Re: [PATCH for-next v3] RDMA/hfi2: Consolidate ABI files and setup uverbs access

[Jason Gunthorpe <jgg@nvidia.com>]

On Thu, Apr 02, 2026 at 04:13:09PM -0400, Dennis Dalessandro wrote:

> diff --git a/include/uapi/rdma/hfi/hfi1_user.h b/include/uapi/rdma/hfi/hfi1_user.h
> deleted file mode 100644
> index 1106a7c90b29..000000000000
> --- a/include/uapi/rdma/hfi/hfi1_user.h
> +++ /dev/null

> -/*
> - * Set of HW and driver capability/feature bits.
> - * These bit values are used to configure enabled/disabled HW and
> - * driver features. The same set of bits are communicated to user
> - * space.
> - */
> -#define HFI1_CAP_DMA_RTAIL        (1UL <<  0) /* Use DMA'ed RTail value */

How is the userspace going to work after these are removed?

Shouldn't some of the drivers/infiniband/hw/hfi1/hfi1_user_compat.h
stuff be here instead?

> diff --git a/include/uapi/rdma/rdma_user_ioctl.h b/include/uapi/rdma/rdma_user_ioctl.h
> index 53c55188dd2a..263cace86f8f 100644
> --- a/include/uapi/rdma/rdma_user_ioctl.h
> +++ b/include/uapi/rdma/rdma_user_ioctl.h
> @@ -39,47 +39,14 @@
>  #include <rdma/rdma_user_ioctl_cmds.h>
>  
>  /* Legacy name, for user space application which already use it */
> -#define IB_IOCTL_MAGIC		RDMA_IOCTL_MAGIC
> -
> -/*
> - * General blocks assignments
> - * It is closed on purpose - do not expose it to user space
> - * #define MAD_CMD_BASE		0x00
> - * #define HFI1_CMD_BAS		0xE0
> - */
> +#define IB_IOCTL_MAGIC RDMA_IOCTL_MAGIC
>  
>  /* MAD specific section */
> -#define IB_USER_MAD_REGISTER_AGENT	_IOWR(RDMA_IOCTL_MAGIC, 0x01, struct ib_user_mad_reg_req)
> -#define IB_USER_MAD_UNREGISTER_AGENT	_IOW(RDMA_IOCTL_MAGIC,  0x02, __u32)
> -#define IB_USER_MAD_ENABLE_PKEY		_IO(RDMA_IOCTL_MAGIC,   0x03)
> -#define IB_USER_MAD_REGISTER_AGENT2	_IOWR(RDMA_IOCTL_MAGIC, 0x04, struct ib_user_mad_reg_req2)

These general lines shouldn't be touched right?

Jason