| Message ID | 20181101161312.14624-1-sagi@grimberg.me (mailing list archive) |
|---|---|
| State | Accepted |
| Headers | show |
| Series | mlx5: remove support for ib_get_vector_affinity | expand |
On Thu, Nov 01, 2018 at 09:13:12AM -0700, Sagi Grimberg wrote: > Devices that does not use managed affinity can not export a vector > affinity as the consumer relies on having a static mapping it can map > to upper layer affinity (e.g. sw queues). If the driver allows the user > to set the device irq affinity, then the affinitization of a long term > existing entites is not relevant. > > For example, nvme-rdma controllers queue-irq affinitization is determined > at init time so if the irq affinity changes over time, we are no longer > aligned. > > Cc: Leon Romanovsky <leonro@mellanox.com> > Signed-off-by: Sagi Grimberg <sagi@grimberg.me> > --- > Hi Sagi - If you dont mind, can do the same patch for i40iw? Since our device IRQs also do not use managed affinity. https://elixir.bootlin.com/linux/latest/source/drivers/infiniband/hw/i40iw/i40iw_verbs.c#L2726 Shiraz
On Thu, Nov 01, 2018 at 09:13:12AM -0700, Sagi Grimberg wrote: > Devices that does not use managed affinity can not export a vector > affinity as the consumer relies on having a static mapping it can map > to upper layer affinity (e.g. sw queues). If the driver allows the user > to set the device irq affinity, then the affinitization of a long term > existing entites is not relevant. > > For example, nvme-rdma controllers queue-irq affinitization is determined > at init time so if the irq affinity changes over time, we are no longer > aligned. > > Cc: Leon Romanovsky <leonro@mellanox.com> Something wrong with your git send-email, this CC wasn't added to CCed list. > Signed-off-by: Sagi Grimberg <sagi@grimberg.me> > --- > drivers/infiniband/hw/mlx5/main.c | 9 --------- > include/linux/mlx5/driver.h | 6 ------ > 2 files changed, 15 deletions(-) > You added, you removed, good :) Thanks, Acked-by: Leon Romanovsky <leonro@mellanox.com>
>> Cc: Leon Romanovsky <leonro@mellanox.com> > > Something wrong with your git send-email, this CC wasn't added to CCed list. Yea, I suppressed that... forgot to explicitly CC you. > >> Signed-off-by: Sagi Grimberg <sagi@grimberg.me> >> --- >> drivers/infiniband/hw/mlx5/main.c | 9 --------- >> include/linux/mlx5/driver.h | 6 ------ >> 2 files changed, 15 deletions(-) >> > > You added, you removed, good :) Not so great because now nvme-rdma now works sub optimally on mlx5 devices, but that is what we need to pay I guess... > Thanks, > Acked-by: Leon Romanovsky <leonro@mellanox.com> Thanks
On Thu, 2018-11-01 at 12:07 -0700, Sagi Grimberg wrote: > > > Cc: Leon Romanovsky <leonro@mellanox.com> > > > > Something wrong with your git send-email, this CC wasn't added to CCed list. > > Yea, I suppressed that... forgot to explicitly CC you. > > > > > > Signed-off-by: Sagi Grimberg <sagi@grimberg.me> > > > --- > > > drivers/infiniband/hw/mlx5/main.c | 9 --------- > > > include/linux/mlx5/driver.h | 6 ------ > > > 2 files changed, 15 deletions(-) > > > > > > > You added, you removed, good :) > > Not so great because now nvme-rdma now works sub optimally on mlx5 > devices, but that is what we need to pay I guess... > > > Thanks, > > Acked-by: Leon Romanovsky <leonro@mellanox.com> > > Thanks Fixed up a minor context issue and applied to for-next.
diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index c414f3809e5c..257d1c7823f3 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -5303,14 +5303,6 @@ static void init_delay_drop(struct mlx5_ib_dev *dev) mlx5_ib_warn(dev, "Failed to init delay drop debugfs\n"); } -static const struct cpumask * -mlx5_ib_get_vector_affinity(struct ib_device *ibdev, int comp_vector) -{ - struct mlx5_ib_dev *dev = to_mdev(ibdev); - - return mlx5_get_vector_affinity_hint(dev->mdev, comp_vector); -} - /* The mlx5_ib_multiport_mutex should be held when calling this function */ static void mlx5_ib_unbind_slave_port(struct mlx5_ib_dev *ibdev, struct mlx5_ib_multiport_info *mpi) @@ -5823,7 +5815,6 @@ int mlx5_ib_stage_caps_init(struct mlx5_ib_dev *dev) dev->ib_dev.map_mr_sg = mlx5_ib_map_mr_sg; dev->ib_dev.check_mr_status = mlx5_ib_check_mr_status; dev->ib_dev.get_dev_fw_str = get_dev_fw_str; - dev->ib_dev.get_vector_affinity = mlx5_ib_get_vector_affinity; if (MLX5_CAP_GEN(mdev, ipoib_enhanced_offloads)) dev->ib_dev.alloc_rdma_netdev = mlx5_ib_alloc_rdma_netdev; diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h index 66d94b4557cf..011e2d195ea8 100644 --- a/include/linux/mlx5/driver.h +++ b/include/linux/mlx5/driver.h @@ -1310,10 +1310,4 @@ enum { MLX5_TRIGGERED_CMD_COMP = (u64)1 << 32, }; -static inline const struct cpumask * -mlx5_get_vector_affinity_hint(struct mlx5_core_dev *dev, int vector) -{ - return dev->priv.irq_info[vector].mask; -} - #endif /* MLX5_DRIVER_H */
Devices that does not use managed affinity can not export a vector affinity as the consumer relies on having a static mapping it can map to upper layer affinity (e.g. sw queues). If the driver allows the user to set the device irq affinity, then the affinitization of a long term existing entites is not relevant. For example, nvme-rdma controllers queue-irq affinitization is determined at init time so if the irq affinity changes over time, we are no longer aligned. Cc: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Sagi Grimberg <sagi@grimberg.me> --- drivers/infiniband/hw/mlx5/main.c | 9 --------- include/linux/mlx5/driver.h | 6 ------ 2 files changed, 15 deletions(-)