[v3] doc: update flow non-cached mode notes
Checks
Commit Message
This patch updates the MLX5 PMD and release notes documentations.
Adding the notes of the behavior change that rte flows organization
is switched into non-cached mode for applications.
Signed-off-by: Bing Zhao <bingz@mellanox.com>
---
v2: update the description in release notes.
v3: update the notes of rte_flow for application.
---
doc/guides/nics/mlx5.rst | 13 +++++++++++++
doc/guides/rel_notes/release_20_05.rst | 2 ++
2 files changed, 15 insertions(+)
Comments
> -----Original Message-----
> From: Bing Zhao <bingz@mellanox.com>
> Sent: Tuesday, April 21, 2020 9:08 AM
> To: john.mcnamara@intel.com; marko.kovacevic@intel.com; Slava Ovsiienko
> <viacheslavo@mellanox.com>; Ori Kam <orika@mellanox.com>; Matan Azrad
> <matan@mellanox.com>
> Cc: Shahaf Shuler <shahafs@mellanox.com>; Raslan Darawsheh
> <rasland@mellanox.com>; dev@dpdk.org; Thomas Monjalon
> <thomas@monjalon.net>
> Subject: [PATCH v3] doc: update flow non-cached mode notes
>
> This patch updates the MLX5 PMD and release notes documentations.
> Adding the notes of the behavior change that rte flows organization
> is switched into non-cached mode for applications.
>
> Signed-off-by: Bing Zhao <bingz@mellanox.com>
> ---
> v2: update the description in release notes.
> v3: update the notes of rte_flow for application.
> ---
Acked-by: Ori Kam <orika@mellanox.com>
Thanks,
Ori
> doc/guides/nics/mlx5.rst | 13 +++++++++++++
> doc/guides/rel_notes/release_20_05.rst | 2 ++
> 2 files changed, 15 insertions(+)
>
> diff --git a/doc/guides/nics/mlx5.rst b/doc/guides/nics/mlx5.rst
> index 759d0ac..ced5e03 100644
> --- a/doc/guides/nics/mlx5.rst
> +++ b/doc/guides/nics/mlx5.rst
> @@ -1300,6 +1300,19 @@ set, and we should allow to specify zero values as
> rte_flow parameters for the
> META and MARK items and actions. In the same time zero mask has no
> meaning and
> should be rejected on validation stage.
>
> +Notes for rte_flow
> +------------------
> +Implementation of flows organization is in non-cached mode. When stopping a
> +device, all the flows inserted from application will be considered invalid and
> +flushed automatically in the background. After restarting of the device, no
> +application's flow exists in the system. All reference pointers to these flows
> +held by the application are not valid anymore and should be discarded by the
> +application.
> +
> +The application should recreate the flows as required after device restarting,
> +and must not try to destroy or flush the invalid flows after stopping (except
> +discarding them).
> +
> Notes for testpmd
> -----------------
>
> diff --git a/doc/guides/rel_notes/release_20_05.rst
> b/doc/guides/rel_notes/release_20_05.rst
> index bacd4c6..367c5df 100644
> --- a/doc/guides/rel_notes/release_20_05.rst
> +++ b/doc/guides/rel_notes/release_20_05.rst
> @@ -63,6 +63,8 @@ New Features
> * Added support for matching on IPv4 Time To Live and IPv6 Hop Limit.
> * Added support for creating Relaxed Ordering Memory Regions.
> * Added support for jumbo frame size (9K MTU) in Multi-Packet RQ mode.
> + * Flow rules caching removed from the driver for memory saving and
> + correctness according to RTE flow and RTE ethdev.
> * Optimized the memory consumption of flow.
>
> * **Updated the Intel ice driver.**
> --
> 1.8.3.1
Hi,
Please prefer splitting lines after punctuation signs.
21/04/2020 08:08, Bing Zhao:
> --- a/doc/guides/nics/mlx5.rst
> +++ b/doc/guides/nics/mlx5.rst
>
> +Notes for rte_flow
A blank line is missing below:
> +------------------
> +Implementation of flows organization is in non-cached mode. When stopping a
This can be simpler in direct form:
Flows are not cached in the driver.
> +device, all the flows inserted from application will be considered invalid and
device -> device port
> +flushed automatically in the background. After restarting of the device, no
restarting of the device -> restarting the device
> +application's flow exists in the system. All reference pointers to these flows
Why "application's flow"?
What about "no flow is represented in the system"?
reference pointers -> references
> +held by the application are not valid anymore and should be discarded by the
> +application.
Invalid and discard explanations are repeated below.
> +
> +The application should recreate the flows as required after device restarting,
device restarting -> port restart
Better to stop the sentence here, and move it at the end of the paragraph after "stop".
> +and must not try to destroy or flush the invalid flows after stopping (except
After stopping, flows become invalid.
They should be discarded but neither destroyed nor flushed.
> +discarding them).
> +
> Notes for testpmd
> -----------------
>
> diff --git a/doc/guides/rel_notes/release_20_05.rst b/doc/guides/rel_notes/release_20_05.rst
> index bacd4c6..367c5df 100644
> --- a/doc/guides/rel_notes/release_20_05.rst
> +++ b/doc/guides/rel_notes/release_20_05.rst
> @@ -63,6 +63,8 @@ New Features
> * Added support for matching on IPv4 Time To Live and IPv6 Hop Limit.
> * Added support for creating Relaxed Ordering Memory Regions.
> * Added support for jumbo frame size (9K MTU) in Multi-Packet RQ mode.
> + * Flow rules caching removed from the driver for memory saving and
> + correctness according to RTE flow and RTE ethdev.
Removed flow rules caching for memory saving and compliance with ethdev API.
@@ -1300,6 +1300,19 @@ set, and we should allow to specify zero values as rte_flow parameters for the
META and MARK items and actions. In the same time zero mask has no meaning and
should be rejected on validation stage.
+Notes for rte_flow
+------------------
+Implementation of flows organization is in non-cached mode. When stopping a
+device, all the flows inserted from application will be considered invalid and
+flushed automatically in the background. After restarting of the device, no
+application's flow exists in the system. All reference pointers to these flows
+held by the application are not valid anymore and should be discarded by the
+application.
+
+The application should recreate the flows as required after device restarting,
+and must not try to destroy or flush the invalid flows after stopping (except
+discarding them).
+
Notes for testpmd
-----------------
@@ -63,6 +63,8 @@ New Features
* Added support for matching on IPv4 Time To Live and IPv6 Hop Limit.
* Added support for creating Relaxed Ordering Memory Regions.
* Added support for jumbo frame size (9K MTU) in Multi-Packet RQ mode.
+ * Flow rules caching removed from the driver for memory saving and
+ correctness according to RTE flow and RTE ethdev.
* Optimized the memory consumption of flow.
* **Updated the Intel ice driver.**