Re: [PATCH rdma-next 2/2] IB/mlx5: Enable UAR to have DevX UID

From: Jason Gunthorpe
Date: Wed Sep 15 2021 - 09:48:01 EST


On Wed, Sep 15, 2021 at 02:11:23AM +0300, Leon Romanovsky wrote:
> From: Meir Lichtinger <meirl@xxxxxxxxxx>
>
> UID field was added to alloc_uar and dealloc_uar PRM command, to specify
> DevX UID for UAR. This change enables firmware validating user access to
> its own UAR resources.
>
> For the kernel allocated UARs the UID will stay 0 as of today.
>
> Signed-off-by: Meir Lichtinger <meirl@xxxxxxxxxx>
> Reviewed-by: Yishai Hadas <yishaih@xxxxxxxxxx>
> Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxx>
> drivers/infiniband/hw/mlx5/cmd.c | 24 ++++++++++++++
> drivers/infiniband/hw/mlx5/cmd.h | 2 ++
> drivers/infiniband/hw/mlx5/main.c | 55 +++++++++++++++++--------------
> 3 files changed, 57 insertions(+), 24 deletions(-)
>
> diff --git a/drivers/infiniband/hw/mlx5/cmd.c b/drivers/infiniband/hw/mlx5/cmd.c
> index a8db8a051170..0fe3c4ceec43 100644
> +++ b/drivers/infiniband/hw/mlx5/cmd.c
> @@ -206,3 +206,27 @@ int mlx5_cmd_mad_ifc(struct mlx5_core_dev *dev, const void *inb, void *outb,
> kfree(in);
> return err;
> }
> +
> +int mlx5_ib_cmd_uar_alloc(struct mlx5_core_dev *dev, u32 *uarn, u16 uid)
> +{
> + u32 out[MLX5_ST_SZ_DW(alloc_uar_out)] = {};
> + u32 in[MLX5_ST_SZ_DW(alloc_uar_in)] = {};
> + int err;
> +
> + MLX5_SET(alloc_uar_in, in, opcode, MLX5_CMD_OP_ALLOC_UAR);
> + MLX5_SET(alloc_uar_in, in, uid, uid);
> + err = mlx5_cmd_exec_inout(dev, alloc_uar, in, out);
> + if (!err)
> + *uarn = MLX5_GET(alloc_uar_out, out, uar);

Success oriented flow:

if (err)
return err;
*uarn = MLX5_GET(alloc_uar_out, out, uar);
return 0;

And why did we add entirely new functions instead of just adding a uid
argument to the core ones? Or, why doesn't this delete the old core
functions that look unused outside of IB anyhow?

Jason