RDMA/cxgb3: Support peer-2-peer connection setup

Open MPI, Intel MPI and other applications don't respect the iWARP
requirement that the client (active) side of the connection send the
first RDMA message.  This class of application connection setup is
called peer-to-peer.  Typically once the connection is setup, _both_
sides want to send data.

This patch enables supporting peer-to-peer over the chelsio RNIC by
enforcing this iWARP requirement in the driver itself as part of RDMA
connection setup.

Connection setup is extended, when the peer2peer module option is 1,
such that the MPA initiator will send a 0B Read (the RTR) just after
connection setup.  The MPA responder will suspend SQ processing until
the RTR message is received and reply-to.

In the longer term, this will be handled in a standardized way by
enhancing the MPA negotiation so peers can indicate whether they
want/need the RTR and what type of RTR (0B read, 0B write, or 0B send)
should be sent.  This will be done by standardizing a few bits of the
private data in order to negotiate all this.  However this patch
enables peer-to-peer applications now and allows most of the required
firmware and driver changes to be done and tested now.

Design:

 - Add a module option, peer2peer, to enable this mode.

 - New firmware support for peer-to-peer mode:

	- a new bit in the rdma_init WR to tell it to do peer-2-peer
	  and what form of RTR message to send or expect.

	- process _all_ preposted recvs before moving the connection
	  into rdma mode.

	- passive side: defer completing the rdma_init WR until all
	  pre-posted recvs are processed.  Suspend SQ processing until
	  the RTR is received.

	- active side: expect and process the 0B read WR on offload TX
	  queue. Defer completing the rdma_init WR until all
	  pre-posted recvs are processed.  Suspend SQ processing until
	  the 0B read WR is processed from the offload TX queue.

 - If peer2peer is set, driver posts 0B read request on offload TX
   queue just after posting the rdma_init WR to the offload TX queue.

 - Add CQ poll logic to ignore unsolicitied read responses.

Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

1 files changed, 50 insertions, 4 deletions

diff --git a/drivers/infiniband/hw/cxgb3/iwch_qp.c b/drivers/infiniband/hw/cxgb3/iwch_qp.c
index 6cd484e11c1..9b4be889c58 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_qp.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_qp.c
@@ -586,6 +586,36 @@ static inline void build_term_codes(struct respQ_msg_t *rsp_msg,
 	}
 }
 
+int iwch_post_zb_read(struct iwch_qp *qhp)
+{
+	union t3_wr *wqe;
+	struct sk_buff *skb;
+	u8 flit_cnt = sizeof(struct t3_rdma_read_wr) >> 3;
+
+	PDBG("%s enter\n", __func__);
+	skb = alloc_skb(40, GFP_KERNEL);
+	if (!skb) {
+		printk(KERN_ERR "%s cannot send zb_read!!\n", __func__);
+		return -ENOMEM;
+	}
+	wqe = (union t3_wr *)skb_put(skb, sizeof(struct t3_rdma_read_wr));
+	memset(wqe, 0, sizeof(struct t3_rdma_read_wr));
+	wqe->read.rdmaop = T3_READ_REQ;
+	wqe->read.reserved[0] = 0;
+	wqe->read.reserved[1] = 0;
+	wqe->read.reserved[2] = 0;
+	wqe->read.rem_stag = cpu_to_be32(1);
+	wqe->read.rem_to = cpu_to_be64(1);
+	wqe->read.local_stag = cpu_to_be32(1);
+	wqe->read.local_len = cpu_to_be32(0);
+	wqe->read.local_to = cpu_to_be64(1);
+	wqe->send.wrh.op_seop_flags = cpu_to_be32(V_FW_RIWR_OP(T3_WR_READ));
+	wqe->send.wrh.gen_tid_len = cpu_to_be32(V_FW_RIWR_TID(qhp->ep->hwtid)|
+						V_FW_RIWR_LEN(flit_cnt));
+	skb->priority = CPL_PRIORITY_DATA;
+	return cxgb3_ofld_send(qhp->rhp->rdev.t3cdev_p, skb);
+}
+
 /*
  * This posts a TERMINATE with layer=RDMA, type=catastrophic.
  */
@@ -671,11 +701,18 @@ static void flush_qp(struct iwch_qp *qhp, unsigned long *flag)
 
 
 /*
- * Return non zero if at least one RECV was pre-posted.
+ * Return count of RECV WRs posted
  */
-static int rqes_posted(struct iwch_qp *qhp)
+u16 iwch_rqes_posted(struct iwch_qp *qhp)
 {
-	return fw_riwrh_opcode((struct fw_riwrh *)qhp->wq.queue) == T3_WR_RCV;
+	union t3_wr *wqe = qhp->wq.queue;
+	u16 count = 0;
+	while ((count+1) != 0 && fw_riwrh_opcode((struct fw_riwrh *)wqe) == T3_WR_RCV) {
+		count++;
+		wqe++;
+	}
+	PDBG("%s qhp %p count %u\n", __func__, qhp, count);
+	return count;
 }
 
 static int rdma_init(struct iwch_dev *rhp, struct iwch_qp *qhp,
@@ -716,8 +753,17 @@ static int rdma_init(struct iwch_dev *rhp, struct iwch_qp *qhp,
 	init_attr.ird = qhp->attr.max_ird;
 	init_attr.qp_dma_addr = qhp->wq.dma_addr;
 	init_attr.qp_dma_size = (1UL << qhp->wq.size_log2);
-	init_attr.flags = rqes_posted(qhp) ? RECVS_POSTED : 0;
+	init_attr.rqe_count = iwch_rqes_posted(qhp);
+	init_attr.flags = qhp->attr.mpa_attr.initiator ? MPA_INITIATOR : 0;
 	init_attr.flags |= capable(CAP_NET_BIND_SERVICE) ? PRIV_QP : 0;
+	if (peer2peer) {
+		init_attr.rtr_type = RTR_READ;
+		if (init_attr.ord == 0 && qhp->attr.mpa_attr.initiator)
+			init_attr.ord = 1;
+		if (init_attr.ird == 0 && !qhp->attr.mpa_attr.initiator)
+			init_attr.ird = 1;
+	} else
+		init_attr.rtr_type = 0;
 	init_attr.irs = qhp->ep->rcv_seq;
 	PDBG("%s init_attr.rq_addr 0x%x init_attr.rq_size = %d "
 	     "flags 0x%x qpcaps 0x%x\n", __func__,