *BITS_PER_BYTE );return 0 ; }
Signature · 1 line
2.26.2
PATCH v2 5/7 net/mlx5e: Properly set default values when disabling adaptive moderation
In reply to Saeed MahameedFrom: Tal Gilboa <[email protected]>
Add a call to mlx5e_reset_rx/tx_moderation() when enabling/disabling adaptive moderation, in order to select the proper default values.
In order to do so, we separate the logic of selecting the moderation values and setting moderion mode (CQE/EQE based).
--- drivers/net/ethernet/mellanox/mlx5/core/en.h | 10 +++++---- .../ethernet/mellanox/mlx5/core/en_ethtool.c | 21 +++++++++++++++---- .../net/ethernet/mellanox/mlx5/core/en_main.c | 20 ++++++++++++------ 3 files changed, 37 insertions(+), 14 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en.h +6 -4
@@ -1068,10 +1068,12 @@ void mlx5e_deactivate_priv_channels(struct mlx5e_priv *priv);void mlx5e_build_default_indir_rqt (u32 *indirection_rqt,int len,int num_channels); -void mlx5e_set_tx_cq_mode_params (struct mlx5e_params *params, -u8 cq_period_mode); -void mlx5e_set_rx_cq_mode_params (struct mlx5e_params *params, -u8 cq_period_mode); + +void mlx5e_reset_tx_moderation (struct mlx5e_params *params,u8 cq_period_mode); +void mlx5e_reset_rx_moderation (struct mlx5e_params *params,u8 cq_period_mode); +void mlx5e_set_tx_cq_mode_params (struct mlx5e_params *params,u8 cq_period_mode); +void mlx5e_set_rx_cq_mode_params (struct mlx5e_params *params,u8 cq_period_mode); +void mlx5e_set_rq_type (struct mlx5_core_dev *mdev,struct mlx5e_params *params);void mlx5e_init_rq_type_params (struct mlx5_core_dev *mdev,struct mlx5e_params *params);
drivers/net/ethernet/mellanox/mlx5/core/en_ethtool.c +17 -4
@@ -527,8 +527,8 @@ int mlx5e_ethtool_set_coalesce(struct mlx5e_priv *priv,struct dim_cq_moder *rx_moder, *tx_moder;struct mlx5_core_dev *mdev = priv->mdev;struct mlx5e_channels new_channels = {}; +bool reset_rx, reset_tx;int err =0 ; -bool reset;if (!MLX5_CAP_GEN (mdev, cq_moderation))return -EOPNOTSUPP ;
@@ -566,15 +566,28 @@ int mlx5e_ethtool_set_coalesce(struct mlx5e_priv *priv, }/* we are opened */ - reset = (!!coal->use_adaptive_rx_coalesce != priv->channels.params.rx_dim_enabled) || - (!!coal->use_adaptive_tx_coalesce != priv->channels.params.tx_dim_enabled); + reset_rx = !!coal->use_adaptive_rx_coalesce != priv->channels.params.rx_dim_enabled; + reset_tx = !!coal->use_adaptive_tx_coalesce != priv->channels.params.tx_dim_enabled; -if (!reset) { +if (!reset_rx && !reset_tx) {mlx5e_set_priv_channels_coalesce (priv, coal); priv->channels.params = new_channels.params; goto out; } +if (reset_rx) { +u8 mode =MLX5E_GET_PFLAG (&new_channels.params, +MLX5E_PFLAG_RX_CQE_BASED_MODER ); + +mlx5e_reset_rx_moderation (&new_channels.params, mode); + } +if (reset_tx) { +u8 mode =MLX5E_GET_PFLAG (&new_channels.params, +MLX5E_PFLAG_TX_CQE_BASED_MODER ); + +mlx5e_reset_tx_moderation (&new_channels.params, mode); + } + err =mlx5e_safe_switch_channels (priv, &new_channels,NULL ,NULL ); out:
drivers/net/ethernet/mellanox/mlx5/core/en_main.c +14 -6
@@ -4716,7 +4716,7 @@ static u8 mlx5_to_net_dim_cq_period_mode(u8 cq_period_mode)DIM_CQ_PERIOD_MODE_START_FROM_EQE ; } -void mlx5e_set_tx_cq_mode_params (struct mlx5e_params *params,u8 cq_period_mode) +void mlx5e_reset_tx_moderation (struct mlx5e_params *params,u8 cq_period_mode) {if (params->tx_dim_enabled) {u8 dim_period_mode =mlx5_to_net_dim_cq_period_mode (cq_period_mode);
@@ -4725,13 +4725,9 @@ void mlx5e_set_tx_cq_mode_params(struct mlx5e_params *params, u8 cq_period_mode) }else { params->tx_cq_moderation =mlx5e_get_def_tx_moderation (cq_period_mode); } - -MLX5E_SET_PFLAG (params,MLX5E_PFLAG_TX_CQE_BASED_MODER , - params->tx_cq_moderation.cq_period_mode == -MLX5_CQ_PERIOD_MODE_START_FROM_CQE ); } -void mlx5e_set_rx_cq_mode_params (struct mlx5e_params *params,u8 cq_period_mode) +void mlx5e_reset_rx_moderation (struct mlx5e_params *params,u8 cq_period_mode) {if (params->rx_dim_enabled) {u8 dim_period_mode =mlx5_to_net_dim_cq_period_mode (cq_period_mode);
@@ -4740,7 +4736,19 @@ void mlx5e_set_rx_cq_mode_params(struct mlx5e_params *params, u8 cq_period_mode) }else { params->rx_cq_moderation =mlx5e_get_def_rx_moderation (cq_period_mode); } +} + +void mlx5e_set_tx_cq_mode_params (struct mlx5e_params *params,u8 cq_period_mode) +{ +mlx5e_reset_tx_moderation (params, cq_period_mode); +MLX5E_SET_PFLAG (params,MLX5E_PFLAG_TX_CQE_BASED_MODER , + params->tx_cq_moderation.cq_period_mode == +MLX5_CQ_PERIOD_MODE_START_FROM_CQE ); +} +void mlx5e_set_rx_cq_mode_params (struct mlx5e_params *params,u8 cq_period_mode) +{ +mlx5e_reset_rx_moderation (params, cq_period_mode);MLX5E_SET_PFLAG (params,MLX5E_PFLAG_RX_CQE_BASED_MODER , params->rx_cq_moderation.cq_period_mode ==MLX5_CQ_PERIOD_MODE_START_FROM_CQE );
Signature · 1 line
2.26.2
PATCH v2 6/7 net/mlx5e: Fix MLX5_TC_CT dependencies
In reply to Saeed MahameedFrom: Vlad Buslov <[email protected]>
Change MLX5_TC_CT config dependencies to include MLX5_ESWITCH instead of MLX5_CORE_EN && NET_SWITCHDEV, which are already required by MLX5_ESWITCH. Without this change mlx5 fails to compile if user disables MLX5_ESWITCH without also manually disabling MLX5_TC_CT.
--- drivers/net/ethernet/mellanox/mlx5/core/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/mellanox/mlx5/core/Kconfig +1 -1
@@ -80,7 +80,7 @@ config MLX5_ESWITCH config MLX5_TC_CT bool "MLX5 TC connection tracking offload support" - depends on MLX5_CORE_EN && NET_SWITCHDEV && NF_FLOW_TABLE && NET_ACT_CT && NET_TC_SKB_EXT + depends on MLX5_ESWITCH && NF_FLOW_TABLE && NET_ACT_CT && NET_TC_SKB_EXT default y help Say Y here if you want to support offloading connection tracking rules
Signature · 1 line
2.26.2
PATCH v2 7/7 net/mlx5e: replace EINVAL in mlx5e_flower_parse_meta()
In reply to Saeed MahameedFrom: Pablo Neira Ayuso <[email protected]>
The drivers reports EINVAL to userspace through netlink on invalid meta match. This is confusing since EINVAL is usually reserved for malformed netlink messages. Replace it by more meaningful codes.
--- drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +3 -3
@@ -2068,7 +2068,7 @@ static int mlx5e_flower_parse_meta(struct net_device *filter_dev,flow_rule_match_meta (rule, &match);if (match.mask->ingress_ifindex !=0xFFFFFFFF ) {NL_SET_ERR_MSG_MOD (extack,"Unsupported ingress ifindex mask" ); -return -EINVAL ; +return -EOPNOTSUPP ; } ingress_dev = __dev_get_by_index(dev_net(filter_dev),
@@ -2076,13 +2076,13 @@ static int mlx5e_flower_parse_meta(struct net_device *filter_dev,if (!ingress_dev) {NL_SET_ERR_MSG_MOD (extack,"Can't find the ingress port to match on" ); -return -EINVAL ; +return -ENOENT ; }if (ingress_dev != filter_dev) {NL_SET_ERR_MSG_MOD (extack,"Can't match on the ingress filter port" ); -return -EINVAL ; +return -EOPNOTSUPP ; }return 0 ;
Signature · 1 line
2.26.2
David Miller
maintainerFrom: Saeed Mahameed <[email protected]>
Date: Fri, 29 May 2020 13:46:03 -0700
Quoted 14 lines
This series introduces some fixes to mlx5 driver.
v1->v2:
- Fix bad sha1, Jakub.
- Added one more patch by Pablo.
net/mlx5e: replace EINVAL inmlx5e_flower_parse_meta()Nothing major, the only patch worth mentioning is the suspend/resume crash fix by adding the missing pci device handlers, the fix is very straight forward and as Dexuan already expressed, the patch is important for Azure users to avoid crash on VM hibernation, patch is marked for -stable v4.6 below.
Quoted 6 lines
Conflict note:
('net/mlx5e: FixMLX5_TC_CTdependencies') has a trivial one line conflict with current net-next, which can be resolved by simply using the line from net-next.Please pull and let me know if there is any problem.
Pulled, and thanks for the conflict info.
Quoted 5 lines
For -stable v4.6
('net/mlx5: Fix crash upon suspend/resume')For -stable v5.6
('net/mlx5e: replace EINVAL inmlx5e_flower_parse_meta()')
Queued up.