Re: [PATCH RFC 1/3] IB/core: Expose a device attribute for rdma_read access flags

[Jason Gunthorpe <jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>]

On Tue, Nov 10, 2015 at 05:41:47AM -0800, Christoph Hellwig wrote:
> FYI, this is the API I'd aim for (only SRP and no HW driver converted
> yet):

>  	n = ib_map_mr_sg(desc->mr, state->sg, state->sg_nents,
> -			 dev->mr_page_size);
> +			 dev->mr_page_size,
> +			 /*
> +			  * XXX: add a bool write argument to this function,
> +			  * so that we only need to open up the required
> +			  * permissions.
> +			  */
> +			 IB_MR_REMOTE | IB_MR_RDMA_READ |
> IB_MR_RDMA_WRITE);

I would call it IB_RDMA_LKEY and IB_RDMA_RKEY. We have other places in
the API where lkey/rkey is used and it makes a lot more sense to think
about a MR as being either a lkey or rkey usable MR - since this is
effectively what we are doing here with these ops.

IB_MR_RKEY | IB_MR_RDMA_READ == The resulting key can be used
in a wr.rdma.rkey field

IB_MR_LKEY | IB_MR_SEND == The key can be used in wr.sg_list[].lkey

etc

It is an error to use a IB_MR_LKEY in a rkey field, etc..

> +enum ib_mr_flags {
> +	/* scope: either remote or local */
> +	IB_MR_REMOTE,
> +	IB_MR_LOCAL,
> +
> +	/* direction: one or both can be ORed into the scope above */
> +	IB_MR_RDMA_READ		= (1 << 10),
> +	IB_MR_RDMA_WRITE	= (1 << 11)

Don't forget SEND too.

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html