[dpdk-dev] [PATCH 07/20] crypto/cnxk: add enqueue burst op

[Anoob Joseph <anoobj@marvell.com>]

Add enqueue_burst op in cn9k & cn10k.

Signed-off-by: Ankur Dwivedi <adwivedi@marvell.com>
Signed-off-by: Anoob Joseph <anoobj@marvell.com>
Signed-off-by: Archana Muniganti <marchana@marvell.com>
Signed-off-by: Tejasree Kondoj <ktejasree@marvell.com>
---
 drivers/crypto/cnxk/cn10k_cryptodev.c     |   2 +
 drivers/crypto/cnxk/cn10k_cryptodev_ops.c | 189 ++++++++++++++++++++++++++++++
 drivers/crypto/cnxk/cn10k_cryptodev_ops.h |   2 +
 drivers/crypto/cnxk/cn9k_cryptodev.c      |   2 +
 drivers/crypto/cnxk/cn9k_cryptodev_ops.c  | 154 ++++++++++++++++++++++++
 drivers/crypto/cnxk/cn9k_cryptodev_ops.h  |   2 +
 drivers/crypto/cnxk/cnxk_cryptodev_ops.h  |   9 ++
 7 files changed, 360 insertions(+)

diff --git a/drivers/crypto/cnxk/cn10k_cryptodev.c b/drivers/crypto/cnxk/cn10k_cryptodev.c
index 79397d5..a34dbbf 100644
--- a/drivers/crypto/cnxk/cn10k_cryptodev.c
+++ b/drivers/crypto/cnxk/cn10k_cryptodev.c
@@ -79,6 +79,8 @@ cn10k_cpt_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
 
 	cnxk_cpt_caps_populate(vf);
 
+	cn10k_cpt_set_enqdeq_fns(dev);
+
 	return 0;
 
 dev_fini:
diff --git a/drivers/crypto/cnxk/cn10k_cryptodev_ops.c b/drivers/crypto/cnxk/cn10k_cryptodev_ops.c
index 34dc107..afdd43c 100644
--- a/drivers/crypto/cnxk/cn10k_cryptodev_ops.c
+++ b/drivers/crypto/cnxk/cn10k_cryptodev_ops.c
@@ -7,7 +7,196 @@
 
 #include "cn10k_cryptodev.h"
 #include "cn10k_cryptodev_ops.h"
+#include "cnxk_cryptodev.h"
 #include "cnxk_cryptodev_ops.h"
+#include "cnxk_se.h"
+
+static inline struct cnxk_se_sess *
+cn10k_cpt_sym_temp_sess_create(struct cnxk_cpt_qp *qp, struct rte_crypto_op *op)
+{
+	const int driver_id = cn10k_cryptodev_driver_id;
+	struct rte_crypto_sym_op *sym_op = op->sym;
+	struct rte_cryptodev_sym_session *sess;
+	struct cnxk_se_sess *priv;
+	int ret;
+
+	/* Create temporary session */
+	sess = rte_cryptodev_sym_session_create(qp->sess_mp);
+	if (sess == NULL)
+		return NULL;
+
+	ret = sym_session_configure(qp->lf.roc_cpt, driver_id, sym_op->xform,
+				    sess, qp->sess_mp_priv);
+	if (ret)
+		goto sess_put;
+
+	priv = get_sym_session_private_data(sess, driver_id);
+
+	sym_op->session = sess;
+
+	return priv;
+
+sess_put:
+	rte_mempool_put(qp->sess_mp, sess);
+	return NULL;
+}
+
+static __rte_always_inline int __rte_hot
+cpt_sym_inst_fill(struct cnxk_cpt_qp *qp, struct rte_crypto_op *op,
+		  struct cnxk_se_sess *sess, struct cpt_inflight_req *infl_req,
+		  struct cpt_inst_s *inst)
+{
+	RTE_SET_USED(qp);
+	RTE_SET_USED(op);
+	RTE_SET_USED(sess);
+	RTE_SET_USED(infl_req);
+	RTE_SET_USED(inst);
+
+	return -ENOTSUP;
+}
+
+static inline int
+cn10k_cpt_fill_inst(struct cnxk_cpt_qp *qp, struct rte_crypto_op *ops[],
+		    struct cpt_inst_s inst[], struct cpt_inflight_req *infl_req)
+{
+	struct rte_crypto_sym_op *sym_op;
+	struct cnxk_se_sess *sess;
+	struct rte_crypto_op *op;
+	uint64_t w7;
+	int ret;
+
+	op = ops[0];
+
+	inst[0].w0.u64 = 0;
+	inst[0].w2.u64 = 0;
+	inst[0].w3.u64 = 0;
+
+	sym_op = op->sym;
+
+	if (op->type == RTE_CRYPTO_OP_TYPE_SYMMETRIC) {
+		if (op->sess_type == RTE_CRYPTO_OP_WITH_SESSION) {
+			sess = get_sym_session_private_data(
+				sym_op->session, cn10k_cryptodev_driver_id);
+			ret = cpt_sym_inst_fill(qp, op, sess, infl_req,
+						&inst[0]);
+			if (unlikely(ret))
+				return 0;
+			w7 = sess->cpt_inst_w7;
+		} else {
+			sess = cn10k_cpt_sym_temp_sess_create(qp, op);
+			if (unlikely(sess == NULL)) {
+				CPT_LOG_DP_ERR("Could not create temp session");
+				return 0;
+			}
+
+			ret = cpt_sym_inst_fill(qp, op, sess, infl_req,
+						&inst[0]);
+			if (unlikely(ret)) {
+				sym_session_clear(cn10k_cryptodev_driver_id,
+						  op->sym->session);
+				rte_mempool_put(qp->sess_mp, op->sym->session);
+				return 0;
+			}
+			w7 = sess->cpt_inst_w7;
+		}
+	} else {
+		CPT_LOG_DP_ERR("Unsupported op type");
+		return 0;
+	}
+
+	inst[0].res_addr = (uint64_t)&infl_req->res;
+	infl_req->res.cn10k.compcode = CPT_COMP_NOT_DONE;
+	infl_req->cop = op;
+
+	inst[0].w7.u64 = w7;
+
+	return 1;
+}
+
+#define PKTS_PER_LOOP	32
+#define PKTS_PER_STEORL 16
+
+static uint16_t
+cn10k_cpt_enqueue_burst(void *qptr, struct rte_crypto_op **ops, uint16_t nb_ops)
+{
+	uint64_t lmt_base, lmt_arg, io_addr;
+	struct cpt_inflight_req *infl_req;
+	uint16_t nb_allowed, count = 0;
+	struct cnxk_cpt_qp *qp = qptr;
+	struct pending_queue *pend_q;
+	struct cpt_inst_s *inst;
+	uint16_t lmt_id;
+	int ret, i;
+
+	pend_q = &qp->pend_q;
+
+	nb_allowed = qp->lf.nb_desc - pend_q->pending_count;
+	nb_ops = RTE_MIN(nb_ops, nb_allowed);
+
+	if (unlikely(nb_ops == 0))
+		return 0;
+
+	lmt_base = qp->lmtline.lmt_base;
+	io_addr = qp->lmtline.io_addr;
+
+	ROC_LMT_BASE_ID_GET(lmt_base, lmt_id);
+	inst = (struct cpt_inst_s *)lmt_base;
+
+again:
+	for (i = 0; i < RTE_MIN(PKTS_PER_LOOP, nb_ops); i++) {
+		infl_req = &pend_q->req_queue[pend_q->enq_tail];
+		infl_req->op_flags = 0;
+
+		ret = cn10k_cpt_fill_inst(qp, ops + i, &inst[2 * i], infl_req);
+		if (unlikely(ret != 1)) {
+			CPT_LOG_DP_ERR("Could not process op: %p", ops + i);
+			if (i == 0)
+				goto update_pending;
+			break;
+		}
+
+		MOD_INC(pend_q->enq_tail, qp->lf.nb_desc);
+	}
+
+	if (i > PKTS_PER_STEORL) {
+		lmt_arg = ROC_CN10K_CPT_LMT_ARG | (PKTS_PER_STEORL - 1) << 12 |
+			  (uint64_t)lmt_id;
+		roc_lmt_submit_steorl(lmt_arg, io_addr);
+		lmt_arg = ROC_CN10K_CPT_LMT_ARG |
+			  (i - PKTS_PER_STEORL - 1) << 12 |
+			  (uint64_t)(lmt_id + PKTS_PER_STEORL);
+		roc_lmt_submit_steorl(lmt_arg, io_addr);
+	} else {
+		lmt_arg = ROC_CN10K_CPT_LMT_ARG | (i - 1) << 12 |
+			  (uint64_t)lmt_id;
+		roc_lmt_submit_steorl(lmt_arg, io_addr);
+	}
+
+	rte_io_wmb();
+
+	if (nb_ops - i > 0 && i == PKTS_PER_LOOP) {
+		nb_ops -= i;
+		ops += i;
+		count += i;
+		goto again;
+	}
+
+update_pending:
+	pend_q->pending_count += count + i;
+
+	pend_q->time_out = rte_get_timer_cycles() +
+			   DEFAULT_COMMAND_TIMEOUT * rte_get_timer_hz();
+
+	return count + i;
+}
+
+void
+cn10k_cpt_set_enqdeq_fns(struct rte_cryptodev *dev)
+{
+	dev->enqueue_burst = cn10k_cpt_enqueue_burst;
+
+	rte_mb();
+}
 
 static void
 cn10k_cpt_dev_info_get(struct rte_cryptodev *dev,
diff --git a/drivers/crypto/cnxk/cn10k_cryptodev_ops.h b/drivers/crypto/cnxk/cn10k_cryptodev_ops.h
index 24611bf..d500b7d 100644
--- a/drivers/crypto/cnxk/cn10k_cryptodev_ops.h
+++ b/drivers/crypto/cnxk/cn10k_cryptodev_ops.h
@@ -10,4 +10,6 @@
 
 extern struct rte_cryptodev_ops cn10k_cpt_ops;
 
+void cn10k_cpt_set_enqdeq_fns(struct rte_cryptodev *dev);
+
 #endif /* _CN10K_CRYPTODEV_OPS_H_ */
diff --git a/drivers/crypto/cnxk/cn9k_cryptodev.c b/drivers/crypto/cnxk/cn9k_cryptodev.c
index 424f812..7470397 100644
--- a/drivers/crypto/cnxk/cn9k_cryptodev.c
+++ b/drivers/crypto/cnxk/cn9k_cryptodev.c
@@ -77,6 +77,8 @@ cn9k_cpt_pci_probe(struct rte_pci_driver *pci_drv __rte_unused,
 
 	cnxk_cpt_caps_populate(vf);
 
+	cn9k_cpt_set_enqdeq_fns(dev);
+
 	return 0;
 
 dev_fini:
diff --git a/drivers/crypto/cnxk/cn9k_cryptodev_ops.c b/drivers/crypto/cnxk/cn9k_cryptodev_ops.c
index bef6159..59e3cb0 100644
--- a/drivers/crypto/cnxk/cn9k_cryptodev_ops.c
+++ b/drivers/crypto/cnxk/cn9k_cryptodev_ops.c
@@ -7,7 +7,161 @@
 
 #include "cn9k_cryptodev.h"
 #include "cn9k_cryptodev_ops.h"
+#include "cnxk_cryptodev.h"
 #include "cnxk_cryptodev_ops.h"
+#include "cnxk_se.h"
+
+static __rte_always_inline int __rte_hot
+cn9k_cpt_sym_inst_fill(struct cnxk_cpt_qp *qp, struct rte_crypto_op *op,
+		       struct cnxk_se_sess *sess,
+		       struct cpt_inflight_req *infl_req,
+		       struct cpt_inst_s *inst)
+{
+	RTE_SET_USED(qp);
+	RTE_SET_USED(op);
+	RTE_SET_USED(sess);
+	RTE_SET_USED(infl_req);
+	RTE_SET_USED(inst);
+
+	return -ENOTSUP;
+}
+
+static inline struct cnxk_se_sess *
+cn9k_cpt_sym_temp_sess_create(struct cnxk_cpt_qp *qp, struct rte_crypto_op *op)
+{
+	const int driver_id = cn9k_cryptodev_driver_id;
+	struct rte_crypto_sym_op *sym_op = op->sym;
+	struct rte_cryptodev_sym_session *sess;
+	struct cnxk_se_sess *priv;
+	int ret;
+
+	/* Create temporary session */
+	sess = rte_cryptodev_sym_session_create(qp->sess_mp);
+	if (sess == NULL)
+		return NULL;
+
+	ret = sym_session_configure(qp->lf.roc_cpt, driver_id, sym_op->xform,
+				    sess, qp->sess_mp_priv);
+	if (ret)
+		goto sess_put;
+
+	priv = get_sym_session_private_data(sess, driver_id);
+
+	sym_op->session = sess;
+
+	return priv;
+
+sess_put:
+	rte_mempool_put(qp->sess_mp, sess);
+	return NULL;
+}
+
+static uint16_t
+cn9k_cpt_enqueue_burst(void *qptr, struct rte_crypto_op **ops, uint16_t nb_ops)
+{
+	struct cpt_inflight_req *infl_req;
+	struct rte_crypto_sym_op *sym_op;
+	uint16_t nb_allowed, count = 0;
+	struct cnxk_cpt_qp *qp = qptr;
+	struct pending_queue *pend_q;
+	struct cnxk_se_sess *sess;
+	struct rte_crypto_op *op;
+	struct cpt_inst_s inst;
+	uint64_t lmt_status;
+	uint64_t lmtline;
+	uint64_t io_addr;
+	int ret;
+
+	pend_q = &qp->pend_q;
+
+	lmtline = qp->lmtline.lmt_base;
+	io_addr = qp->lmtline.io_addr;
+
+	inst.w0.u64 = 0;
+	inst.w2.u64 = 0;
+	inst.w3.u64 = 0;
+
+	nb_allowed = qp->lf.nb_desc - pend_q->pending_count;
+	nb_ops = RTE_MIN(nb_ops, nb_allowed);
+
+	for (count = 0; count < nb_ops; count++) {
+		op = ops[count];
+		infl_req = &pend_q->req_queue[pend_q->enq_tail];
+		infl_req->op_flags = 0;
+
+		if (op->type == RTE_CRYPTO_OP_TYPE_SYMMETRIC) {
+			if (op->sess_type == RTE_CRYPTO_OP_WITH_SESSION) {
+				sym_op = op->sym;
+				sess = get_sym_session_private_data(
+					sym_op->session,
+					cn9k_cryptodev_driver_id);
+				ret = cn9k_cpt_sym_inst_fill(qp, op, sess,
+							     infl_req, &inst);
+			} else {
+				sess = cn9k_cpt_sym_temp_sess_create(qp, op);
+				if (unlikely(sess == NULL)) {
+					CPT_LOG_DP_ERR(
+						"Could not create temp session");
+					break;
+				}
+
+				ret = cn9k_cpt_sym_inst_fill(qp, op, sess,
+							     infl_req, &inst);
+				if (unlikely(ret)) {
+					sym_session_clear(
+						cn9k_cryptodev_driver_id,
+						op->sym->session);
+					rte_mempool_put(qp->sess_mp,
+							op->sym->session);
+				}
+			}
+		} else {
+			CPT_LOG_DP_ERR("Unsupported op type");
+			break;
+		}
+
+		if (unlikely(ret)) {
+			CPT_LOG_DP_ERR("Could not process op: %p", op);
+			break;
+		}
+
+		infl_req->cop = op;
+
+		infl_req->res.cn9k.compcode = CPT_COMP_NOT_DONE;
+		inst.res_addr = (uint64_t)&infl_req->res;
+		inst.w7.u64 = sess->cpt_inst_w7;
+
+		do {
+			/* Copy CPT command to LMTLINE */
+			memcpy((void *)lmtline, &inst, sizeof(inst));
+
+			/*
+			 * Make sure compiler does not reorder memcpy and ldeor.
+			 * LMTST transactions are always flushed from the write
+			 * buffer immediately, a DMB is not required to push out
+			 * LMTSTs.
+			 */
+			rte_io_wmb();
+			lmt_status = roc_lmt_submit_ldeor(io_addr);
+		} while (lmt_status == 0);
+
+		MOD_INC(pend_q->enq_tail, qp->lf.nb_desc);
+	}
+
+	pend_q->pending_count += count;
+	pend_q->time_out = rte_get_timer_cycles() +
+			   DEFAULT_COMMAND_TIMEOUT * rte_get_timer_hz();
+
+	return count;
+}
+
+void
+cn9k_cpt_set_enqdeq_fns(struct rte_cryptodev *dev)
+{
+	dev->enqueue_burst = cn9k_cpt_enqueue_burst;
+
+	rte_mb();
+}
 
 static void
 cn9k_cpt_dev_info_get(struct rte_cryptodev *dev,
diff --git a/drivers/crypto/cnxk/cn9k_cryptodev_ops.h b/drivers/crypto/cnxk/cn9k_cryptodev_ops.h
index 72fc297..2277f6b 100644
--- a/drivers/crypto/cnxk/cn9k_cryptodev_ops.h
+++ b/drivers/crypto/cnxk/cn9k_cryptodev_ops.h
@@ -9,4 +9,6 @@
 
 extern struct rte_cryptodev_ops cn9k_cpt_ops;
 
+void cn9k_cpt_set_enqdeq_fns(struct rte_cryptodev *dev);
+
 #endif /* _CN9K_CRYPTODEV_OPS_H_ */
diff --git a/drivers/crypto/cnxk/cnxk_cryptodev_ops.h b/drivers/crypto/cnxk/cnxk_cryptodev_ops.h
index 8f9b4fe..b252c52 100644
--- a/drivers/crypto/cnxk/cnxk_cryptodev_ops.h
+++ b/drivers/crypto/cnxk/cnxk_cryptodev_ops.h
@@ -12,6 +12,11 @@
 #define CNXK_CPT_MIN_HEADROOM_REQ 24
 #define CNXK_CPT_MIN_TAILROOM_REQ 8
 
+/* Default command timeout in seconds */
+#define DEFAULT_COMMAND_TIMEOUT 4
+
+#define MOD_INC(i, l) ((i) == (l - 1) ? (i) = 0 : (i)++)
+
 struct cpt_qp_meta_info {
 	struct rte_mempool *pool;
 	int mlen;
@@ -27,6 +32,10 @@ enum sym_xform_type {
 	CNXK_CPT_CIPHER_DEC_AUTH_VRFY
 };
 
+#define CPT_OP_FLAGS_METABUF	       (1 << 1)
+#define CPT_OP_FLAGS_AUTH_VERIFY       (1 << 0)
+#define CPT_OP_FLAGS_IPSEC_DIR_INBOUND (1 << 2)
+
 struct cpt_inflight_req {
 	union cpt_res_s res;
 	struct rte_crypto_op *cop;
-- 
2.7.4