IB/isert: allocate RW ctxs according to max IO size

Current iSER target code allocates MR pool budget based on queue size.
Since there is no handshake between iSER initiator and target on max IO
size, we'll set the iSER target to support upto 16MiB IO operations and
allocate the correct number of RDMA ctxs according to the factor of MR's
per IO operation. This would guarantee sufficient size of the MR pool for
the required IO queue depth and IO size.

Link: https://lore.kernel.org/r/20200708091908.162263-1-maxg@mellanox.com
Reported-by: Krishnamraju Eraparaju <krishna2@chelsio.com>
Tested-by: Krishnamraju Eraparaju <krishna2@chelsio.com>
Signed-off-by: Max Gurtovoy <maxg@mellanox.com>
Acked-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

drivers/infiniband/ulp/isert/ib_isert.c

@@ -119,7 +119,7 @@ isert_create_qp(struct isert_conn *isert_conn,
 {
 	struct isert_device *device = isert_conn->device;
 	struct ib_qp_init_attr attr;
-	int ret;
+	int ret, factor;
 
 	memset(&attr, 0, sizeof(struct ib_qp_init_attr));
 	attr.event_handler = isert_qp_event_callback;
@@ -128,7 +128,9 @@ isert_create_qp(struct isert_conn *isert_conn,
 	attr.recv_cq = comp->cq;
 	attr.cap.max_send_wr = ISERT_QP_MAX_REQ_DTOS + 1;
 	attr.cap.max_recv_wr = ISERT_QP_MAX_RECV_DTOS + 1;
-	attr.cap.max_rdma_ctxs = ISCSI_DEF_XMIT_CMDS_MAX;
+	factor = rdma_rw_mr_factor(device->ib_device, cma_id->port_num,
+				   ISCSI_ISER_MAX_SG_TABLESIZE);
+	attr.cap.max_rdma_ctxs = ISCSI_DEF_XMIT_CMDS_MAX * factor;
 	attr.cap.max_send_sge = device->ib_device->attrs.max_send_sge;
 	attr.cap.max_recv_sge = 1;
 	attr.sq_sig_type = IB_SIGNAL_REQ_WR;

drivers/infiniband/ulp/isert/ib_isert.h

@@ -63,7 +63,8 @@
 		(ISER_RX_PAYLOAD_SIZE + sizeof(u64) + sizeof(struct ib_sge) + \
 		 sizeof(struct ib_cqe) + sizeof(bool)))
 
-#define ISCSI_ISER_SG_TABLESIZE		256
+/* Maximum support is 16MB I/O size */
+#define ISCSI_ISER_MAX_SG_TABLESIZE	4096
 
 enum isert_desc_type {
 	ISCSI_TX_CONTROL,