Skip to contentSkip to conversation

devlink: Add port function attribute for IO EQs

PATCH Applied Parav Pandit 5 participants 8 messages Apr 6, 2024 – Apr 8, 2024

Subsystems (6)
lore
Share & tools
lore
#0 lore raw

v4 cover devlink: Add port function attribute for IO EQs

Currently, PCI SFs and VFs use IO event queues to deliver netdev per channel events. The number of netdev channels is a function of IO event queues. In the second scenario of an RDMA device, the completion vectors are also a function of IO event queues. Currently, an administrator on the hypervisor has no means to provision the number of IO event queues for the SF device or the VF device. Device/firmware determines some arbitrary value for these IO event queues. Due to this, the SF netdev channels are unpredictable, and consequently, the performance is too.

This short series introduces a new port function attribute: max_io_eqs. The goal is to provide administrators at the hypervisor level with the ability to provision the maximum number of IO event queues for a function. This gives the control to the administrator to provision right number of IO event queues and have predictable performance.

Examples of when an administrator provisions (set) maximum number of IO event queues when using switchdev mode:

  $ devlink port show pci/0000:06:00.0/1
      pci/0000:06:00.0/1: type eth netdev enp6s0pf0vf0 flavour pcivf pfnum 0 vfnum 0
          function:
          hw_addr 00:00:00:00:00:00 roce enable max_io_eqs 10
  $ devlink port function set pci/0000:06:00.0/1 max_io_eqs 20
  $ devlink port show pci/0000:06:00.0/1
      pci/0000:06:00.0/1: type eth netdev enp6s0pf0vf0 flavour pcivf pfnum 0 vfnum 0
          function:
          hw_addr 00:00:00:00:00:00 roce enable max_io_eqs 20

This sets the corresponding maximum IO event queues of the function before it is enumerated. Thus, when the VF/SF driver reads the capability from the device, it sees the value provisioned by the hypervisor. The driver is then able to configure the number of channels for the net device, as well as the number of completion vectors for the RDMA device. The device/firmware also honors the provisioned value, hence any VF/SF driver attempting to create IO EQs beyond provisioned value results in an error.

With above setting now, the administrator is able to achieve the 2x performance on SFs with 20 channels. In second example when SF was provisioned for a container with 2 cpus, the administrator provisioned only 2 IO event queues, thereby saving device resources.

With the above settings now in place, the administrator achieved 2x performance with the SF device with 20 channels. In the second example, when the SF was provisioned for a container with 2 CPUs, the administrator provisioned only 2 IO event queues, thereby saving device resources.

changelog:
v2->v3:
- limited to 80 chars per line in devlink
- fixed comments from Jakub in mlx5 driver to fix missing mutex unlock on error path
v1->v2:
- limited comment to 80 chars per line in header file
- fixed set function variables for reverse christmas tree
- fixed comments from Kalesh
- fixed missing kfree in get call
- returning error code for get cmd failure
- fixed error msg copy paste error in set on cmd failure

Parav Pandit (2):
devlink: Support setting max_io_eqs
mlx5/core: Support max_io_eqs for a function

 .../networking/devlink/devlink-port.rst       | 33 +++++++
 .../mellanox/mlx5/core/esw/devlink_port.c     |  4 +
 .../net/ethernet/mellanox/mlx5/core/eswitch.h |  7 ++
 .../mellanox/mlx5/core/eswitch_offloads.c     | 97 +++++++++++++++++++
 include/net/devlink.h                         | 14 +++
 include/uapi/linux/devlink.h                  |  1 +
 net/devlink/port.c                            | 53 ++++++++++
 7 files changed, 209 insertions(+)
Signature · 1 line
2.26.2
#1 lore raw

PATCH v4 1/2 devlink: Support setting max_io_eqs

Many devices send event notifications for the IO queues, such as tx and rx queues, through event queues.

Enable a privileged owner, such as a hypervisor PF, to set the number of IO event queues for the VF and SF during the provisioning stage.

example:
Get maximum IO event queues of the VF device::

  $ devlink port show pci/0000:06:00.0/2
  pci/0000:06:00.0/2: type eth netdev enp6s0pf0vf1 flavour pcivf pfnum 0 vfnum 1
      function:
          hw_addr 00:00:00:00:00:00 ipsec_packet disabled max_io_eqs 10

Set maximum IO event queues of the VF device::

  $ devlink port function set pci/0000:06:00.0/2 max_io_eqs 32
  $ devlink port show pci/0000:06:00.0/2
  pci/0000:06:00.0/2: type eth netdev enp6s0pf0vf1 flavour pcivf pfnum 0 vfnum 1
      function:
          hw_addr 00:00:00:00:00:00 ipsec_packet disabled max_io_eqs 32

---
changelog:
v3->v4:
- addressed comment from Jakub to improve devlink documentation v2->v3:
- limited 80 chars per line
v1->v2:
- limited comment to 80 chars per line in header file

---
 .../networking/devlink/devlink-port.rst       | 33 ++++++++++++
 include/net/devlink.h                         | 14 +++++
 include/uapi/linux/devlink.h                  |  1 +
 net/devlink/port.c                            | 53 +++++++++++++++++++
 4 files changed, 101 insertions(+)
Documentation/networking/devlink/devlink-port.rst +33 -0
index 562f46b41274..9d22d41a7cd1 100644
@@ -134,6 +134,9 @@ Users may also set the IPsec crypto capability of the function using
 Users may also set the IPsec packet capability of the function using
 `devlink port function set ipsec_packet` command.
 
+Users may also set the maximum IO event queues of the function
+using `devlink port function set max_io_eqs` command.
+
 Function attributes
 ===================
 
@@ -295,6 +298,36 @@ policy is processed in software by the kernel.
         function:
             hw_addr 00:00:00:00:00:00 ipsec_packet enabled
 
+Maximum IO events queues setup
+------------------------------
+When user sets maximum number of IO event queues for a SF or
+a VF, such function driver is limited to consume only enforced
+number of IO event queues.
+
+IO event queues deliver events related to IO queues, including network
+device transmit and receive queues (txq and rxq) and RDMA Queue Pairs (QPs).
+For example, the number of netdevice channels and RDMA device completion
+vectors are derived from the function's IO event queues. Usually, the number
+of interrupt vectors consumed by the driver is limited by the number of IO
+event queues per device, as each of the IO event queues is connected to an
+interrupt vector.
+
+- Get maximum IO event queues of the VF device::
+
+    $ devlink port show pci/0000:06:00.0/2
+    pci/0000:06:00.0/2: type eth netdev enp6s0pf0vf1 flavour pcivf pfnum 0 vfnum 1
+        function:
+            hw_addr 00:00:00:00:00:00 ipsec_packet disabled max_io_eqs 10
+
+- Set maximum IO event queues of the VF device::
+
+    $ devlink port function set pci/0000:06:00.0/2 max_io_eqs 32
+
+    $ devlink port show pci/0000:06:00.0/2
+    pci/0000:06:00.0/2: type eth netdev enp6s0pf0vf1 flavour pcivf pfnum 0 vfnum 1
+        function:
+            hw_addr 00:00:00:00:00:00 ipsec_packet disabled max_io_eqs 32
+
 Subfunction
 ============
 
include/net/devlink.h +14 -0
index 9ac394bdfbe4..bb1af599d101 100644
@@ -1602,6 +1602,14 @@ void devlink_free(struct devlink *devlink);
  *			      capability. Should be used by device drivers to
  *			      enable/disable ipsec_packet capability of a
  *			      function managed by the devlink port.
+ * @port_fn_max_io_eqs_get: Callback used to get port function's maximum number
+ *			    of event queues. Should be used by device drivers to
+ *			    report the maximum event queues of a function
+ *			    managed by the devlink port.
+ * @port_fn_max_io_eqs_set: Callback used to set port function's maximum number
+ *			    of event queues. Should be used by device drivers to
+ *			    configure maximum number of event queues
+ *			    of a function managed by the devlink port.
  *
  * Note: Driver should return -EOPNOTSUPP if it doesn't support
  * port function (@port_fn_*) handling for a particular port.
@@ -1651,6 +1659,12 @@ struct devlink_port_ops {
 	int (*port_fn_ipsec_packet_set)(struct devlink_port *devlink_port,
 					bool enable,
 					struct netlink_ext_ack *extack);
+	int (*port_fn_max_io_eqs_get)(struct devlink_port *devlink_port,
+				      u32 *max_eqs,
+				      struct netlink_ext_ack *extack);
+	int (*port_fn_max_io_eqs_set)(struct devlink_port *devlink_port,
+				      u32 max_eqs,
+				      struct netlink_ext_ack *extack);
 };
 
 void devlink_port_init(struct devlink *devlink,
include/uapi/linux/devlink.h +1 -0
index 2da0c7eb6710..9401aa343673 100644
@@ -686,6 +686,7 @@ enum devlink_port_function_attr {
 	DEVLINK_PORT_FN_ATTR_OPSTATE,	/* u8 */
 	DEVLINK_PORT_FN_ATTR_CAPS,	/* bitfield32 */
 	DEVLINK_PORT_FN_ATTR_DEVLINK,	/* nested */
+	DEVLINK_PORT_FN_ATTR_MAX_IO_EQS,	/* u32 */
 
 	__DEVLINK_PORT_FUNCTION_ATTR_MAX,
 	DEVLINK_PORT_FUNCTION_ATTR_MAX = __DEVLINK_PORT_FUNCTION_ATTR_MAX - 1
net/devlink/port.c +53 -0

89 lines not on this page · raw message

Signature · 1 line
2.26.2
#2 lore raw

PATCH v4 2/2 mlx5/core: Support max_io_eqs for a function

In reply to Parav Pandit

Implement get and set for the maximum IO event queues for SF and VF. This enables administrator on the hypervisor to control the maximum IO event queues which are typically used to derive the maximum and default number of net device channels or rdma device completion vectors.

---
changelog:
v3->v4:
- addressed comment from David
- replaced open coded overflow check with kernel api v2->v3:
- limited to 80 chars per line in devlink
- fixed comments from Jakub in mlx5 driver to fix missing mutex unlock on error path
v1->v2:
- fixed comments from Kalesh
- fixed missing kfree in get call
- returning error code for get cmd failure
- fixed error msg copy paste error in set on cmd failure
- limited code to 80 chars limit
- fixed set function variables for reverse christmas tree

---
 .../mellanox/mlx5/core/esw/devlink_port.c     |  4 +
 .../net/ethernet/mellanox/mlx5/core/eswitch.h |  7 ++
 .../mellanox/mlx5/core/eswitch_offloads.c     | 97 +++++++++++++++++++
 3 files changed, 108 insertions(+)
drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c +4 -0
index d8e739cbcbce..f8869c9b6802 100644
@@ -98,6 +98,8 @@ static const struct devlink_port_ops mlx5_esw_pf_vf_dl_port_ops = {
 	.port_fn_ipsec_packet_get = mlx5_devlink_port_fn_ipsec_packet_get,
 	.port_fn_ipsec_packet_set = mlx5_devlink_port_fn_ipsec_packet_set,
 #endif /* CONFIG_XFRM_OFFLOAD */
+	.port_fn_max_io_eqs_get = mlx5_devlink_port_fn_max_io_eqs_get,
+	.port_fn_max_io_eqs_set = mlx5_devlink_port_fn_max_io_eqs_set,
 };
 
 static void mlx5_esw_offloads_sf_devlink_port_attrs_set(struct mlx5_eswitch *esw,
@@ -143,6 +145,8 @@ static const struct devlink_port_ops mlx5_esw_dl_sf_port_ops = {
 	.port_fn_state_get = mlx5_devlink_sf_port_fn_state_get,
 	.port_fn_state_set = mlx5_devlink_sf_port_fn_state_set,
 #endif
+	.port_fn_max_io_eqs_get = mlx5_devlink_port_fn_max_io_eqs_get,
+	.port_fn_max_io_eqs_set = mlx5_devlink_port_fn_max_io_eqs_set,
 };
 
 int mlx5_esw_offloads_devlink_port_register(struct mlx5_eswitch *esw, struct mlx5_vport *vport)
drivers/net/ethernet/mellanox/mlx5/core/eswitch.h +7 -0
index 349e28a6dd8d..50ce1ea20dd4 100644
@@ -573,6 +573,13 @@ int mlx5_devlink_port_fn_ipsec_packet_get(struct devlink_port *port, bool *is_en
 int mlx5_devlink_port_fn_ipsec_packet_set(struct devlink_port *port, bool enable,
 					  struct netlink_ext_ack *extack);
 #endif /* CONFIG_XFRM_OFFLOAD */
+int mlx5_devlink_port_fn_max_io_eqs_get(struct devlink_port *port,
+					u32 *max_io_eqs,
+					struct netlink_ext_ack *extack);
+int mlx5_devlink_port_fn_max_io_eqs_set(struct devlink_port *port,
+					u32 max_io_eqs,
+					struct netlink_ext_ack *extack);
+
 void *mlx5_eswitch_get_uplink_priv(struct mlx5_eswitch *esw, u8 rep_type);
 
 int __mlx5_eswitch_set_vport_vlan(struct mlx5_eswitch *esw,
drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c +97 -0

106 lines not on this page · raw message

Signature · 1 line
2.26.2
#3 lore raw

Re: [net-next v4 2/2] mlx5/core: Support max_io_eqs for a function

Quoted 24 lines · Parav Pandit

Implement get and set for the maximum IO event queues for SF and VF. This enables administrator on the hypervisor to control the maximum IO event queues which are typically used to derive the maximum and default number of net device channels or rdma device completion vectors.

---
changelog:
v3->v4:
- addressed comment from David
- replaced open coded overflow check with kernel api v2->v3:
- limited to 80 chars per line in devlink
- fixed comments from Jakub in mlx5 driver to fix missing mutex unlock on error path
v1->v2:
- fixed comments from Kalesh
- fixed missing kfree in get call
- returning error code for get cmd failure
- fixed error msg copy paste error in set on cmd failure
- limited code to 80 chars limit
- fixed set function variables for reverse christmas tree

Quoted 6 lines
---
 .../mellanox/mlx5/core/esw/devlink_port.c     |  4 +
 .../net/ethernet/mellanox/mlx5/core/eswitch.h |  7 ++
 .../mellanox/mlx5/core/eswitch_offloads.c     | 97 +++++++++++++++++++
 3 files changed, 108 insertions(+)

LGTM, thanks for addressing.

maintainer Reviewed-by on v4 2/2
#4 lore raw

Re: [net-next v4 2/2] mlx5/core: Support max_io_eqs for a function

In reply to Parav Pandit
Quoted 7 lines · Sat, Apr 06, 2024 at 03:05:38AM CEST, [email protected] wrote:

Implement get and set for the maximum IO event queues for SF and VF. This enables administrator on the hypervisor to control the maximum IO event queues which are typically used to derive the maximum and default number of net device channels or rdma device completion vectors.

#5 lore raw
In reply to Parav Pandit
Quoted 40 lines · Parav Pandit

Currently, PCI SFs and VFs use IO event queues to deliver netdev per channel events. The number of netdev channels is a function of IO event queues. In the second scenario of an RDMA device, the completion vectors are also a function of IO event queues. Currently, an administrator on the hypervisor has no means to provision the number of IO event queues for the SF device or the VF device. Device/firmware determines some arbitrary value for these IO event queues. Due to this, the SF netdev channels are unpredictable, and consequently, the performance is too.

This short series introduces a new port function attribute: max_io_eqs. The goal is to provide administrators at the hypervisor level with the ability to provision the maximum number of IO event queues for a function. This gives the control to the administrator to provision right number of IO event queues and have predictable performance.

Examples of when an administrator provisions (set) maximum number of IO event queues when using switchdev mode:

   $ devlink port show pci/0000:06:00.0/1
       pci/0000:06:00.0/1: type eth netdev enp6s0pf0vf0 flavour pcivf pfnum 0 vfnum 0
           function:
           hw_addr 00:00:00:00:00:00 roce enable max_io_eqs 10
   $ devlink port function set pci/0000:06:00.0/1 max_io_eqs 20
   $ devlink port show pci/0000:06:00.0/1
       pci/0000:06:00.0/1: type eth netdev enp6s0pf0vf0 flavour pcivf pfnum 0 vfnum 0
           function:
           hw_addr 00:00:00:00:00:00 roce enable max_io_eqs 20

This sets the corresponding maximum IO event queues of the function before it is enumerated. Thus, when the VF/SF driver reads the capability from the device, it sees the value provisioned by the hypervisor. The driver is then able to configure the number of channels for the net device, as well as the number of completion vectors for the RDMA device. The device/firmware also honors the provisioned value, hence any VF/SF driver attempting to create IO EQs beyond provisioned value results in an error.

Quoted 5 lines

With above setting now, the administrator is able to achieve the 2x performance on SFs with 20 channels. In second example when SF was provisioned for a container with 2 cpus, the administrator provisioned only 2 IO event queues, thereby saving device resources.

The following paragraph is the same with the above paragraph?

Quoted 31 lines

With the above settings now in place, the administrator achieved 2x performance with the SF device with 20 channels. In the second example, when the SF was provisioned for a container with 2 CPUs, the administrator provisioned only 2 IO event queues, thereby saving device resources.

changelog:
v2->v3:
- limited to 80 chars per line in devlink
- fixed comments from Jakub in mlx5 driver to fix missing mutex unlock on error path
v1->v2:
- limited comment to 80 chars per line in header file
- fixed set function variables for reverse christmas tree
- fixed comments from Kalesh
- fixed missing kfree in get call
- returning error code for get cmd failure
- fixed error msg copy paste error in set on cmd failure

Parav Pandit (2):
devlink: Support setting max_io_eqs
mlx5/core: Support max_io_eqs for a function

  .../networking/devlink/devlink-port.rst       | 33 +++++++
  .../mellanox/mlx5/core/esw/devlink_port.c     |  4 +
  .../net/ethernet/mellanox/mlx5/core/eswitch.h |  7 ++
  .../mellanox/mlx5/core/eswitch_offloads.c     | 97 +++++++++++++++++++
  include/net/devlink.h                         | 14 +++
  include/uapi/linux/devlink.h                  |  1 +
  net/devlink/port.c                            | 53 ++++++++++
  7 files changed, 209 insertions(+)
#6 lore raw
Quoted 47 lines

From: Zhu Yanjun <[email protected]>
Sent: Saturday, April 6, 2024 2:36 PM

Quoted 16 lines · Parav Pandit

Currently, PCI SFs and VFs use IO event queues to deliver netdev per channel events. The number of netdev channels is a function of IO event queues. In the second scenario of an RDMA device, the completion vectors are also a function of IO event queues. Currently, an administrator on the hypervisor has no means to provision the number of IO event queues for the SF device or the VF device. Device/firmware determines some arbitrary value for these IO event queues. Due to this, the SF netdev channels are unpredictable, and consequently, the performance is too.

This short series introduces a new port function attribute: max_io_eqs. The goal is to provide administrators at the hypervisor level with the ability to provision the maximum number of IO event queues for a function. This gives the control to the administrator to provision right number of IO event queues and have predictable performance.

Quoted 5 lines

Examples of when an administrator provisions (set) maximum number of IO event queues when using switchdev mode:

   $ devlink port show pci/0000:06:00.0/1
       pci/0000:06:00.0/1: type eth netdev enp6s0pf0vf0 flavour pcivf pfnum

0 vfnum 0

Quoted 7 lines
           function:
           hw_addr 00:00:00:00:00:00 roce enable max_io_eqs 10
   $ devlink port function set pci/0000:06:00.0/1 max_io_eqs 20
   $ devlink port show pci/0000:06:00.0/1
       pci/0000:06:00.0/1: type eth netdev enp6s0pf0vf0 flavour pcivf pfnum

0 vfnum 0

Quoted 12 lines
           function:
           hw_addr 00:00:00:00:00:00 roce enable max_io_eqs 20

This sets the corresponding maximum IO event queues of the function before it is enumerated. Thus, when the VF/SF driver reads the capability from the device, it sees the value provisioned by the hypervisor. The driver is then able to configure the number of channels for the net device, as well as the number of completion vectors for the RDMA device. The device/firmware also honors the provisioned value, hence any VF/SF driver attempting to create IO EQs beyond provisioned value results in an error.

Quoted 9 lines
Quoted 6 lines

With above setting now, the administrator is able to achieve the 2x performance on SFs with 20 channels. In second example when SF was provisioned for a container with 2 cpus, the administrator provisioned only
2 IO event queues, thereby saving device resources.

The following paragraph is the same with the above paragraph?

Ah, yes. I forgot to remove one of them while doing minor grammar changes.

Quoted 32 lines
Quoted 4 lines

With the above settings now in place, the administrator achieved 2x performance with the SF device with 20 channels. In the second example, when the SF was provisioned for a container with 2 CPUs, the administrator provisioned only 2 IO event queues, thereby saving device

resources.

Quoted 27 lines

changelog:
v2->v3:
- limited to 80 chars per line in devlink
- fixed comments from Jakub in mlx5 driver to fix missing mutex unlock on error path
v1->v2:
- limited comment to 80 chars per line in header file
- fixed set function variables for reverse christmas tree
- fixed comments from Kalesh
- fixed missing kfree in get call
- returning error code for get cmd failure
- fixed error msg copy paste error in set on cmd failure

Parav Pandit (2):
devlink: Support setting max_io_eqs
mlx5/core: Support max_io_eqs for a function

  .../networking/devlink/devlink-port.rst       | 33 +++++++
  .../mellanox/mlx5/core/esw/devlink_port.c     |  4 +
  .../net/ethernet/mellanox/mlx5/core/eswitch.h |  7 ++
  .../mellanox/mlx5/core/eswitch_offloads.c     | 97 +++++++++++++++++++
  include/net/devlink.h                         | 14 +++
  include/uapi/linux/devlink.h                  |  1 +
  net/devlink/port.c                            | 53 ++++++++++
  7 files changed, 209 insertions(+)
#7 lore raw
In reply to Parav Pandit
This series was applied to netdev/net-next.git (main) by David S. Miller <[email protected]>:

Hello:

This series was applied to netdev/net-next.git (main) by David S. Miller <[email protected]>:

Quoted 11 lines · On Sat, 6 Apr 2024 04:05:36 +0300 you wrote:

Currently, PCI SFs and VFs use IO event queues to deliver netdev per channel events. The number of netdev channels is a function of IO event queues. In the second scenario of an RDMA device, the completion vectors are also a function of IO event queues. Currently, an administrator on the hypervisor has no means to provision the number of IO event queues for the SF device or the VF device. Device/firmware determines some arbitrary value for these IO event queues. Due to this, the SF netdev channels are unpredictable, and consequently, the performance is too.

[...]

Here is the summary with links:
- [net-next,v4,1/2] devlink: Support setting max_io_eqs

    https://git.kernel.org/netdev/net-next/c/5af3e3876d56
  - [net-next,v4,2/2] mlx5/core: Support max_io_eqs for a function
    https://git.kernel.org/netdev/net-next/c/93197c7c509d

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html