[dpdk-dev] [PATCH v5 4/8] crypto/nitrox: add hardware queue management functionality

[Nagadheeraj Rottela <rnagadheeraj@marvell.com>]

Add hardware queue management code corresponding to queue pair setup
and release functions.

Signed-off-by: Nagadheeraj Rottela <rnagadheeraj@marvell.com>
---
 drivers/crypto/nitrox/nitrox_csr.h |  12 +++
 drivers/crypto/nitrox/nitrox_hal.c | 151 +++++++++++++++++++++++++++++++++++++
 drivers/crypto/nitrox/nitrox_hal.h | 128 +++++++++++++++++++++++++++++++
 drivers/crypto/nitrox/nitrox_qp.c  |  51 +++++++++++--
 drivers/crypto/nitrox/nitrox_qp.h  |   8 ++
 5 files changed, 345 insertions(+), 5 deletions(-)

diff --git a/drivers/crypto/nitrox/nitrox_csr.h b/drivers/crypto/nitrox/nitrox_csr.h
index 879104515..8cd92e38b 100644
--- a/drivers/crypto/nitrox/nitrox_csr.h
+++ b/drivers/crypto/nitrox/nitrox_csr.h
@@ -9,6 +9,18 @@
 #include <rte_io.h>
 
 #define CSR_DELAY	30
+#define NITROX_CSR_ADDR(bar_addr, offset) (bar_addr + (offset))
+
+/* NPS packet registers */
+#define NPS_PKT_IN_INSTR_CTLX(_i)	(0x10060 + ((_i) * 0x40000))
+#define NPS_PKT_IN_INSTR_BADDRX(_i)	(0x10068 + ((_i) * 0x40000))
+#define NPS_PKT_IN_INSTR_RSIZEX(_i)	(0x10070 + ((_i) * 0x40000))
+#define NPS_PKT_IN_DONE_CNTSX(_i)	(0x10080 + ((_i) * 0x40000))
+#define NPS_PKT_IN_INSTR_BAOFF_DBELLX(_i)	(0x10078 + ((_i) * 0x40000))
+#define NPS_PKT_IN_INT_LEVELSX(_i)		(0x10088 + ((_i) * 0x40000))
+#define NPS_PKT_SLC_CTLX(_i)		(0x10000 + ((_i) * 0x40000))
+#define NPS_PKT_SLC_CNTSX(_i)		(0x10008 + ((_i) * 0x40000))
+#define NPS_PKT_SLC_INT_LEVELSX(_i)	(0x10010 + ((_i) * 0x40000))
 
 /* AQM Virtual Function Registers */
 #define AQMQ_QSZX(_i)			(0x20008 + ((_i)*0x40000))
diff --git a/drivers/crypto/nitrox/nitrox_hal.c b/drivers/crypto/nitrox/nitrox_hal.c
index ed1882016..433f3adb2 100644
--- a/drivers/crypto/nitrox/nitrox_hal.c
+++ b/drivers/crypto/nitrox/nitrox_hal.c
@@ -12,6 +12,157 @@
 
 #define MAX_VF_QUEUES	8
 #define MAX_PF_QUEUES	64
+#define NITROX_TIMER_THOLD	0x3FFFFF
+#define NITROX_COUNT_THOLD      0xFFFFFFFF
+
+void
+nps_pkt_input_ring_disable(uint8_t *bar_addr, uint16_t ring)
+{
+	union nps_pkt_in_instr_ctl pkt_in_instr_ctl;
+	uint64_t reg_addr;
+	int max_retries = 5;
+
+	reg_addr = NPS_PKT_IN_INSTR_CTLX(ring);
+	pkt_in_instr_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	pkt_in_instr_ctl.s.enb = 0;
+	nitrox_write_csr(bar_addr, reg_addr, pkt_in_instr_ctl.u64);
+	rte_delay_us_block(100);
+
+	/* wait for enable bit to be cleared */
+	pkt_in_instr_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	while (pkt_in_instr_ctl.s.enb && max_retries--) {
+		rte_delay_ms(10);
+		pkt_in_instr_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	}
+}
+
+void
+nps_pkt_solicited_port_disable(uint8_t *bar_addr, uint16_t port)
+{
+	union nps_pkt_slc_ctl pkt_slc_ctl;
+	uint64_t reg_addr;
+	int max_retries = 5;
+
+	/* clear enable bit */
+	reg_addr = NPS_PKT_SLC_CTLX(port);
+	pkt_slc_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	pkt_slc_ctl.s.enb = 0;
+	nitrox_write_csr(bar_addr, reg_addr, pkt_slc_ctl.u64);
+	rte_delay_us_block(100);
+
+	pkt_slc_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	while (pkt_slc_ctl.s.enb && max_retries--) {
+		rte_delay_ms(10);
+		pkt_slc_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	}
+}
+
+void
+setup_nps_pkt_input_ring(uint8_t *bar_addr, uint16_t ring, uint32_t rsize,
+			 phys_addr_t raddr)
+{
+	union nps_pkt_in_instr_ctl pkt_in_instr_ctl;
+	union nps_pkt_in_instr_rsize pkt_in_instr_rsize;
+	union nps_pkt_in_instr_baoff_dbell pkt_in_instr_baoff_dbell;
+	union nps_pkt_in_done_cnts pkt_in_done_cnts;
+	uint64_t base_addr, reg_addr;
+	int max_retries = 5;
+
+	nps_pkt_input_ring_disable(bar_addr, ring);
+
+	/* write base address */
+	reg_addr = NPS_PKT_IN_INSTR_BADDRX(ring);
+	base_addr = raddr;
+	nitrox_write_csr(bar_addr, reg_addr, base_addr);
+	rte_delay_us_block(CSR_DELAY);
+
+	/* write ring size */
+	reg_addr = NPS_PKT_IN_INSTR_RSIZEX(ring);
+	pkt_in_instr_rsize.u64 = 0;
+	pkt_in_instr_rsize.s.rsize = rsize;
+	nitrox_write_csr(bar_addr, reg_addr, pkt_in_instr_rsize.u64);
+	rte_delay_us_block(CSR_DELAY);
+
+	/* clear door bell */
+	reg_addr = NPS_PKT_IN_INSTR_BAOFF_DBELLX(ring);
+	pkt_in_instr_baoff_dbell.u64 = 0;
+	pkt_in_instr_baoff_dbell.s.dbell = 0xFFFFFFFF;
+	nitrox_write_csr(bar_addr, reg_addr, pkt_in_instr_baoff_dbell.u64);
+	rte_delay_us_block(CSR_DELAY);
+
+	/* clear done count */
+	reg_addr = NPS_PKT_IN_DONE_CNTSX(ring);
+	pkt_in_done_cnts.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	nitrox_write_csr(bar_addr, reg_addr, pkt_in_done_cnts.u64);
+	rte_delay_us_block(CSR_DELAY);
+
+	/* Setup PKT IN RING Interrupt Threshold */
+	reg_addr = NPS_PKT_IN_INT_LEVELSX(ring);
+	nitrox_write_csr(bar_addr, reg_addr, 0xFFFFFFFF);
+	rte_delay_us_block(CSR_DELAY);
+
+	/* enable ring */
+	reg_addr = NPS_PKT_IN_INSTR_CTLX(ring);
+	pkt_in_instr_ctl.u64 = 0;
+	pkt_in_instr_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	pkt_in_instr_ctl.s.is64b = 1;
+	pkt_in_instr_ctl.s.enb = 1;
+	nitrox_write_csr(bar_addr, reg_addr, pkt_in_instr_ctl.u64);
+	rte_delay_us_block(100);
+
+	pkt_in_instr_ctl.u64 = 0;
+	pkt_in_instr_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	/* wait for ring to be enabled */
+	while (!pkt_in_instr_ctl.s.enb && max_retries--) {
+		rte_delay_ms(10);
+		pkt_in_instr_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	}
+}
+
+void
+setup_nps_pkt_solicit_output_port(uint8_t *bar_addr, uint16_t port)
+{
+	union nps_pkt_slc_ctl pkt_slc_ctl;
+	union nps_pkt_slc_cnts pkt_slc_cnts;
+	union nps_pkt_slc_int_levels pkt_slc_int_levels;
+	uint64_t reg_addr;
+	int max_retries = 5;
+
+	nps_pkt_solicited_port_disable(bar_addr, port);
+
+	/* clear pkt counts */
+	reg_addr = NPS_PKT_SLC_CNTSX(port);
+	pkt_slc_cnts.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	nitrox_write_csr(bar_addr, reg_addr, pkt_slc_cnts.u64);
+	rte_delay_us_block(CSR_DELAY);
+
+	/* slc interrupt levels */
+	reg_addr = NPS_PKT_SLC_INT_LEVELSX(port);
+	pkt_slc_int_levels.u64 = 0;
+	pkt_slc_int_levels.s.bmode = 0;
+	pkt_slc_int_levels.s.timet = NITROX_TIMER_THOLD;
+
+	if (NITROX_COUNT_THOLD > 0)
+		pkt_slc_int_levels.s.cnt = NITROX_COUNT_THOLD - 1;
+
+	nitrox_write_csr(bar_addr, reg_addr, pkt_slc_int_levels.u64);
+	rte_delay_us_block(CSR_DELAY);
+
+	/* enable ring */
+	reg_addr = NPS_PKT_SLC_CTLX(port);
+	pkt_slc_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	pkt_slc_ctl.s.rh = 1;
+	pkt_slc_ctl.s.z = 1;
+	pkt_slc_ctl.s.enb = 1;
+	nitrox_write_csr(bar_addr, reg_addr, pkt_slc_ctl.u64);
+	rte_delay_us_block(100);
+
+	pkt_slc_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	while (!pkt_slc_ctl.s.enb && max_retries--) {
+		rte_delay_ms(10);
+		pkt_slc_ctl.u64 = nitrox_read_csr(bar_addr, reg_addr);
+	}
+}
 
 int
 vf_get_vf_config_mode(uint8_t *bar_addr)
diff --git a/drivers/crypto/nitrox/nitrox_hal.h b/drivers/crypto/nitrox/nitrox_hal.h
index 6184211a5..dcfbd11d8 100644
--- a/drivers/crypto/nitrox/nitrox_hal.h
+++ b/drivers/crypto/nitrox/nitrox_hal.h
@@ -10,6 +10,129 @@
 
 #include "nitrox_csr.h"
 
+union nps_pkt_slc_cnts {
+	uint64_t u64;
+	struct {
+#if RTE_BYTE_ORDER == RTE_BIG_ENDIAN
+		uint64_t slc_int : 1;
+		uint64_t uns_int : 1;
+		uint64_t in_int : 1;
+		uint64_t mbox_int : 1;
+		uint64_t resend : 1;
+		uint64_t raz : 5;
+		uint64_t timer : 22;
+		uint64_t cnt : 32;
+#else
+		uint64_t cnt : 32;
+		uint64_t timer : 22;
+		uint64_t raz : 5;
+		uint64_t resend : 1;
+		uint64_t mbox_int : 1;
+		uint64_t in_int : 1;
+		uint64_t uns_int : 1;
+		uint64_t slc_int : 1;
+#endif
+	} s;
+};
+
+union nps_pkt_slc_int_levels {
+	uint64_t u64;
+	struct {
+#if RTE_BYTE_ORDER == RTE_BIG_ENDIAN
+		uint64_t bmode : 1;
+		uint64_t raz : 9;
+		uint64_t timet : 22;
+		uint64_t cnt : 32;
+#else
+		uint64_t cnt : 32;
+		uint64_t timet : 22;
+		uint64_t raz : 9;
+		uint64_t bmode : 1;
+#endif
+	} s;
+};
+
+union nps_pkt_slc_ctl {
+	uint64_t u64;
+	struct {
+#if RTE_BYTE_ORDER == RTE_BIG_ENDIAN
+		uint64_t raz : 61;
+		uint64_t rh : 1;
+		uint64_t z : 1;
+		uint64_t enb : 1;
+#else
+		uint64_t enb : 1;
+		uint64_t z : 1;
+		uint64_t rh : 1;
+		uint64_t raz : 61;
+#endif
+	} s;
+};
+
+union nps_pkt_in_instr_ctl {
+	uint64_t u64;
+	struct {
+#if RTE_BYTE_ORDER == RTE_BIG_ENDIAN
+		uint64_t raz : 62;
+		uint64_t is64b : 1;
+		uint64_t enb : 1;
+#else
+		uint64_t enb : 1;
+		uint64_t is64b : 1;
+		uint64_t raz : 62;
+#endif
+	} s;
+};
+
+union nps_pkt_in_instr_rsize {
+	uint64_t u64;
+	struct {
+#if RTE_BYTE_ORDER == RTE_BIG_ENDIAN
+		uint64_t raz : 32;
+		uint64_t rsize : 32;
+#else
+		uint64_t rsize : 32;
+		uint64_t raz : 32;
+#endif
+	} s;
+};
+
+union nps_pkt_in_instr_baoff_dbell {
+	uint64_t u64;
+	struct {
+#if RTE_BYTE_ORDER == RTE_BIG_ENDIAN
+		uint64_t aoff : 32;
+		uint64_t dbell : 32;
+#else
+		uint64_t dbell : 32;
+		uint64_t aoff : 32;
+#endif
+	} s;
+};
+
+union nps_pkt_in_done_cnts {
+	uint64_t u64;
+	struct {
+#if RTE_BYTE_ORDER == RTE_BIG_ENDIAN
+		uint64_t slc_int : 1;
+		uint64_t uns_int : 1;
+		uint64_t in_int : 1;
+		uint64_t mbox_int : 1;
+		uint64_t resend : 1;
+		uint64_t raz : 27;
+		uint64_t cnt : 32;
+#else
+		uint64_t cnt : 32;
+		uint64_t raz : 27;
+		uint64_t resend : 1;
+		uint64_t mbox_int : 1;
+		uint64_t in_int : 1;
+		uint64_t uns_int : 1;
+		uint64_t slc_int : 1;
+#endif
+	} s;
+};
+
 union aqmq_qsz {
 	uint64_t u64;
 	struct {
@@ -33,5 +156,10 @@ enum nitrox_vf_mode {
 
 int vf_get_vf_config_mode(uint8_t *bar_addr);
 int vf_config_mode_to_nr_queues(enum nitrox_vf_mode vf_mode);
+void setup_nps_pkt_input_ring(uint8_t *bar_addr, uint16_t ring, uint32_t rsize,
+			      phys_addr_t raddr);
+void setup_nps_pkt_solicit_output_port(uint8_t *bar_addr, uint16_t port);
+void nps_pkt_input_ring_disable(uint8_t *bar_addr, uint16_t ring);
+void nps_pkt_solicited_port_disable(uint8_t *bar_addr, uint16_t port);
 
 #endif /* _NITROX_HAL_H_ */
diff --git a/drivers/crypto/nitrox/nitrox_qp.c b/drivers/crypto/nitrox/nitrox_qp.c
index 9673bb4f3..5e85ccbd5 100644
--- a/drivers/crypto/nitrox/nitrox_qp.c
+++ b/drivers/crypto/nitrox/nitrox_qp.c
@@ -10,6 +10,38 @@
 #include "nitrox_logs.h"
 
 #define MAX_CMD_QLEN 16384
+#define CMDQ_PKT_IN_ALIGN 16
+
+static int
+nitrox_setup_cmdq(struct nitrox_qp *qp, uint8_t *bar_addr,
+		  const char *dev_name, uint8_t instr_size, int socket_id)
+{
+	char mz_name[RTE_MEMZONE_NAMESIZE];
+	const struct rte_memzone *mz;
+	size_t cmdq_size = qp->count * instr_size;
+	uint64_t offset;
+
+	snprintf(mz_name, sizeof(mz_name), "%s_cmdq_%d", dev_name, qp->qno);
+	mz = rte_memzone_reserve_aligned(mz_name, cmdq_size, socket_id,
+					 RTE_MEMZONE_SIZE_HINT_ONLY |
+					 RTE_MEMZONE_256MB,
+					 CMDQ_PKT_IN_ALIGN);
+	if (!mz) {
+		NITROX_LOG(ERR, "cmdq memzone reserve failed for %s queue\n",
+			   mz_name);
+		return -ENOMEM;
+	}
+
+	qp->cmdq.mz = mz;
+	offset = NPS_PKT_IN_INSTR_BAOFF_DBELLX(qp->qno);
+	qp->cmdq.dbell_csr_addr = NITROX_CSR_ADDR(bar_addr, offset);
+	qp->cmdq.ring = mz->addr;
+	qp->cmdq.instr_size = instr_size;
+	setup_nps_pkt_input_ring(bar_addr, qp->qno, qp->count, mz->iova);
+	setup_nps_pkt_solicit_output_port(bar_addr, qp->qno);
+
+	return 0;
+}
 
 static int
 nitrox_setup_ridq(struct nitrox_qp *qp, int socket_id)
@@ -27,6 +59,14 @@ nitrox_setup_ridq(struct nitrox_qp *qp, int socket_id)
 	return 0;
 }
 
+static int
+nitrox_release_cmdq(struct nitrox_qp *qp, uint8_t *bar_addr)
+{
+	nps_pkt_solicited_port_disable(bar_addr, qp->qno);
+	nps_pkt_input_ring_disable(bar_addr, qp->qno);
+	return rte_memzone_free(qp->cmdq.mz);
+}
+
 int
 nitrox_qp_setup(struct nitrox_qp *qp, uint8_t *bar_addr, const char *dev_name,
 		uint32_t nb_descriptors, uint8_t instr_size, int socket_id)
@@ -34,8 +74,6 @@ nitrox_qp_setup(struct nitrox_qp *qp, uint8_t *bar_addr, const char *dev_name,
 	int err;
 	uint32_t count;
 
-	RTE_SET_USED(bar_addr);
-	RTE_SET_USED(instr_size);
 	count = rte_align32pow2(nb_descriptors);
 	if (count > MAX_CMD_QLEN) {
 		NITROX_LOG(ERR, "%s: Number of descriptors too big %d,"
@@ -48,6 +86,10 @@ nitrox_qp_setup(struct nitrox_qp *qp, uint8_t *bar_addr, const char *dev_name,
 	qp->count = count;
 	qp->head = qp->tail = 0;
 	rte_atomic16_init(&qp->pending_count);
+	err = nitrox_setup_cmdq(qp, bar_addr, dev_name, instr_size, socket_id);
+	if (err)
+		return err;
+
 	err = nitrox_setup_ridq(qp, socket_id);
 	if (err)
 		goto ridq_err;
@@ -55,8 +97,8 @@ nitrox_qp_setup(struct nitrox_qp *qp, uint8_t *bar_addr, const char *dev_name,
 	return 0;
 
 ridq_err:
+	nitrox_release_cmdq(qp, bar_addr);
 	return err;
-
 }
 
 static void
@@ -68,7 +110,6 @@ nitrox_release_ridq(struct nitrox_qp *qp)
 int
 nitrox_qp_release(struct nitrox_qp *qp, uint8_t *bar_addr)
 {
-	RTE_SET_USED(bar_addr);
 	nitrox_release_ridq(qp);
-	return 0;
+	return nitrox_release_cmdq(qp, bar_addr);
 }
diff --git a/drivers/crypto/nitrox/nitrox_qp.h b/drivers/crypto/nitrox/nitrox_qp.h
index cf0102ff9..0244c4dbf 100644
--- a/drivers/crypto/nitrox/nitrox_qp.h
+++ b/drivers/crypto/nitrox/nitrox_qp.h
@@ -11,11 +11,19 @@
 
 struct nitrox_softreq;
 
+struct command_queue {
+	const struct rte_memzone *mz;
+	uint8_t *dbell_csr_addr;
+	uint8_t *ring;
+	uint8_t instr_size;
+};
+
 struct rid {
 	struct nitrox_softreq *sr;
 };
 
 struct nitrox_qp {
+	struct command_queue cmdq;
 	struct rid *ridq;
 	uint32_t count;
 	uint32_t head;
-- 
2.13.6