| Message ID | 20240223192658.45893-1-rrameshbabu@nvidia.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | ethtool HW timestamping statistics | expand |
On 2/23/2024 11:24 AM, Rahul Rameshbabu wrote: > The goal of this patch series is to introduce a common set of ethtool statistics > for hardware timestamping that a driver implementer can hook into. The > statistics counters added are based on what I believe are common > patterns/behaviors found across various hardware timestamping implementations > seen in the kernel tree today. The mlx5 family of devices is used as the PoC for > this patch series. Other vendors are more than welcome to chim in on this > series. > > Statistics can be queried from either the DMA or PHY layers. I think this > concept of layer-based statistics selection and the general timestamping layer > selection work Kory Maincent is working on can be converged together [1]. > > [1] https://lore.kernel.org/netdev/20240216-feature_ptp_netnext-v8-0-510f42f444fb@bootlin.com/ > Makes sense! I like this direction, I had meant to work on this for the Intel parts, but got sidetracked by other tasks. I look forward to seeing what you've done here.
On 2/23/2024 1:00 PM, Jacob Keller wrote: > > > On 2/23/2024 11:24 AM, Rahul Rameshbabu wrote: >> The goal of this patch series is to introduce a common set of ethtool statistics >> for hardware timestamping that a driver implementer can hook into. The >> statistics counters added are based on what I believe are common >> patterns/behaviors found across various hardware timestamping implementations >> seen in the kernel tree today. The mlx5 family of devices is used as the PoC for >> this patch series. Other vendors are more than welcome to chim in on this >> series. >> >> Statistics can be queried from either the DMA or PHY layers. I think this >> concept of layer-based statistics selection and the general timestamping layer >> selection work Kory Maincent is working on can be converged together [1]. >> >> [1] https://lore.kernel.org/netdev/20240216-feature_ptp_netnext-v8-0-510f42f444fb@bootlin.com/ >> > > Makes sense! I like this direction, I had meant to work on this for the > Intel parts, but got sidetracked by other tasks. I look forward to > seeing what you've done here. > I'm fairly happy with the series overall, and the entire thing read pretty straight forward. I'd be happy to follow up with patches to convert the ice driver to report these statistics as well.
On Fri, 23 Feb, 2024 13:00:10 -0800 Jacob Keller <jacob.e.keller@intel.com> wrote: > On 2/23/2024 11:24 AM, Rahul Rameshbabu wrote: >> The goal of this patch series is to introduce a common set of ethtool statistics >> for hardware timestamping that a driver implementer can hook into. The >> statistics counters added are based on what I believe are common >> patterns/behaviors found across various hardware timestamping implementations >> seen in the kernel tree today. The mlx5 family of devices is used as the PoC for >> this patch series. Other vendors are more than welcome to chim in on this >> series. >> >> Statistics can be queried from either the DMA or PHY layers. I think this >> concept of layer-based statistics selection and the general timestamping layer >> selection work Kory Maincent is working on can be converged together [1]. >> >> [1] https://lore.kernel.org/netdev/20240216-feature_ptp_netnext-v8-0-510f42f444fb@bootlin.com/ >> > > Makes sense! I like this direction, I had meant to work on this for the > Intel parts, but got sidetracked by other tasks. I still have a back burner task for a cyclecounter API change you suggested a while back.... Hoping to get to that after getting some large features out of the way in the next two months. -- Thanks, Rahul Rameshbabu
The goal of this patch series is to introduce a common set of ethtool statistics for hardware timestamping that a driver implementer can hook into. The statistics counters added are based on what I believe are common patterns/behaviors found across various hardware timestamping implementations seen in the kernel tree today. The mlx5 family of devices is used as the PoC for this patch series. Other vendors are more than welcome to chim in on this series. Statistics can be queried from either the DMA or PHY layers. I think this concept of layer-based statistics selection and the general timestamping layer selection work Kory Maincent is working on can be converged together [1]. [1] https://lore.kernel.org/netdev/20240216-feature_ptp_netnext-v8-0-510f42f444fb@bootlin.com/ Signed-off-by: Rahul Rameshbabu <rrameshbabu@nvidia.com> --- Rahul Rameshbabu (6): ethtool: add interface to read Tx hardware timestamping statistics net/mlx5e: Introduce lost_cqe statistic counter for PTP Tx port timestamping CQ net/mlx5e: Introduce timestamps statistic counter for Tx DMA layer net/mlx5e: Implement ethtool hardware timestamping statistics tools: ynl: ethtool.py: Make tool invokable from any CWD tools: ynl: ethtool.py: Add ts ethtool statistics group .../ethernet/mellanox/mlx5/counters.rst | 11 +++ .../net/ethernet/mellanox/mlx5/core/en/ptp.c | 1 + .../ethernet/mellanox/mlx5/core/en_ethtool.c | 9 +++ .../ethernet/mellanox/mlx5/core/en_stats.c | 71 +++++++++++++++++++ .../ethernet/mellanox/mlx5/core/en_stats.h | 4 ++ .../net/ethernet/mellanox/mlx5/core/en_tx.c | 6 +- include/linux/ethtool.h | 28 ++++++++ include/uapi/linux/ethtool.h | 20 ++++++ include/uapi/linux/ethtool_netlink.h | 17 +++++ net/ethtool/netlink.h | 3 +- net/ethtool/stats.c | 53 ++++++++++++-- net/ethtool/strset.c | 5 ++ tools/net/ynl/ethtool.py | 9 ++- 13 files changed, 227 insertions(+), 10 deletions(-)