| Message ID | 20240403174133.37587-2-parav@nvidia.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | devlink: Add port function attribute for IO EQs | expand |
On 2024-04-03 10:41, Parav Pandit wrote: > 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 > > Reviewed-by: Jiri Pirko <jiri@nvidia.com> > Reviewed-by: Shay Drory <shayd@nvidia.com> > Signed-off-by: Parav Pandit <parav@nvidia.com> > --- > changelog: > v2->v3: > - limited 80 chars per line > v1->v2: > - limited comment to 80 chars per line in header file > --- > .../networking/devlink/devlink-port.rst | 25 +++++++++ > include/net/devlink.h | 14 +++++ > include/uapi/linux/devlink.h | 1 + > net/devlink/port.c | 53 +++++++++++++++++++ > 4 files changed, 93 insertions(+) > > diff --git a/Documentation/networking/devlink/devlink-port.rst b/Documentation/networking/devlink/devlink-port.rst > index 562f46b41274..451f57393f11 100644 > --- a/Documentation/networking/devlink/devlink-port.rst > +++ b/Documentation/networking/devlink/devlink-port.rst > @@ -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 +298,28 @@ 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. > + > +- 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 > ============ > > diff --git a/include/net/devlink.h b/include/net/devlink.h > index 9ac394bdfbe4..bb1af599d101 100644 > --- a/include/net/devlink.h > +++ b/include/net/devlink.h > @@ -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, > diff --git a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h > index 2da0c7eb6710..9401aa343673 100644 > --- a/include/uapi/linux/devlink.h > +++ b/include/uapi/linux/devlink.h > @@ -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 > diff --git a/net/devlink/port.c b/net/devlink/port.c > index 118d130d2afd..be9158b4453c 100644 > --- a/net/devlink/port.c > +++ b/net/devlink/port.c > @@ -16,6 +16,7 @@ static const struct nla_policy devlink_function_nl_policy[DEVLINK_PORT_FUNCTION_ > DEVLINK_PORT_FN_STATE_ACTIVE), > [DEVLINK_PORT_FN_ATTR_CAPS] = > NLA_POLICY_BITFIELD32(DEVLINK_PORT_FN_CAPS_VALID_MASK), > + [DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] = { .type = NLA_U32 }, > }; > > #define ASSERT_DEVLINK_PORT_REGISTERED(devlink_port) \ > @@ -182,6 +183,30 @@ static int devlink_port_fn_caps_fill(struct devlink_port *devlink_port, > return 0; > } > > +static int devlink_port_fn_max_io_eqs_fill(struct devlink_port *port, > + struct sk_buff *msg, > + struct netlink_ext_ack *extack, > + bool *msg_updated) > +{ > + u32 max_io_eqs; > + int err; > + > + if (!port->ops->port_fn_max_io_eqs_get) > + return 0; > + > + err = port->ops->port_fn_max_io_eqs_get(port, &max_io_eqs, extack); > + if (err) { > + if (err == -EOPNOTSUPP) > + return 0; Docs above says: * Note: Driver should return -EOPNOTSUPP if it doesn't support * port function (@port_fn_*) handling for a particular port. But here you're returning 0 in both cases of no port_fn_max_io_eqs_get or port_fn_max_io_eqs_get() returns EOPNOTSUPP. > + return err; > + } > + err = nla_put_u32(msg, DEVLINK_PORT_FN_ATTR_MAX_IO_EQS, max_io_eqs); > + if (err) > + return err; > + *msg_updated = true; > + return 0; > +} > + > int devlink_nl_port_handle_fill(struct sk_buff *msg, struct devlink_port *devlink_port) > { > if (devlink_nl_put_handle(msg, devlink_port->devlink)) > @@ -409,6 +434,18 @@ static int devlink_port_fn_caps_set(struct devlink_port *devlink_port, > return 0; > } > > +static int > +devlink_port_fn_max_io_eqs_set(struct devlink_port *devlink_port, > + const struct nlattr *attr, > + struct netlink_ext_ack *extack) > +{ > + u32 max_io_eqs; > + > + max_io_eqs = nla_get_u32(attr); > + return devlink_port->ops->port_fn_max_io_eqs_set(devlink_port, > + max_io_eqs, extack); > +} > + > static int > devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct devlink_port *port, > struct netlink_ext_ack *extack) > @@ -428,6 +465,9 @@ devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct devlink_port *por > if (err) > goto out; > err = devlink_port_fn_state_fill(port, msg, extack, &msg_updated); > + if (err) > + goto out; > + err = devlink_port_fn_max_io_eqs_fill(port, msg, extack, &msg_updated); > if (err) > goto out; > err = devlink_rel_devlink_handle_put(msg, port->devlink, > @@ -726,6 +766,12 @@ static int devlink_port_function_validate(struct devlink_port *devlink_port, > } > } > } > + if (tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] && > + !ops->port_fn_max_io_eqs_set) { > + NL_SET_ERR_MSG_ATTR(extack, tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS], > + "Function does not support max_io_eqs setting"); > + return -EOPNOTSUPP; > + } > return 0; > } > > @@ -761,6 +807,13 @@ static int devlink_port_function_set(struct devlink_port *port, > return err; > } > > + attr = tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS]; > + if (attr) { > + err = devlink_port_fn_max_io_eqs_set(port, attr, extack); > + if (err) > + return err; > + } > + > /* Keep this as the last function attribute set, so that when > * multiple port function attributes are set along with state, > * Those can be applied first before activating the state.
Hi David, > From: David Wei <dw@davidwei.uk> > Sent: Thursday, April 4, 2024 10:29 PM > > On 2024-04-03 10:41, Parav Pandit wrote: > > 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 > > > > Reviewed-by: Jiri Pirko <jiri@nvidia.com> > > Reviewed-by: Shay Drory <shayd@nvidia.com> > > Signed-off-by: Parav Pandit <parav@nvidia.com> > > --- > > changelog: > > v2->v3: > > - limited 80 chars per line > > v1->v2: > > - limited comment to 80 chars per line in header file > > --- > > .../networking/devlink/devlink-port.rst | 25 +++++++++ > > include/net/devlink.h | 14 +++++ > > include/uapi/linux/devlink.h | 1 + > > net/devlink/port.c | 53 +++++++++++++++++++ > > 4 files changed, 93 insertions(+) > > > > diff --git a/Documentation/networking/devlink/devlink-port.rst > > b/Documentation/networking/devlink/devlink-port.rst > > index 562f46b41274..451f57393f11 100644 > > --- a/Documentation/networking/devlink/devlink-port.rst > > +++ b/Documentation/networking/devlink/devlink-port.rst > > @@ -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 +298,28 @@ 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. > > + > > +- 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 > > ============ > > > > diff --git a/include/net/devlink.h b/include/net/devlink.h index > > 9ac394bdfbe4..bb1af599d101 100644 > > --- a/include/net/devlink.h > > +++ b/include/net/devlink.h > > @@ -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, diff --git > > a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h index > > 2da0c7eb6710..9401aa343673 100644 > > --- a/include/uapi/linux/devlink.h > > +++ b/include/uapi/linux/devlink.h > > @@ -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 diff --git a/net/devlink/port.c b/net/devlink/port.c index > > 118d130d2afd..be9158b4453c 100644 > > --- a/net/devlink/port.c > > +++ b/net/devlink/port.c > > @@ -16,6 +16,7 @@ static const struct nla_policy > devlink_function_nl_policy[DEVLINK_PORT_FUNCTION_ > > DEVLINK_PORT_FN_STATE_ACTIVE), > > [DEVLINK_PORT_FN_ATTR_CAPS] = > > > NLA_POLICY_BITFIELD32(DEVLINK_PORT_FN_CAPS_VALID_MASK), > > + [DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] = { .type = NLA_U32 }, > > }; > > > > #define ASSERT_DEVLINK_PORT_REGISTERED(devlink_port) > \ > > @@ -182,6 +183,30 @@ static int devlink_port_fn_caps_fill(struct > devlink_port *devlink_port, > > return 0; > > } > > > > +static int devlink_port_fn_max_io_eqs_fill(struct devlink_port *port, > > + struct sk_buff *msg, > > + struct netlink_ext_ack *extack, > > + bool *msg_updated) > > +{ > > + u32 max_io_eqs; > > + int err; > > + > > + if (!port->ops->port_fn_max_io_eqs_get) > > + return 0; > > + > > + err = port->ops->port_fn_max_io_eqs_get(port, &max_io_eqs, > extack); > > + if (err) { > > + if (err == -EOPNOTSUPP) > > + return 0; > > Docs above says: > * Note: Driver should return -EOPNOTSUPP if it doesn't support > * port function (@port_fn_*) handling for a particular port. > > But here you're returning 0 in both cases of no port_fn_max_io_eqs_get or > port_fn_max_io_eqs_get() returns EOPNOTSUPP. > When the port does not support this op, the function pointer is null and, 0 is returned as expected. When the port for some reason has the ops function pointer set for a port, but if the port does not support the ops, it will return ENOPNOTSUPP. This may be possible when the driver has chosen to use same ops callback structure for multiple port flavors. This code pattern is likely left over code of relatively recent work that moved ops from devlink instance to per port ops. I propose to keep the current check as done in this patch, and run a full audit of all the drivers, if all drivers have moved to per port ops, then simplify the code to drop the check for EOPNOTSUPP in a new series that may touch more drivers. Otherwise, we may end up failing the port show operation when it returns - ENOPNOTSUPP. > > + return err; > > + } > > + err = nla_put_u32(msg, DEVLINK_PORT_FN_ATTR_MAX_IO_EQS, > max_io_eqs); > > + if (err) > > + return err; > > + *msg_updated = true; > > + return 0; > > +} > > + > > int devlink_nl_port_handle_fill(struct sk_buff *msg, struct > > devlink_port *devlink_port) { > > if (devlink_nl_put_handle(msg, devlink_port->devlink)) @@ -409,6 > > +434,18 @@ static int devlink_port_fn_caps_set(struct devlink_port > *devlink_port, > > return 0; > > } > > > > +static int > > +devlink_port_fn_max_io_eqs_set(struct devlink_port *devlink_port, > > + const struct nlattr *attr, > > + struct netlink_ext_ack *extack) { > > + u32 max_io_eqs; > > + > > + max_io_eqs = nla_get_u32(attr); > > + return devlink_port->ops->port_fn_max_io_eqs_set(devlink_port, > > + max_io_eqs, extack); > > +} > > + > > static int > > devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct > devlink_port *port, > > struct netlink_ext_ack *extack) @@ -428,6 > +465,9 @@ > > devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct > devlink_port *por > > if (err) > > goto out; > > err = devlink_port_fn_state_fill(port, msg, extack, &msg_updated); > > + if (err) > > + goto out; > > + err = devlink_port_fn_max_io_eqs_fill(port, msg, extack, > > +&msg_updated); > > if (err) > > goto out; > > err = devlink_rel_devlink_handle_put(msg, port->devlink, @@ -726,6 > > +766,12 @@ static int devlink_port_function_validate(struct devlink_port > *devlink_port, > > } > > } > > } > > + if (tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] && > > + !ops->port_fn_max_io_eqs_set) { > > + NL_SET_ERR_MSG_ATTR(extack, > tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS], > > + "Function does not support max_io_eqs > setting"); > > + return -EOPNOTSUPP; > > + } > > return 0; > > } > > > > @@ -761,6 +807,13 @@ static int devlink_port_function_set(struct > devlink_port *port, > > return err; > > } > > > > + attr = tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS]; > > + if (attr) { > > + err = devlink_port_fn_max_io_eqs_set(port, attr, extack); > > + if (err) > > + return err; > > + } > > + > > /* Keep this as the last function attribute set, so that when > > * multiple port function attributes are set along with state, > > * Those can be applied first before activating the state.
On 2024-04-04 10:58, Parav Pandit wrote: > Hi David, > >> From: David Wei <dw@davidwei.uk> >> Sent: Thursday, April 4, 2024 10:29 PM >> >> On 2024-04-03 10:41, Parav Pandit wrote: >>> 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 >>> >>> Reviewed-by: Jiri Pirko <jiri@nvidia.com> >>> Reviewed-by: Shay Drory <shayd@nvidia.com> >>> Signed-off-by: Parav Pandit <parav@nvidia.com> >>> --- >>> changelog: >>> v2->v3: >>> - limited 80 chars per line >>> v1->v2: >>> - limited comment to 80 chars per line in header file >>> --- >>> .../networking/devlink/devlink-port.rst | 25 +++++++++ >>> include/net/devlink.h | 14 +++++ >>> include/uapi/linux/devlink.h | 1 + >>> net/devlink/port.c | 53 +++++++++++++++++++ >>> 4 files changed, 93 insertions(+) >>> >>> diff --git a/Documentation/networking/devlink/devlink-port.rst >>> b/Documentation/networking/devlink/devlink-port.rst >>> index 562f46b41274..451f57393f11 100644 >>> --- a/Documentation/networking/devlink/devlink-port.rst >>> +++ b/Documentation/networking/devlink/devlink-port.rst >>> @@ -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 +298,28 @@ 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. >>> + >>> +- 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 >>> ============ >>> >>> diff --git a/include/net/devlink.h b/include/net/devlink.h index >>> 9ac394bdfbe4..bb1af599d101 100644 >>> --- a/include/net/devlink.h >>> +++ b/include/net/devlink.h >>> @@ -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, diff --git >>> a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h index >>> 2da0c7eb6710..9401aa343673 100644 >>> --- a/include/uapi/linux/devlink.h >>> +++ b/include/uapi/linux/devlink.h >>> @@ -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 diff --git a/net/devlink/port.c b/net/devlink/port.c index >>> 118d130d2afd..be9158b4453c 100644 >>> --- a/net/devlink/port.c >>> +++ b/net/devlink/port.c >>> @@ -16,6 +16,7 @@ static const struct nla_policy >> devlink_function_nl_policy[DEVLINK_PORT_FUNCTION_ >>> DEVLINK_PORT_FN_STATE_ACTIVE), >>> [DEVLINK_PORT_FN_ATTR_CAPS] = >>> >> NLA_POLICY_BITFIELD32(DEVLINK_PORT_FN_CAPS_VALID_MASK), >>> + [DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] = { .type = NLA_U32 }, >>> }; >>> >>> #define ASSERT_DEVLINK_PORT_REGISTERED(devlink_port) >> \ >>> @@ -182,6 +183,30 @@ static int devlink_port_fn_caps_fill(struct >> devlink_port *devlink_port, >>> return 0; >>> } >>> >>> +static int devlink_port_fn_max_io_eqs_fill(struct devlink_port *port, >>> + struct sk_buff *msg, >>> + struct netlink_ext_ack *extack, >>> + bool *msg_updated) >>> +{ >>> + u32 max_io_eqs; >>> + int err; >>> + >>> + if (!port->ops->port_fn_max_io_eqs_get) >>> + return 0; >>> + >>> + err = port->ops->port_fn_max_io_eqs_get(port, &max_io_eqs, >> extack); >>> + if (err) { >>> + if (err == -EOPNOTSUPP) >>> + return 0; >> >> Docs above says: >> * Note: Driver should return -EOPNOTSUPP if it doesn't support >> * port function (@port_fn_*) handling for a particular port. >> >> But here you're returning 0 in both cases of no port_fn_max_io_eqs_get or >> port_fn_max_io_eqs_get() returns EOPNOTSUPP. >> > When the port does not support this op, the function pointer is null and, 0 is returned as expected. > > When the port for some reason has the ops function pointer set for a port, but if the port does not support the ops, it will return ENOPNOTSUPP. > This may be possible when the driver has chosen to use same ops callback structure for multiple port flavors. > > This code pattern is likely left over code of relatively recent work that moved ops from devlink instance to per port ops. > I propose to keep the current check as done in this patch, > and run a full audit of all the drivers, if all drivers have moved to per port ops, then simplify the code to drop the check for EOPNOTSUPP in a new series that may touch more drivers. > Otherwise, we may end up failing the port show operation when it returns - ENOPNOTSUPP. Thanks for the explanation. So ideally each port flavour has its own unique set of struct ops, and if something is not supported then don't set the func ptr in the struct ops. Yes, I see that 0 has to be returned for devlink_port_fn_caps_fill() to succeed. Had a brief look and there's only a handful of drivers (mlx, nfp, ice) that use devlink_port_ops. > >>> + return err; >>> + } >>> + err = nla_put_u32(msg, DEVLINK_PORT_FN_ATTR_MAX_IO_EQS, >> max_io_eqs); >>> + if (err) >>> + return err; >>> + *msg_updated = true; >>> + return 0; >>> +} >>> + >>> int devlink_nl_port_handle_fill(struct sk_buff *msg, struct >>> devlink_port *devlink_port) { >>> if (devlink_nl_put_handle(msg, devlink_port->devlink)) @@ -409,6 >>> +434,18 @@ static int devlink_port_fn_caps_set(struct devlink_port >> *devlink_port, >>> return 0; >>> } >>> >>> +static int >>> +devlink_port_fn_max_io_eqs_set(struct devlink_port *devlink_port, >>> + const struct nlattr *attr, >>> + struct netlink_ext_ack *extack) { >>> + u32 max_io_eqs; >>> + >>> + max_io_eqs = nla_get_u32(attr); >>> + return devlink_port->ops->port_fn_max_io_eqs_set(devlink_port, >>> + max_io_eqs, extack); >>> +} >>> + >>> static int >>> devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct >> devlink_port *port, >>> struct netlink_ext_ack *extack) @@ -428,6 >> +465,9 @@ >>> devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct >> devlink_port *por >>> if (err) >>> goto out; >>> err = devlink_port_fn_state_fill(port, msg, extack, &msg_updated); >>> + if (err) >>> + goto out; >>> + err = devlink_port_fn_max_io_eqs_fill(port, msg, extack, >>> +&msg_updated); >>> if (err) >>> goto out; >>> err = devlink_rel_devlink_handle_put(msg, port->devlink, @@ -726,6 >>> +766,12 @@ static int devlink_port_function_validate(struct devlink_port >> *devlink_port, >>> } >>> } >>> } >>> + if (tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] && >>> + !ops->port_fn_max_io_eqs_set) { >>> + NL_SET_ERR_MSG_ATTR(extack, >> tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS], >>> + "Function does not support max_io_eqs >> setting"); >>> + return -EOPNOTSUPP; >>> + } >>> return 0; >>> } >>> >>> @@ -761,6 +807,13 @@ static int devlink_port_function_set(struct >> devlink_port *port, >>> return err; >>> } >>> >>> + attr = tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS]; >>> + if (attr) { >>> + err = devlink_port_fn_max_io_eqs_set(port, attr, extack); >>> + if (err) >>> + return err; >>> + } >>> + >>> /* Keep this as the last function attribute set, so that when >>> * multiple port function attributes are set along with state, >>> * Those can be applied first before activating the state.
> From: David Wei <dw@davidwei.uk> > Sent: Friday, April 5, 2024 12:10 AM > > On 2024-04-04 10:58, Parav Pandit wrote: > > Hi David, > > > >> From: David Wei <dw@davidwei.uk> > >> Sent: Thursday, April 4, 2024 10:29 PM > >> > >> On 2024-04-03 10:41, Parav Pandit wrote: > >>> 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 > >>> > >>> Reviewed-by: Jiri Pirko <jiri@nvidia.com> > >>> Reviewed-by: Shay Drory <shayd@nvidia.com> > >>> Signed-off-by: Parav Pandit <parav@nvidia.com> > >>> --- > >>> changelog: > >>> v2->v3: > >>> - limited 80 chars per line > >>> v1->v2: > >>> - limited comment to 80 chars per line in header file > >>> --- > >>> .../networking/devlink/devlink-port.rst | 25 +++++++++ > >>> include/net/devlink.h | 14 +++++ > >>> include/uapi/linux/devlink.h | 1 + > >>> net/devlink/port.c | 53 +++++++++++++++++++ > >>> 4 files changed, 93 insertions(+) > >>> > >>> diff --git a/Documentation/networking/devlink/devlink-port.rst > >>> b/Documentation/networking/devlink/devlink-port.rst > >>> index 562f46b41274..451f57393f11 100644 > >>> --- a/Documentation/networking/devlink/devlink-port.rst > >>> +++ b/Documentation/networking/devlink/devlink-port.rst > >>> @@ -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 +298,28 @@ 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. > >>> + > >>> +- 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 > >>> ============ > >>> > >>> diff --git a/include/net/devlink.h b/include/net/devlink.h index > >>> 9ac394bdfbe4..bb1af599d101 100644 > >>> --- a/include/net/devlink.h > >>> +++ b/include/net/devlink.h > >>> @@ -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, diff --git > >>> a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h index > >>> 2da0c7eb6710..9401aa343673 100644 > >>> --- a/include/uapi/linux/devlink.h > >>> +++ b/include/uapi/linux/devlink.h > >>> @@ -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 diff --git a/net/devlink/port.c b/net/devlink/port.c index > >>> 118d130d2afd..be9158b4453c 100644 > >>> --- a/net/devlink/port.c > >>> +++ b/net/devlink/port.c > >>> @@ -16,6 +16,7 @@ static const struct nla_policy > >> devlink_function_nl_policy[DEVLINK_PORT_FUNCTION_ > >>> DEVLINK_PORT_FN_STATE_ACTIVE), > >>> [DEVLINK_PORT_FN_ATTR_CAPS] = > >>> > >> NLA_POLICY_BITFIELD32(DEVLINK_PORT_FN_CAPS_VALID_MASK), > >>> + [DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] = { .type = NLA_U32 }, > >>> }; > >>> > >>> #define ASSERT_DEVLINK_PORT_REGISTERED(devlink_port) > >> \ > >>> @@ -182,6 +183,30 @@ static int devlink_port_fn_caps_fill(struct > >> devlink_port *devlink_port, > >>> return 0; > >>> } > >>> > >>> +static int devlink_port_fn_max_io_eqs_fill(struct devlink_port *port, > >>> + struct sk_buff *msg, > >>> + struct netlink_ext_ack *extack, > >>> + bool *msg_updated) > >>> +{ > >>> + u32 max_io_eqs; > >>> + int err; > >>> + > >>> + if (!port->ops->port_fn_max_io_eqs_get) > >>> + return 0; > >>> + > >>> + err = port->ops->port_fn_max_io_eqs_get(port, &max_io_eqs, > >> extack); > >>> + if (err) { > >>> + if (err == -EOPNOTSUPP) > >>> + return 0; > >> > >> Docs above says: > >> * Note: Driver should return -EOPNOTSUPP if it doesn't support > >> * port function (@port_fn_*) handling for a particular port. > >> > >> But here you're returning 0 in both cases of no > >> port_fn_max_io_eqs_get or > >> port_fn_max_io_eqs_get() returns EOPNOTSUPP. > >> > > When the port does not support this op, the function pointer is null and, 0 > is returned as expected. > > > > When the port for some reason has the ops function pointer set for a port, > but if the port does not support the ops, it will return ENOPNOTSUPP. > > This may be possible when the driver has chosen to use same ops callback > structure for multiple port flavors. > > > > This code pattern is likely left over code of relatively recent work that > moved ops from devlink instance to per port ops. > > I propose to keep the current check as done in this patch, and run a > > full audit of all the drivers, if all drivers have moved to per port ops, then > simplify the code to drop the check for EOPNOTSUPP in a new series that > may touch more drivers. > > Otherwise, we may end up failing the port show operation when it returns > - ENOPNOTSUPP. > > Thanks for the explanation. So ideally each port flavour has its own unique > set of struct ops, and if something is not supported then don't set the func > ptr in the struct ops. > > Yes, I see that 0 has to be returned for devlink_port_fn_caps_fill() to succeed. > > Had a brief look and there's only a handful of drivers (mlx, nfp, ice) that use > devlink_port_ops. > And netdevsim too. :) Can we please the cleanup in a separate series? Post this may be later in the month, I may have cycles to audit and simplify. Would it be ok with you? > > > >>> + return err; > >>> + } > >>> + err = nla_put_u32(msg, DEVLINK_PORT_FN_ATTR_MAX_IO_EQS, > >> max_io_eqs); > >>> + if (err) > >>> + return err; > >>> + *msg_updated = true; > >>> + return 0; > >>> +} > >>> + > >>> int devlink_nl_port_handle_fill(struct sk_buff *msg, struct > >>> devlink_port *devlink_port) { > >>> if (devlink_nl_put_handle(msg, devlink_port->devlink)) @@ -409,6 > >>> +434,18 @@ static int devlink_port_fn_caps_set(struct devlink_port > >> *devlink_port, > >>> return 0; > >>> } > >>> > >>> +static int > >>> +devlink_port_fn_max_io_eqs_set(struct devlink_port *devlink_port, > >>> + const struct nlattr *attr, > >>> + struct netlink_ext_ack *extack) { > >>> + u32 max_io_eqs; > >>> + > >>> + max_io_eqs = nla_get_u32(attr); > >>> + return devlink_port->ops->port_fn_max_io_eqs_set(devlink_port, > >>> + max_io_eqs, extack); > >>> +} > >>> + > >>> static int > >>> devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct > >> devlink_port *port, > >>> struct netlink_ext_ack *extack) @@ -428,6 > >> +465,9 @@ > >>> devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct > >> devlink_port *por > >>> if (err) > >>> goto out; > >>> err = devlink_port_fn_state_fill(port, msg, extack, &msg_updated); > >>> + if (err) > >>> + goto out; > >>> + err = devlink_port_fn_max_io_eqs_fill(port, msg, extack, > >>> +&msg_updated); > >>> if (err) > >>> goto out; > >>> err = devlink_rel_devlink_handle_put(msg, port->devlink, @@ -726,6 > >>> +766,12 @@ static int devlink_port_function_validate(struct > >>> +devlink_port > >> *devlink_port, > >>> } > >>> } > >>> } > >>> + if (tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] && > >>> + !ops->port_fn_max_io_eqs_set) { > >>> + NL_SET_ERR_MSG_ATTR(extack, > >> tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS], > >>> + "Function does not support max_io_eqs > >> setting"); > >>> + return -EOPNOTSUPP; > >>> + } > >>> return 0; > >>> } > >>> > >>> @@ -761,6 +807,13 @@ static int devlink_port_function_set(struct > >> devlink_port *port, > >>> return err; > >>> } > >>> > >>> + attr = tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS]; > >>> + if (attr) { > >>> + err = devlink_port_fn_max_io_eqs_set(port, attr, extack); > >>> + if (err) > >>> + return err; > >>> + } > >>> + > >>> /* Keep this as the last function attribute set, so that when > >>> * multiple port function attributes are set along with state, > >>> * Those can be applied first before activating the state.
On 2024-04-04 11:50, Parav Pandit wrote: > > >> From: David Wei <dw@davidwei.uk> >> Sent: Friday, April 5, 2024 12:10 AM >> >> On 2024-04-04 10:58, Parav Pandit wrote: >>> Hi David, >>> >>>> From: David Wei <dw@davidwei.uk> >>>> Sent: Thursday, April 4, 2024 10:29 PM >>>> >>>> On 2024-04-03 10:41, Parav Pandit wrote: >>>>> 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 >>>>> >>>>> Reviewed-by: Jiri Pirko <jiri@nvidia.com> >>>>> Reviewed-by: Shay Drory <shayd@nvidia.com> >>>>> Signed-off-by: Parav Pandit <parav@nvidia.com> >>>>> --- >>>>> changelog: >>>>> v2->v3: >>>>> - limited 80 chars per line >>>>> v1->v2: >>>>> - limited comment to 80 chars per line in header file >>>>> --- >>>>> .../networking/devlink/devlink-port.rst | 25 +++++++++ >>>>> include/net/devlink.h | 14 +++++ >>>>> include/uapi/linux/devlink.h | 1 + >>>>> net/devlink/port.c | 53 +++++++++++++++++++ >>>>> 4 files changed, 93 insertions(+) >>>>> >>>>> diff --git a/Documentation/networking/devlink/devlink-port.rst >>>>> b/Documentation/networking/devlink/devlink-port.rst >>>>> index 562f46b41274..451f57393f11 100644 >>>>> --- a/Documentation/networking/devlink/devlink-port.rst >>>>> +++ b/Documentation/networking/devlink/devlink-port.rst >>>>> @@ -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 +298,28 @@ 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. >>>>> + >>>>> +- 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 >>>>> ============ >>>>> >>>>> diff --git a/include/net/devlink.h b/include/net/devlink.h index >>>>> 9ac394bdfbe4..bb1af599d101 100644 >>>>> --- a/include/net/devlink.h >>>>> +++ b/include/net/devlink.h >>>>> @@ -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, diff --git >>>>> a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h index >>>>> 2da0c7eb6710..9401aa343673 100644 >>>>> --- a/include/uapi/linux/devlink.h >>>>> +++ b/include/uapi/linux/devlink.h >>>>> @@ -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 diff --git a/net/devlink/port.c b/net/devlink/port.c index >>>>> 118d130d2afd..be9158b4453c 100644 >>>>> --- a/net/devlink/port.c >>>>> +++ b/net/devlink/port.c >>>>> @@ -16,6 +16,7 @@ static const struct nla_policy >>>> devlink_function_nl_policy[DEVLINK_PORT_FUNCTION_ >>>>> DEVLINK_PORT_FN_STATE_ACTIVE), >>>>> [DEVLINK_PORT_FN_ATTR_CAPS] = >>>>> >>>> NLA_POLICY_BITFIELD32(DEVLINK_PORT_FN_CAPS_VALID_MASK), >>>>> + [DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] = { .type = NLA_U32 }, >>>>> }; >>>>> >>>>> #define ASSERT_DEVLINK_PORT_REGISTERED(devlink_port) >>>> \ >>>>> @@ -182,6 +183,30 @@ static int devlink_port_fn_caps_fill(struct >>>> devlink_port *devlink_port, >>>>> return 0; >>>>> } >>>>> >>>>> +static int devlink_port_fn_max_io_eqs_fill(struct devlink_port *port, >>>>> + struct sk_buff *msg, >>>>> + struct netlink_ext_ack *extack, >>>>> + bool *msg_updated) >>>>> +{ >>>>> + u32 max_io_eqs; >>>>> + int err; >>>>> + >>>>> + if (!port->ops->port_fn_max_io_eqs_get) >>>>> + return 0; >>>>> + >>>>> + err = port->ops->port_fn_max_io_eqs_get(port, &max_io_eqs, >>>> extack); >>>>> + if (err) { >>>>> + if (err == -EOPNOTSUPP) >>>>> + return 0; >>>> >>>> Docs above says: >>>> * Note: Driver should return -EOPNOTSUPP if it doesn't support >>>> * port function (@port_fn_*) handling for a particular port. >>>> >>>> But here you're returning 0 in both cases of no >>>> port_fn_max_io_eqs_get or >>>> port_fn_max_io_eqs_get() returns EOPNOTSUPP. >>>> >>> When the port does not support this op, the function pointer is null and, 0 >> is returned as expected. >>> >>> When the port for some reason has the ops function pointer set for a port, >> but if the port does not support the ops, it will return ENOPNOTSUPP. >>> This may be possible when the driver has chosen to use same ops callback >> structure for multiple port flavors. >>> >>> This code pattern is likely left over code of relatively recent work that >> moved ops from devlink instance to per port ops. >>> I propose to keep the current check as done in this patch, and run a >>> full audit of all the drivers, if all drivers have moved to per port ops, then >> simplify the code to drop the check for EOPNOTSUPP in a new series that >> may touch more drivers. >>> Otherwise, we may end up failing the port show operation when it returns >> - ENOPNOTSUPP. >> >> Thanks for the explanation. So ideally each port flavour has its own unique >> set of struct ops, and if something is not supported then don't set the func >> ptr in the struct ops. >> >> Yes, I see that 0 has to be returned for devlink_port_fn_caps_fill() to succeed. >> >> Had a brief look and there's only a handful of drivers (mlx, nfp, ice) that use >> devlink_port_ops. >> > And netdevsim too. :) > Can we please the cleanup in a separate series? > Post this may be later in the month, I may have cycles to audit and simplify. > Would it be ok with you? Yeah of course, thanks. I'm making some changes in netdevsim and can do a drive-by cleanup there too. > >>> >>>>> + return err; >>>>> + } >>>>> + err = nla_put_u32(msg, DEVLINK_PORT_FN_ATTR_MAX_IO_EQS, >>>> max_io_eqs); >>>>> + if (err) >>>>> + return err; >>>>> + *msg_updated = true; >>>>> + return 0; >>>>> +} >>>>> + >>>>> int devlink_nl_port_handle_fill(struct sk_buff *msg, struct >>>>> devlink_port *devlink_port) { >>>>> if (devlink_nl_put_handle(msg, devlink_port->devlink)) @@ -409,6 >>>>> +434,18 @@ static int devlink_port_fn_caps_set(struct devlink_port >>>> *devlink_port, >>>>> return 0; >>>>> } >>>>> >>>>> +static int >>>>> +devlink_port_fn_max_io_eqs_set(struct devlink_port *devlink_port, >>>>> + const struct nlattr *attr, >>>>> + struct netlink_ext_ack *extack) { >>>>> + u32 max_io_eqs; >>>>> + >>>>> + max_io_eqs = nla_get_u32(attr); >>>>> + return devlink_port->ops->port_fn_max_io_eqs_set(devlink_port, >>>>> + max_io_eqs, extack); >>>>> +} >>>>> + >>>>> static int >>>>> devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct >>>> devlink_port *port, >>>>> struct netlink_ext_ack *extack) @@ -428,6 >>>> +465,9 @@ >>>>> devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct >>>> devlink_port *por >>>>> if (err) >>>>> goto out; >>>>> err = devlink_port_fn_state_fill(port, msg, extack, &msg_updated); >>>>> + if (err) >>>>> + goto out; >>>>> + err = devlink_port_fn_max_io_eqs_fill(port, msg, extack, >>>>> +&msg_updated); >>>>> if (err) >>>>> goto out; >>>>> err = devlink_rel_devlink_handle_put(msg, port->devlink, @@ -726,6 >>>>> +766,12 @@ static int devlink_port_function_validate(struct >>>>> +devlink_port >>>> *devlink_port, >>>>> } >>>>> } >>>>> } >>>>> + if (tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] && >>>>> + !ops->port_fn_max_io_eqs_set) { >>>>> + NL_SET_ERR_MSG_ATTR(extack, >>>> tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS], >>>>> + "Function does not support max_io_eqs >>>> setting"); >>>>> + return -EOPNOTSUPP; >>>>> + } >>>>> return 0; >>>>> } >>>>> >>>>> @@ -761,6 +807,13 @@ static int devlink_port_function_set(struct >>>> devlink_port *port, >>>>> return err; >>>>> } >>>>> >>>>> + attr = tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS]; >>>>> + if (attr) { >>>>> + err = devlink_port_fn_max_io_eqs_set(port, attr, extack); >>>>> + if (err) >>>>> + return err; >>>>> + } >>>>> + >>>>> /* Keep this as the last function attribute set, so that when >>>>> * multiple port function attributes are set along with state, >>>>> * Those can be applied first before activating the state.
> From: David Wei <dw@davidwei.uk> > Sent: Friday, April 5, 2024 12:23 AM > > On 2024-04-04 11:50, Parav Pandit wrote: > > > > > >> From: David Wei <dw@davidwei.uk> > >> Sent: Friday, April 5, 2024 12:10 AM > >> > >> On 2024-04-04 10:58, Parav Pandit wrote: > >>> Hi David, > >>> > >>>> From: David Wei <dw@davidwei.uk> > >>>> Sent: Thursday, April 4, 2024 10:29 PM > >>>> > >>>> On 2024-04-03 10:41, Parav Pandit wrote: > >>>>> 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 > >>>>> > >>>>> Reviewed-by: Jiri Pirko <jiri@nvidia.com> > >>>>> Reviewed-by: Shay Drory <shayd@nvidia.com> > >>>>> Signed-off-by: Parav Pandit <parav@nvidia.com> > >>>>> --- > >>>>> changelog: > >>>>> v2->v3: > >>>>> - limited 80 chars per line > >>>>> v1->v2: > >>>>> - limited comment to 80 chars per line in header file > >>>>> --- > >>>>> .../networking/devlink/devlink-port.rst | 25 +++++++++ > >>>>> include/net/devlink.h | 14 +++++ > >>>>> include/uapi/linux/devlink.h | 1 + > >>>>> net/devlink/port.c | 53 +++++++++++++++++++ > >>>>> 4 files changed, 93 insertions(+) > >>>>> > >>>>> diff --git a/Documentation/networking/devlink/devlink-port.rst > >>>>> b/Documentation/networking/devlink/devlink-port.rst > >>>>> index 562f46b41274..451f57393f11 100644 > >>>>> --- a/Documentation/networking/devlink/devlink-port.rst > >>>>> +++ b/Documentation/networking/devlink/devlink-port.rst > >>>>> @@ -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 +298,28 @@ 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. > >>>>> + > >>>>> +- 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 > >>>>> ============ > >>>>> > >>>>> diff --git a/include/net/devlink.h b/include/net/devlink.h index > >>>>> 9ac394bdfbe4..bb1af599d101 100644 > >>>>> --- a/include/net/devlink.h > >>>>> +++ b/include/net/devlink.h > >>>>> @@ -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, diff --git > >>>>> a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h > >>>>> index > >>>>> 2da0c7eb6710..9401aa343673 100644 > >>>>> --- a/include/uapi/linux/devlink.h > >>>>> +++ b/include/uapi/linux/devlink.h > >>>>> @@ -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 diff --git a/net/devlink/port.c b/net/devlink/port.c index > >>>>> 118d130d2afd..be9158b4453c 100644 > >>>>> --- a/net/devlink/port.c > >>>>> +++ b/net/devlink/port.c > >>>>> @@ -16,6 +16,7 @@ static const struct nla_policy > >>>> devlink_function_nl_policy[DEVLINK_PORT_FUNCTION_ > >>>>> DEVLINK_PORT_FN_STATE_ACTIVE), > >>>>> [DEVLINK_PORT_FN_ATTR_CAPS] = > >>>>> > >>>> NLA_POLICY_BITFIELD32(DEVLINK_PORT_FN_CAPS_VALID_MASK), > >>>>> + [DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] = { .type = > NLA_U32 }, > >>>>> }; > >>>>> > >>>>> #define ASSERT_DEVLINK_PORT_REGISTERED(devlink_port) > >>>> \ > >>>>> @@ -182,6 +183,30 @@ static int devlink_port_fn_caps_fill(struct > >>>> devlink_port *devlink_port, > >>>>> return 0; > >>>>> } > >>>>> > >>>>> +static int devlink_port_fn_max_io_eqs_fill(struct devlink_port *port, > >>>>> + struct sk_buff *msg, > >>>>> + struct netlink_ext_ack > *extack, > >>>>> + bool *msg_updated) > >>>>> +{ > >>>>> + u32 max_io_eqs; > >>>>> + int err; > >>>>> + > >>>>> + if (!port->ops->port_fn_max_io_eqs_get) > >>>>> + return 0; > >>>>> + > >>>>> + err = port->ops->port_fn_max_io_eqs_get(port, > &max_io_eqs, > >>>> extack); > >>>>> + if (err) { > >>>>> + if (err == -EOPNOTSUPP) > >>>>> + return 0; > >>>> > >>>> Docs above says: > >>>> * Note: Driver should return -EOPNOTSUPP if it doesn't support > >>>> * port function (@port_fn_*) handling for a particular port. > >>>> > >>>> But here you're returning 0 in both cases of no > >>>> port_fn_max_io_eqs_get or > >>>> port_fn_max_io_eqs_get() returns EOPNOTSUPP. > >>>> > >>> When the port does not support this op, the function pointer is null > >>> and, 0 > >> is returned as expected. > >>> > >>> When the port for some reason has the ops function pointer set for a > >>> port, > >> but if the port does not support the ops, it will return ENOPNOTSUPP. > >>> This may be possible when the driver has chosen to use same ops > >>> callback > >> structure for multiple port flavors. > >>> > >>> This code pattern is likely left over code of relatively recent work > >>> that > >> moved ops from devlink instance to per port ops. > >>> I propose to keep the current check as done in this patch, and run a > >>> full audit of all the drivers, if all drivers have moved to per port > >>> ops, then > >> simplify the code to drop the check for EOPNOTSUPP in a new series > >> that may touch more drivers. > >>> Otherwise, we may end up failing the port show operation when it > >>> returns > >> - ENOPNOTSUPP. > >> > >> Thanks for the explanation. So ideally each port flavour has its own > >> unique set of struct ops, and if something is not supported then > >> don't set the func ptr in the struct ops. > >> > >> Yes, I see that 0 has to be returned for devlink_port_fn_caps_fill() to > succeed. > >> > >> Had a brief look and there's only a handful of drivers (mlx, nfp, > >> ice) that use devlink_port_ops. > >> > > And netdevsim too. :) > > Can we please the cleanup in a separate series? > > Post this may be later in the month, I may have cycles to audit and simplify. > > Would it be ok with you? > > Yeah of course, thanks. I'm making some changes in netdevsim and can do a > drive-by cleanup there too. > Ok. yes, that will be helpful. Thanks. > > > >>> > >>>>> + return err; > >>>>> + } > >>>>> + err = nla_put_u32(msg, > DEVLINK_PORT_FN_ATTR_MAX_IO_EQS, > >>>> max_io_eqs); > >>>>> + if (err) > >>>>> + return err; > >>>>> + *msg_updated = true; > >>>>> + return 0; > >>>>> +} > >>>>> + > >>>>> int devlink_nl_port_handle_fill(struct sk_buff *msg, struct > >>>>> devlink_port *devlink_port) { > >>>>> if (devlink_nl_put_handle(msg, devlink_port->devlink)) @@ -409,6 > >>>>> +434,18 @@ static int devlink_port_fn_caps_set(struct devlink_port > >>>> *devlink_port, > >>>>> return 0; > >>>>> } > >>>>> > >>>>> +static int > >>>>> +devlink_port_fn_max_io_eqs_set(struct devlink_port *devlink_port, > >>>>> + const struct nlattr *attr, > >>>>> + struct netlink_ext_ack *extack) { > >>>>> + u32 max_io_eqs; > >>>>> + > >>>>> + max_io_eqs = nla_get_u32(attr); > >>>>> + return devlink_port->ops- > >port_fn_max_io_eqs_set(devlink_port, > >>>>> + max_io_eqs, > extack); > >>>>> +} > >>>>> + > >>>>> static int > >>>>> devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct > >>>> devlink_port *port, > >>>>> struct netlink_ext_ack *extack) @@ -428,6 > >>>> +465,9 @@ > >>>>> devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct > >>>> devlink_port *por > >>>>> if (err) > >>>>> goto out; > >>>>> err = devlink_port_fn_state_fill(port, msg, extack, > >>>>> &msg_updated); > >>>>> + if (err) > >>>>> + goto out; > >>>>> + err = devlink_port_fn_max_io_eqs_fill(port, msg, extack, > >>>>> +&msg_updated); > >>>>> if (err) > >>>>> goto out; > >>>>> err = devlink_rel_devlink_handle_put(msg, port->devlink, @@ > >>>>> -726,6 > >>>>> +766,12 @@ static int devlink_port_function_validate(struct > >>>>> +devlink_port > >>>> *devlink_port, > >>>>> } > >>>>> } > >>>>> } > >>>>> + if (tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] && > >>>>> + !ops->port_fn_max_io_eqs_set) { > >>>>> + NL_SET_ERR_MSG_ATTR(extack, > >>>> tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS], > >>>>> + "Function does not support > max_io_eqs > >>>> setting"); > >>>>> + return -EOPNOTSUPP; > >>>>> + } > >>>>> return 0; > >>>>> } > >>>>> > >>>>> @@ -761,6 +807,13 @@ static int devlink_port_function_set(struct > >>>> devlink_port *port, > >>>>> return err; > >>>>> } > >>>>> > >>>>> + attr = tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS]; > >>>>> + if (attr) { > >>>>> + err = devlink_port_fn_max_io_eqs_set(port, attr, > extack); > >>>>> + if (err) > >>>>> + return err; > >>>>> + } > >>>>> + > >>>>> /* Keep this as the last function attribute set, so that when > >>>>> * multiple port function attributes are set along with state, > >>>>> * Those can be applied first before activating the state.
On Wed, 3 Apr 2024 20:41:32 +0300 Parav Pandit wrote: > 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. What's the relationship between EQ and queue pairs and IRQs? The next patch says "maximum IO event queues which are typically used to derive the maximum and default number of net device channels" It may not be obvious to non-mlx5 experts, I think it needs to be better documented.
On Thu, 4 Apr 2024 18:55:09 +0000 Parav Pandit wrote: > > Yeah of course, thanks. I'm making some changes in netdevsim and can do a > > drive-by cleanup there too. > > > Ok. yes, that will be helpful. Thanks. I think that part is fine, TBH. Drivers may not support particular control on particular HW/FW for any number of reasons, we should let them return EOPNOTSUPP. Not sure if there's much benefit for splitting the ops..
Hi Jakub, > From: Jakub Kicinski <kuba@kernel.org> > Sent: Friday, April 5, 2024 7:51 AM > > On Wed, 3 Apr 2024 20:41:32 +0300 Parav Pandit wrote: > > 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. > > What's the relationship between EQ and queue pairs and IRQs? Netdev qps (txq, rxq pair) channels created are typically upto num cpus by driver, provided it has enough IO event queues upto cpu count. Rdma qps are far more than netdev qps as multiple process uses them and they are per user space process resource. Those applications use number of QPs based on number of cpus and number of event channels to deliver notifications to user space. Driver uses the IRQs dynamically upto the PCI's limit based on supported IO event queues. > The next patch says "maximum IO event queues which are typically used to > derive the maximum and default number of net device channels" > It may not be obvious to non-mlx5 experts, I think it needs to be better > documented. I will expand the documentation in .../networking/devlink/devlink-port.rst. I will add below change to the v4 that has David's comments also addressed. Is this ok for you? --- a/Documentation/networking/devlink/devlink-port.rst +++ b/Documentation/networking/devlink/devlink-port.rst @@ -304,6 +304,11 @@ 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. +
On Fri, 5 Apr 2024 03:13:36 +0000 Parav Pandit wrote: > Netdev qps (txq, rxq pair) channels created are typically upto num cpus by driver, provided it has enough IO event queues upto cpu count. > > Rdma qps are far more than netdev qps as multiple process uses them and they are per user space process resource. > Those applications use number of QPs based on number of cpus and number of event channels to deliver notifications to user space. Some other drivers (e.g. intel) support multiple queues per core in netdev. For mlx5 I think AF_XDP may be a good example (or used to be) where there may be more than one queue? So I think the question still stands even for netdev. We should document whether number of EQs contains the number of Rx/Tx queues. > Driver uses the IRQs dynamically upto the PCI's limit based on supported IO event queues. Right but one IRQ <> one EQ? Typically / always? SFs "share" the IRQs with PF IIRC, do they share EQs? > > The next patch says "maximum IO event queues which are typically used to > > derive the maximum and default number of net device channels" > > It may not be obvious to non-mlx5 experts, I think it needs to be better > > documented. > I will expand the documentation in .../networking/devlink/devlink-port.rst. > > I will add below change to the v4 that has David's comments also addressed. > Is this ok for you? Looks like a good start but I think a few more sentences describing the relation to other resources would be good. > --- a/Documentation/networking/devlink/devlink-port.rst > +++ b/Documentation/networking/devlink/devlink-port.rst > @@ -304,6 +304,11 @@ 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.
> From: Jakub Kicinski <kuba@kernel.org> > Sent: Friday, April 5, 2024 7:44 PM > > On Fri, 5 Apr 2024 03:13:36 +0000 Parav Pandit wrote: > > Netdev qps (txq, rxq pair) channels created are typically upto num cpus by > driver, provided it has enough IO event queues upto cpu count. > > > > Rdma qps are far more than netdev qps as multiple process uses them and > they are per user space process resource. > > Those applications use number of QPs based on number of cpus and > number of event channels to deliver notifications to user space. > > Some other drivers (e.g. intel) support multiple queues per core in netdev. > For mlx5 I think AF_XDP may be a good example (or used to be) where there > may be more than one queue? > Yes, there may be multiple netdev queues which can be connected to a eq. For example, as you described, mlx5 xdp, also mlx5 creates a multiple txq per traffic class per channel which are linked to a single eq of the channel. But still those txq are per channel AFAIK. > So I think the question still stands even for netdev. > We should document whether number of EQs contains the number of Rx/Tx > queues. > I believe number of txq, rxqs can be more than the number of EQs connecting to same EQ. Netdev channels have more accurate linkage to EQs, compared to raw txq/rxqs. > > Driver uses the IRQs dynamically upto the PCI's limit based on supported IO > event queues. > > Right but one IRQ <> one EQ? Typically / always? Typically yes, one IRQ <> one EQ. > SFs "share" the IRQs with PF IIRC, do they share EQs? > SFs do not share EQs. Yes, SFs have their own dedicated EQs. You remember right, that they share IRQs. > > > The next patch says "maximum IO event queues which are typically > > > used to derive the maximum and default number of net device channels" > > > It may not be obvious to non-mlx5 experts, I think it needs to be > > > better documented. > > I will expand the documentation in .../networking/devlink/devlink-port.rst. > > > > I will add below change to the v4 that has David's comments also > addressed. > > Is this ok for you? > > Looks like a good start but I think a few more sentences describing the > relation to other resources would be good. > I think EQs limited object that does not have more wider relation in the stack. Relation to IRQ is probably a good addition to do. Along with below changes, will add the reference to IRQ too in v4. > > --- a/Documentation/networking/devlink/devlink-port.rst > > +++ b/Documentation/networking/devlink/devlink-port.rst > > @@ -304,6 +304,11 @@ 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.
diff --git a/Documentation/networking/devlink/devlink-port.rst b/Documentation/networking/devlink/devlink-port.rst index 562f46b41274..451f57393f11 100644 --- a/Documentation/networking/devlink/devlink-port.rst +++ b/Documentation/networking/devlink/devlink-port.rst @@ -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 +298,28 @@ 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. + +- 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 ============ diff --git a/include/net/devlink.h b/include/net/devlink.h index 9ac394bdfbe4..bb1af599d101 100644 --- a/include/net/devlink.h +++ b/include/net/devlink.h @@ -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, diff --git a/include/uapi/linux/devlink.h b/include/uapi/linux/devlink.h index 2da0c7eb6710..9401aa343673 100644 --- a/include/uapi/linux/devlink.h +++ b/include/uapi/linux/devlink.h @@ -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 diff --git a/net/devlink/port.c b/net/devlink/port.c index 118d130d2afd..be9158b4453c 100644 --- a/net/devlink/port.c +++ b/net/devlink/port.c @@ -16,6 +16,7 @@ static const struct nla_policy devlink_function_nl_policy[DEVLINK_PORT_FUNCTION_ DEVLINK_PORT_FN_STATE_ACTIVE), [DEVLINK_PORT_FN_ATTR_CAPS] = NLA_POLICY_BITFIELD32(DEVLINK_PORT_FN_CAPS_VALID_MASK), + [DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] = { .type = NLA_U32 }, }; #define ASSERT_DEVLINK_PORT_REGISTERED(devlink_port) \ @@ -182,6 +183,30 @@ static int devlink_port_fn_caps_fill(struct devlink_port *devlink_port, return 0; } +static int devlink_port_fn_max_io_eqs_fill(struct devlink_port *port, + struct sk_buff *msg, + struct netlink_ext_ack *extack, + bool *msg_updated) +{ + u32 max_io_eqs; + int err; + + if (!port->ops->port_fn_max_io_eqs_get) + return 0; + + err = port->ops->port_fn_max_io_eqs_get(port, &max_io_eqs, extack); + if (err) { + if (err == -EOPNOTSUPP) + return 0; + return err; + } + err = nla_put_u32(msg, DEVLINK_PORT_FN_ATTR_MAX_IO_EQS, max_io_eqs); + if (err) + return err; + *msg_updated = true; + return 0; +} + int devlink_nl_port_handle_fill(struct sk_buff *msg, struct devlink_port *devlink_port) { if (devlink_nl_put_handle(msg, devlink_port->devlink)) @@ -409,6 +434,18 @@ static int devlink_port_fn_caps_set(struct devlink_port *devlink_port, return 0; } +static int +devlink_port_fn_max_io_eqs_set(struct devlink_port *devlink_port, + const struct nlattr *attr, + struct netlink_ext_ack *extack) +{ + u32 max_io_eqs; + + max_io_eqs = nla_get_u32(attr); + return devlink_port->ops->port_fn_max_io_eqs_set(devlink_port, + max_io_eqs, extack); +} + static int devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct devlink_port *port, struct netlink_ext_ack *extack) @@ -428,6 +465,9 @@ devlink_nl_port_function_attrs_put(struct sk_buff *msg, struct devlink_port *por if (err) goto out; err = devlink_port_fn_state_fill(port, msg, extack, &msg_updated); + if (err) + goto out; + err = devlink_port_fn_max_io_eqs_fill(port, msg, extack, &msg_updated); if (err) goto out; err = devlink_rel_devlink_handle_put(msg, port->devlink, @@ -726,6 +766,12 @@ static int devlink_port_function_validate(struct devlink_port *devlink_port, } } } + if (tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS] && + !ops->port_fn_max_io_eqs_set) { + NL_SET_ERR_MSG_ATTR(extack, tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS], + "Function does not support max_io_eqs setting"); + return -EOPNOTSUPP; + } return 0; } @@ -761,6 +807,13 @@ static int devlink_port_function_set(struct devlink_port *port, return err; } + attr = tb[DEVLINK_PORT_FN_ATTR_MAX_IO_EQS]; + if (attr) { + err = devlink_port_fn_max_io_eqs_set(port, attr, extack); + if (err) + return err; + } + /* Keep this as the last function attribute set, so that when * multiple port function attributes are set along with state, * Those can be applied first before activating the state.
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 Reviewed-by: Jiri Pirko <jiri@nvidia.com> Reviewed-by: Shay Drory <shayd@nvidia.com> Signed-off-by: Parav Pandit <parav@nvidia.com> --- changelog: v2->v3: - limited 80 chars per line v1->v2: - limited comment to 80 chars per line in header file --- .../networking/devlink/devlink-port.rst | 25 +++++++++ include/net/devlink.h | 14 +++++ include/uapi/linux/devlink.h | 1 + net/devlink/port.c | 53 +++++++++++++++++++ 4 files changed, 93 insertions(+)