[v4,1/6] ethdev: add capability to keep flow rules on restart

  Previously, it was not specified what happens to the flow rules
when the device is stopped, possibly reconfigured, then started.
If flow rules were kept, it could be convenient for application
developers, because they wouldn't need to save and restore them.
However, due to the number of flows and possible creation rate it is
impractical to save all flow rules in DPDK layer. This means that flow
rules persistence really depends on whether PMD and HW can implement it
efficiently. It can also be limited by the rule item and action types,
and its attributes transfer bit (a combination of an item/action type
and a value of the transfer bit is called a ruel feature).

Add a device capability bit for PMDs that can keep at least some
of the flow rules across restart. Without this capability behavior
is still unspecified and it is declared that the application must
flush the rules before stopping the device.
Allow the application to test for persistence of rules using
a particular feature by attempting to create a flow rule
using that feature when the device is stopped
and checking for the specific error.
This is logical because if the PMD can to create the flow rule
when the device is not started and use it after the start happens,
it is natural that it can move its internal flow rule object
to the same state when the device is stopped and restore the state
when the device is started.

Rule persistence across a reconfigurations is not required,
because tracking all the rules and configuration-dependent resources
they use may be infeasible. In case a PMD cannot keep the rules
across reconfiguration, it is allowed just to report an error.
Application must then flush the rules before attempting it.

Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
---
 doc/guides/prog_guide/rte_flow.rst | 31 ++++++++++++++++++++++++++++++
 lib/ethdev/rte_ethdev.h            |  7 +++++++
 lib/ethdev/rte_flow.h              |  1 +
 3 files changed, 39 insertions(+)
  

Hi Dmitry,

> -----Original Message-----
> From: dev <dev-bounces@dpdk.org> On Behalf Of Dmitry Kozlyuk
> Sent: Thursday, October 21, 2021 9:35 AM
> To: dev@dpdk.org
> Cc: Ori Kam <orika@oss.nvidia.com>; NBU-Contact-Thomas Monjalon <thomas@monjalon.net>;
> Ferruh Yigit <ferruh.yigit@intel.com>; Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
> Subject: [dpdk-dev] [PATCH v4 1/6] ethdev: add capability to keep flow rules on restart
> 
> Previously, it was not specified what happens to the flow rules
> when the device is stopped, possibly reconfigured, then started.
> If flow rules were kept, it could be convenient for application
> developers, because they wouldn't need to save and restore them.
> However, due to the number of flows and possible creation rate it is
> impractical to save all flow rules in DPDK layer. This means that flow
> rules persistence really depends on whether PMD and HW can implement it
> efficiently. It can also be limited by the rule item and action types,
> and its attributes transfer bit (a combination of an item/action type
> and a value of the transfer bit is called a ruel feature).
> 
> Add a device capability bit for PMDs that can keep at least some
> of the flow rules across restart. Without this capability behavior
> is still unspecified and it is declared that the application must
> flush the rules before stopping the device.
> Allow the application to test for persistence of rules using
> a particular feature by attempting to create a flow rule
> using that feature when the device is stopped
> and checking for the specific error.
> This is logical because if the PMD can to create the flow rule
> when the device is not started and use it after the start happens,
> it is natural that it can move its internal flow rule object
> to the same state when the device is stopped and restore the state
> when the device is started.
> 
> Rule persistence across a reconfigurations is not required,
> because tracking all the rules and configuration-dependent resources
> they use may be infeasible. In case a PMD cannot keep the rules
> across reconfiguration, it is allowed just to report an error.
> Application must then flush the rules before attempting it.
> 
> Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
> ---
>  doc/guides/prog_guide/rte_flow.rst | 31 ++++++++++++++++++++++++++++++
>  lib/ethdev/rte_ethdev.h            |  7 +++++++
>  lib/ethdev/rte_flow.h              |  1 +
>  3 files changed, 39 insertions(+)
> 
> diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst
> index aeba374182..9beaae3df3 100644
> --- a/doc/guides/prog_guide/rte_flow.rst
> +++ b/doc/guides/prog_guide/rte_flow.rst
> @@ -87,6 +87,37 @@ To avoid resource leaks on the PMD side, handles must be explicitly
>  destroyed by the application before releasing associated resources such as
>  queues and ports.
> 
> +When the device is stopped, its rules do not process the traffic.
> +In particular, transfer rules created using some device
> +stop affecting the traffic even if they refer to different ports.
> +
> +If ``RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP`` is not advertised,
> +rules cannot be created until the device is started for the first time
> +and cannot be kept when the device is stopped.
> +However, PMD also does not flush them automatically on stop,
> +so the application must call ``rte_flow_flush()`` or ``rte_flow_destroy()``
> +before stopping the device to ensure no rules remain.
> +
> +If ``RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP`` is advertised, this means
> +the PMD can keep at least some rules across the device stop and start.
> +However, ``rte_eth_dev_configure()`` may fail if any rules remain,
> +so the application must flush them before attempting a reconfiguration.
> +Keeping may be unsupported for some types of rule items and actions,
> +as well as depending on the value of flow attributes transfer bit.
> +A combination of a single an item or action type
> +and a value of the transfer bit is called a rule feature.
> +For example: a COUNT action with the transfer bit set.
> +To test if rules with a particular feature are kept, the application must try
> +to create a valid rule using this feature when the device is not started
> +(either before the first start or after a stop).
> +If it fails with an error of type ``RTE_FLOW_ERROR_TYPE_STATE``,
> +all rules using this feature must be flushed by the application
> +before stopping the device.
> +If it succeeds, such rules will be kept when the device is stopped,
> +provided they do not use other features that are not supported.
> +Rules that are created when the device is stopped, including the rules
> +created for the test, will be kept after the device is started.
> +
>  The following sections cover:
> 
>  - **Attributes** (represented by ``struct rte_flow_attr``): properties of a
> diff --git a/lib/ethdev/rte_ethdev.h b/lib/ethdev/rte_ethdev.h
> index 014270d316..9cf23fecce 100644
> --- a/lib/ethdev/rte_ethdev.h
> +++ b/lib/ethdev/rte_ethdev.h
> @@ -90,6 +90,11 @@
>   *     - flow director filtering mode (but not filtering rules)
>   *     - NIC queue statistics mappings
>   *
> + * The following configuration may be retained or not
> + * depending on the device capabilities:
> + *
> + *     - flow rules
> + *
>   * Any other configuration will not be stored and will need to be re-entered
>   * before a call to rte_eth_dev_start().
>   *
> @@ -1445,6 +1450,8 @@ struct rte_eth_conf {
>  #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
>  /** Device supports Tx queue setup after device started. */
>  #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
> +/** Device supports keeping flow rules across restart. */
> +#define RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP 0x00000004
>  /**@}*/
> 
>  /*
> diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h
> index 2b6efeef8c..16ef33819b 100644
> --- a/lib/ethdev/rte_flow.h
> +++ b/lib/ethdev/rte_flow.h
> @@ -3676,6 +3676,7 @@ enum rte_flow_error_type {
>  	RTE_FLOW_ERROR_TYPE_ACTION_NUM, /**< Number of actions. */
>  	RTE_FLOW_ERROR_TYPE_ACTION_CONF, /**< Action configuration. */
>  	RTE_FLOW_ERROR_TYPE_ACTION, /**< Specific action. */
> +	RTE_FLOW_ERROR_TYPE_STATE, /**< Current device state. */
>  };
> 
>  /**
> --
> 2.25.1

Acked-by: Ori Kam <orika@nvidia.com>
Best,
Ori
  

On Wed, Oct 20, 2021 at 11:35 PM Dmitry Kozlyuk <dkozlyuk@oss.nvidia.com> wrote:
>
> Previously, it was not specified what happens to the flow rules
> when the device is stopped, possibly reconfigured, then started.
> If flow rules were kept, it could be convenient for application
> developers, because they wouldn't need to save and restore them.
> However, due to the number of flows and possible creation rate it is
> impractical to save all flow rules in DPDK layer. This means that flow
> rules persistence really depends on whether PMD and HW can implement it
> efficiently. It can also be limited by the rule item and action types,
> and its attributes transfer bit (a combination of an item/action type
> and a value of the transfer bit is called a ruel feature).
>
> Add a device capability bit for PMDs that can keep at least some
> of the flow rules across restart. Without this capability behavior
> is still unspecified and it is declared that the application must
> flush the rules before stopping the device.
> Allow the application to test for persistence of rules using
> a particular feature by attempting to create a flow rule
> using that feature when the device is stopped
> and checking for the specific error.
> This is logical because if the PMD can to create the flow rule
> when the device is not started and use it after the start happens,
> it is natural that it can move its internal flow rule object
> to the same state when the device is stopped and restore the state
> when the device is started.
>
> Rule persistence across a reconfigurations is not required,
> because tracking all the rules and configuration-dependent resources
> they use may be infeasible. In case a PMD cannot keep the rules
> across reconfiguration, it is allowed just to report an error.
> Application must then flush the rules before attempting it.
>
> Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>

Acked-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
  

On 10/21/21 9:34 AM, Dmitry Kozlyuk wrote:
> Previously, it was not specified what happens to the flow rules
> when the device is stopped, possibly reconfigured, then started.
> If flow rules were kept, it could be convenient for application
> developers, because they wouldn't need to save and restore them.
> However, due to the number of flows and possible creation rate it is
> impractical to save all flow rules in DPDK layer. This means that flow
> rules persistence really depends on whether PMD and HW can implement it
> efficiently. It can also be limited by the rule item and action types,
> and its attributes transfer bit (a combination of an item/action type
> and a value of the transfer bit is called a ruel feature).
> 
> Add a device capability bit for PMDs that can keep at least some
> of the flow rules across restart. Without this capability behavior
> is still unspecified and it is declared that the application must
> flush the rules before stopping the device.
> Allow the application to test for persistence of rules using
> a particular feature by attempting to create a flow rule
> using that feature when the device is stopped
> and checking for the specific error.
> This is logical because if the PMD can to create the flow rule
> when the device is not started and use it after the start happens,
> it is natural that it can move its internal flow rule object
> to the same state when the device is stopped and restore the state
> when the device is started.
> 
> Rule persistence across a reconfigurations is not required,
> because tracking all the rules and configuration-dependent resources
> they use may be infeasible. In case a PMD cannot keep the rules
> across reconfiguration, it is allowed just to report an error.
> Application must then flush the rules before attempting it.
> 
> Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>

Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>

with some review notes below

> ---
>   doc/guides/prog_guide/rte_flow.rst | 31 ++++++++++++++++++++++++++++++
>   lib/ethdev/rte_ethdev.h            |  7 +++++++
>   lib/ethdev/rte_flow.h              |  1 +
>   3 files changed, 39 insertions(+)
> 
> diff --git a/doc/guides/prog_guide/rte_flow.rst b/doc/guides/prog_guide/rte_flow.rst
> index aeba374182..9beaae3df3 100644
> --- a/doc/guides/prog_guide/rte_flow.rst
> +++ b/doc/guides/prog_guide/rte_flow.rst
> @@ -87,6 +87,37 @@ To avoid resource leaks on the PMD side, handles must be explicitly
>   destroyed by the application before releasing associated resources such as
>   queues and ports.
>   
> +When the device is stopped, its rules do not process the traffic.
> +In particular, transfer rules created using some device
> +stop affecting the traffic even if they refer to different ports.
> +
> +If ``RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP`` is not advertised,
> +rules cannot be created until the device is started for the first time
> +and cannot be kept when the device is stopped.
> +However, PMD also does not flush them automatically on stop,
> +so the application must call ``rte_flow_flush()`` or ``rte_flow_destroy()``
> +before stopping the device to ensure no rules remain.
> +
> +If ``RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP`` is advertised, this means
> +the PMD can keep at least some rules across the device stop and start.
> +However, ``rte_eth_dev_configure()`` may fail if any rules remain,
> +so the application must flush them before attempting a reconfiguration.
> +Keeping may be unsupported for some types of rule items and actions,
> +as well as depending on the value of flow attributes transfer bit.
> +A combination of a single an item or action type
> +and a value of the transfer bit is called a rule feature.
> +For example: a COUNT action with the transfer bit set.
> +To test if rules with a particular feature are kept, the application must try
> +to create a valid rule using this feature when the device is not started
> +(either before the first start or after a stop).
> +If it fails with an error of type ``RTE_FLOW_ERROR_TYPE_STATE``,
> +all rules using this feature must be flushed by the application
> +before stopping the device.
> +If it succeeds, such rules will be kept when the device is stopped,
> +provided they do not use other features that are not supported.
> +Rules that are created when the device is stopped, including the rules
> +created for the test, will be kept after the device is started.
> +
>   The following sections cover:
>   
>   - **Attributes** (represented by ``struct rte_flow_attr``): properties of a
> diff --git a/lib/ethdev/rte_ethdev.h b/lib/ethdev/rte_ethdev.h
> index 014270d316..9cf23fecce 100644
> --- a/lib/ethdev/rte_ethdev.h
> +++ b/lib/ethdev/rte_ethdev.h
> @@ -90,6 +90,11 @@
>    *     - flow director filtering mode (but not filtering rules)
>    *     - NIC queue statistics mappings
>    *
> + * The following configuration may be retained or not
> + * depending on the device capabilities:
> + *
> + *     - flow rules
> + *
>    * Any other configuration will not be stored and will need to be re-entered
>    * before a call to rte_eth_dev_start().
>    *
> @@ -1445,6 +1450,8 @@ struct rte_eth_conf {
>   #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
>   /** Device supports Tx queue setup after device started. */
>   #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
> +/** Device supports keeping flow rules across restart. */
> +#define RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP 0x00000004

RTE_BIT64(2) since previous two are already defined using RTE_BIT32()
in next-net

Don't we need an experimental markup in the documentation to
make it possible to refine the future in the nearest future
without API breakage? If yes, it must be mentioned in the
rte_flow.rst documentation as well.

>   /**@}*/
>   
>   /*
> diff --git a/lib/ethdev/rte_flow.h b/lib/ethdev/rte_flow.h
> index 2b6efeef8c..16ef33819b 100644
> --- a/lib/ethdev/rte_flow.h
> +++ b/lib/ethdev/rte_flow.h
> @@ -3676,6 +3676,7 @@ enum rte_flow_error_type {
>   	RTE_FLOW_ERROR_TYPE_ACTION_NUM, /**< Number of actions. */
>   	RTE_FLOW_ERROR_TYPE_ACTION_CONF, /**< Action configuration. */
>   	RTE_FLOW_ERROR_TYPE_ACTION, /**< Specific action. */
> +	RTE_FLOW_ERROR_TYPE_STATE, /**< Current device state. */
>   };
>   
>   /**
>
  

> > @@ -1445,6 +1450,8 @@ struct rte_eth_conf {
> >   #define RTE_ETH_DEV_CAPA_RUNTIME_RX_QUEUE_SETUP 0x00000001
> >   /** Device supports Tx queue setup after device started. */
> >   #define RTE_ETH_DEV_CAPA_RUNTIME_TX_QUEUE_SETUP 0x00000002
> > +/** Device supports keeping flow rules across restart. */
> > +#define RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP 0x00000004
> 
> RTE_BIT64(2) since previous two are already defined using RTE_BIT32()
> in next-net
> 
> Don't we need an experimental markup in the documentation to
> make it possible to refine the future in the nearest future
> without API breakage? If yes, it must be mentioned in the
> rte_flow.rst documentation as well.

It seems constants are not usually marked as experimental,
but I will add warnings to rte_flow.rst anyway.