Commit 20cf4e02 authored by Chuck Lever's avatar Chuck Lever Committed by Doug Ledford
Browse files

rdma: Enable ib_alloc_cq to spread work over a device's comp_vectors 



Send and Receive completion is handled on a single CPU selected at
the time each Completion Queue is allocated. Typically this is when
an initiator instantiates an RDMA transport, or when a target
accepts an RDMA connection.

Some ULPs cannot open a connection per CPU to spread completion
workload across available CPUs and MSI vectors. For such ULPs,
provide an API that allows the RDMA core to select a completion
vector based on the device's complement of available comp_vecs.

ULPs that invoke ib_alloc_cq() with only comp_vector 0 are converted
to use the new API so that their completion workloads interfere less
with each other.

Suggested-by: default avatarHåkon Bugge <haakon.bugge@oracle.com>
Signed-off-by: default avatarChuck Lever <chuck.lever@oracle.com>
Reviewed-by: default avatarLeon Romanovsky <leonro@mellanox.com>
Cc: <linux-cifs@vger.kernel.org>
Cc: <v9fs-developer@lists.sourceforge.net>
Link: https://lore.kernel.org/r/20190729171923.13428.52555.stgit@manet.1015granger.net


Signed-off-by: default avatarDoug Ledford <dledford@redhat.com>
parent 31d0e6c1

drivers/infiniband/core/cq.c

+28 −0
Original line number Diff line number Diff line
@@ -252,6 +252,34 @@ out_free_cq: 
}

EXPORT_SYMBOL(__ib_alloc_cq_user);



/**

 * __ib_alloc_cq_any - allocate a completion queue

 * @dev:		device to allocate the CQ for

 * @private:		driver private data, accessible from cq->cq_context

 * @nr_cqe:		number of CQEs to allocate

 * @poll_ctx:		context to poll the CQ from

 * @caller:		module owner name

 *

 * Attempt to spread ULP Completion Queues over each device's interrupt

 * vectors. A simple best-effort mechanism is used.

 */

struct ib_cq *__ib_alloc_cq_any(struct ib_device *dev, void *private,

				int nr_cqe, enum ib_poll_context poll_ctx,

				const char *caller)

{

	static atomic_t counter;

	int comp_vector = 0;



	if (dev->num_comp_vectors > 1)

		comp_vector =

			atomic_inc_return(&counter) %

			min_t(int, dev->num_comp_vectors, num_online_cpus());



	return __ib_alloc_cq_user(dev, private, nr_cqe, comp_vector, poll_ctx,

				  caller, NULL);

}

EXPORT_SYMBOL(__ib_alloc_cq_any);



/**

 * ib_free_cq_user - free a completion queue

 * @cq:		completion queue to free.

drivers/infiniband/ulp/srpt/ib_srpt.c

+2 −2
Original line number Diff line number Diff line
@@ -1767,8 +1767,8 @@ static int srpt_create_ch_ib(struct srpt_rdma_ch *ch) 
		goto out;



retry:

	ch->cq = ib_alloc_cq(sdev->device, ch, ch->rq_size + sq_size,

			0 /* XXX: spread CQs */, IB_POLL_WORKQUEUE);

	ch->cq = ib_alloc_cq_any(sdev->device, ch, ch->rq_size + sq_size,

				 IB_POLL_WORKQUEUE);

	if (IS_ERR(ch->cq)) {

		ret = PTR_ERR(ch->cq);

		pr_err("failed to create CQ cqe= %d ret= %d\n",

fs/cifs/smbdirect.c

+6 −4
Original line number Diff line number Diff line
@@ -1654,15 +1654,17 @@ static struct smbd_connection *_smbd_get_connection( 


	info->send_cq = NULL;

	info->recv_cq = NULL;

	info->send_cq = ib_alloc_cq(info->id->device, info,

			info->send_credit_target, 0, IB_POLL_SOFTIRQ);

	info->send_cq =

		ib_alloc_cq_any(info->id->device, info,

				info->send_credit_target, IB_POLL_SOFTIRQ);

	if (IS_ERR(info->send_cq)) {

		info->send_cq = NULL;

		goto alloc_cq_failed;

	}



	info->recv_cq = ib_alloc_cq(info->id->device, info,

			info->receive_credit_max, 0, IB_POLL_SOFTIRQ);

	info->recv_cq =

		ib_alloc_cq_any(info->id->device, info,

				info->receive_credit_max, IB_POLL_SOFTIRQ);

	if (IS_ERR(info->recv_cq)) {

		info->recv_cq = NULL;

		goto alloc_cq_failed;

include/rdma/ib_verbs.h

+19 −0
Original line number Diff line number Diff line
@@ -3711,6 +3711,25 @@ static inline struct ib_cq *ib_alloc_cq(struct ib_device *dev, void *private, 
				NULL);

}



struct ib_cq *__ib_alloc_cq_any(struct ib_device *dev, void *private,

				int nr_cqe, enum ib_poll_context poll_ctx,

				const char *caller);



/**

 * ib_alloc_cq_any: Allocate kernel CQ

 * @dev: The IB device

 * @private: Private data attached to the CQE

 * @nr_cqe: Number of CQEs in the CQ

 * @poll_ctx: Context used for polling the CQ

 */

static inline struct ib_cq *ib_alloc_cq_any(struct ib_device *dev,

					    void *private, int nr_cqe,

					    enum ib_poll_context poll_ctx)

{

	return __ib_alloc_cq_any(dev, private, nr_cqe, poll_ctx,

				 KBUILD_MODNAME);

}



/**

 * ib_free_cq_user - Free kernel/user CQ

 * @cq: The CQ to free

net/9p/trans_rdma.c

+3 −3
Original line number Diff line number Diff line
@@ -685,9 +685,9 @@ rdma_create_trans(struct p9_client *client, const char *addr, char *args) 
		goto error;



	/* Create the Completion Queue */

	rdma->cq = ib_alloc_cq(rdma->cm_id->device, client,

	rdma->cq = ib_alloc_cq_any(rdma->cm_id->device, client,

				   opts.sq_depth + opts.rq_depth + 1,

			0, IB_POLL_SOFTIRQ);

				   IB_POLL_SOFTIRQ);

	if (IS_ERR(rdma->cq))

		goto error;

CRA Git | Maintained and supported by SUSTech CRA and CCSE