IB/rdmavt: Add data structures and routines for table driven post send

Add flexibility for driver dependent operations in post send
because different drivers will have differing post send
operation support.

This includes data structure definitions to support a table
driven scheme along with the necessary validation routine
using the new table.

Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Jianxin Xiong <jianxin.xiong@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

3 files changed, 89 insertions, 9 deletions

diff --git a/drivers/infiniband/sw/rdmavt/qp.c b/drivers/infiniband/sw/rdmavt/qp.c
index 41ba7e9cadaa..d2b5b547c5a0 100644
--- a/drivers/infiniband/sw/rdmavt/qp.c
+++ b/drivers/infiniband/sw/rdmavt/qp.c
@@ -613,6 +613,7 @@ struct ib_qp *rvt_create_qp(struct ib_pd *ibpd,
 	struct rvt_dev_info *rdi = ib_to_rvt(ibpd->device);
 	void *priv = NULL;
 	gfp_t gfp;
+	size_t sqsize;
 
 	if (!rdi)
 		return ERR_PTR(-EINVAL);
@@ -643,7 +644,8 @@ struct ib_qp *rvt_create_qp(struct ib_pd *ibpd,
 		    init_attr->cap.max_recv_wr == 0)
 			return ERR_PTR(-EINVAL);
 	}
-
+	sqsize =
+		init_attr->cap.max_send_wr + 1;
 	switch (init_attr->qp_type) {
 	case IB_QPT_SMI:
 	case IB_QPT_GSI:
@@ -658,11 +660,11 @@ struct ib_qp *rvt_create_qp(struct ib_pd *ibpd,
 			sizeof(struct rvt_swqe);
 		if (gfp == GFP_NOIO)
 			swq = __vmalloc(
-				(init_attr->cap.max_send_wr + 1) * sz,
+				sqsize * sz,
 				gfp | __GFP_ZERO, PAGE_KERNEL);
 		else
 			swq = vzalloc_node(
-				(init_attr->cap.max_send_wr + 1) * sz,
+				sqsize * sz,
 				rdi->dparms.node);
 		if (!swq)
 			return ERR_PTR(-ENOMEM);
@@ -747,7 +749,7 @@ struct ib_qp *rvt_create_qp(struct ib_pd *ibpd,
 		INIT_LIST_HEAD(&qp->rspwait);
 		qp->state = IB_QPS_RESET;
 		qp->s_wq = swq;
-		qp->s_size = init_attr->cap.max_send_wr + 1;
+		qp->s_size = sqsize;
 		qp->s_avail = init_attr->cap.max_send_wr;
 		qp->s_max_sge = init_attr->cap.max_send_sge;
 		if (init_attr->sq_sig_type == IB_SIGNAL_REQ_WR)
@@ -1440,12 +1442,65 @@ int rvt_post_recv(struct ib_qp *ibqp, struct ib_recv_wr *wr,
 }
 
 /**
- * qp_get_savail - return number of avail send entries
+ * rvt_qp_valid_operation - validate post send wr request
+ * @qp - the qp
+ * @post-parms - the post send table for the driver
+ * @wr - the work request
  *
+ * The routine validates the operation based on the
+ * validation table an returns the length of the operation
+ * which can extend beyond the ib_send_bw.  Operation
+ * dependent flags key atomic operation validation.
+ *
+ * There is an exception for UD qps that validates the pd and
+ * overrides the length to include the additional UD specific
+ * length.
+ *
+ * Returns a negative error or the length of the work request
+ * for building the swqe.
+ */
+static inline int rvt_qp_valid_operation(
+	struct rvt_qp *qp,
+	const struct rvt_operation_params *post_parms,
+	struct ib_send_wr *wr)
+{
+	int len;
+
+	if (wr->opcode >= RVT_OPERATION_MAX || !post_parms[wr->opcode].length)
+		return -EINVAL;
+	if (!(post_parms[wr->opcode].qpt_support & BIT(qp->ibqp.qp_type)))
+		return -EINVAL;
+	if ((post_parms[wr->opcode].flags & RVT_OPERATION_PRIV) &&
+	    ibpd_to_rvtpd(qp->ibqp.pd)->user)
+		return -EINVAL;
+	if (post_parms[wr->opcode].flags & RVT_OPERATION_ATOMIC_SGE &&
+	    (wr->num_sge == 0 ||
+	     wr->sg_list[0].length < sizeof(u64) ||
+	     wr->sg_list[0].addr & (sizeof(u64) - 1)))
+		return -EINVAL;
+	if (post_parms[wr->opcode].flags & RVT_OPERATION_ATOMIC &&
+	    !qp->s_max_rd_atomic)
+		return -EINVAL;
+	len = post_parms[wr->opcode].length;
+	/* UD specific */
+	if (qp->ibqp.qp_type != IB_QPT_UC &&
+	    qp->ibqp.qp_type != IB_QPT_RC) {
+		if (qp->ibqp.pd != ud_wr(wr)->ah->pd)
+			return -EINVAL;
+		len = sizeof(struct ib_ud_wr);
+	}
+	return len;
+}
+
+/**
+ * qp_get_savail - return number of avail send entries
  * @qp - the qp
  *
  * This assumes the s_hlock is held but the s_last
  * qp variable is uncontrolled.
+ *
+ * The return is adjusted to not count device specific
+ * reserved operations.
  */
 static inline u32 qp_get_savail(struct rvt_qp *qp)
 {
@@ -1481,6 +1536,8 @@ static int rvt_post_one_wr(struct rvt_qp *qp,
 	u8 log_pmtu;
 	int ret;
 
+	BUILD_BUG_ON(IB_QPT_MAX >= (sizeof(u32) * BITS_PER_BYTE));
+
 	/* IB spec says that num_sge == 0 is OK. */
 	if (unlikely(wr->num_sge > qp->s_max_sge))
 		return -EINVAL;
diff --git a/include/rdma/rdma_vt.h b/include/rdma/rdma_vt.h
index 9c9a27d42aaa..3a70dc047314 100644
--- a/include/rdma/rdma_vt.h
+++ b/include/rdma/rdma_vt.h
@@ -351,6 +351,9 @@ struct rvt_dev_info {
 	/* Driver specific properties */
 	struct rvt_driver_params dparms;
 
+	/* post send table */
+	const struct rvt_operation_params *post_parms;
+
 	struct rvt_mregion __rcu *dma_mr;
 	struct rvt_lkey_table lkey_table;
 
diff --git a/include/rdma/rdmavt_qp.h b/include/rdma/rdmavt_qp.h
index 6d23b879416a..a90d1e941504 100644
--- a/include/rdma/rdmavt_qp.h
+++ b/include/rdma/rdmavt_qp.h
@@ -228,11 +228,31 @@ struct rvt_ack_entry {
 
 #define	RC_QP_SCALING_INTERVAL	5
 
-/*
- * Variables prefixed with s_ are for the requester (sender).
- * Variables prefixed with r_ are for the responder (receiver).
- * Variables prefixed with ack_ are for responder replies.
+#define RVT_OPERATION_PRIV        0x00000001
+#define RVT_OPERATION_ATOMIC      0x00000002
+#define RVT_OPERATION_ATOMIC_SGE  0x00000004
+
+#define RVT_OPERATION_MAX (IB_WR_RESERVED10 + 1)
+
+/**
+ * rvt_operation_params - op table entry
+ * @length - the length to copy into the swqe entry
+ * @qpt_support - a bit mask indicating QP type support
+ * @flags - RVT_OPERATION flags (see above)
+ *
+ * This supports table driven post send so that
+ * the driver can have differing an potentially
+ * different sets of operations.
  *
+ **/
+
+struct rvt_operation_params {
+	size_t length;
+	u32 qpt_support;
+	u32 flags;
+};
+
+/*
  * Common variables are protected by both r_rq.lock and s_lock in that order
  * which only happens in modify_qp() or changing the QP 'state'.
  */