[v5,01/11] sched: remove wrr from strict priority tc queues

Message ID 20190717144245.138876-2-jasvinder.singh@intel.com
State Superseded, archived
Delegated to: Cristian Dumitrescu
Headers show
Series
  • sched: feature enhancements
Related show

Checks

Context Check Description
ci/intel-Performance-Testing success Performance Testing PASS
ci/mellanox-Performance-Testing success Performance Testing PASS
ci/Intel-compilation success Compilation OK
ci/checkpatch success coding style OK

Commit Message

Jasvinder Singh July 17, 2019, 2:42 p.m.
All higher priority traffic classes contain only one queue, thus
remove wrr function for them. The lowest priority best-effort
traffic class conitnue to have multiple queues and packet are
scheduled from its queues using wrr function.

Signed-off-by: Jasvinder Singh <jasvinder.singh@intel.com>
Signed-off-by: Abraham Tovar <abrahamx.tovar@intel.com>
Signed-off-by: Lukasz Krakowiak <lukaszx.krakowiak@intel.com>
---
 app/test/test_sched.c        |   2 +-
 examples/qos_sched/init.c    |   2 +-
 lib/librte_sched/Makefile    |   2 +-
 lib/librte_sched/meson.build |   2 +-
 lib/librte_sched/rte_sched.c | 152 +++++++++++++++++++----------------
 lib/librte_sched/rte_sched.h |  22 +++--
 6 files changed, 104 insertions(+), 78 deletions(-)

Patch

diff --git a/app/test/test_sched.c b/app/test/test_sched.c
index 49bb9ea6f..36fa2d425 100644
--- a/app/test/test_sched.c
+++ b/app/test/test_sched.c
@@ -40,7 +40,7 @@  static struct rte_sched_pipe_params pipe_profile[] = {
 		.tc_rate = {305175, 305175, 305175, 305175},
 		.tc_period = 40,
 
-		.wrr_weights = {1, 1, 1, 1,  1, 1, 1, 1,  1, 1, 1, 1,  1, 1, 1, 1},
+		.wrr_weights = {1, 1, 1, 1},
 	},
 };
 
diff --git a/examples/qos_sched/init.c b/examples/qos_sched/init.c
index 1209bd7ce..6b63d4e0e 100644
--- a/examples/qos_sched/init.c
+++ b/examples/qos_sched/init.c
@@ -186,7 +186,7 @@  static struct rte_sched_pipe_params pipe_profiles[RTE_SCHED_PIPE_PROFILES_PER_PO
 		.tc_ov_weight = 1,
 #endif
 
-		.wrr_weights = {1, 1, 1, 1,  1, 1, 1, 1,  1, 1, 1, 1,  1, 1, 1, 1},
+		.wrr_weights = {1, 1, 1, 1},
 	},
 };
 
diff --git a/lib/librte_sched/Makefile b/lib/librte_sched/Makefile
index 644fd9d15..3d7f410e1 100644
--- a/lib/librte_sched/Makefile
+++ b/lib/librte_sched/Makefile
@@ -18,7 +18,7 @@  LDLIBS += -lrte_timer
 
 EXPORT_MAP := rte_sched_version.map
 
-LIBABIVER := 2
+LIBABIVER := 3
 
 #
 # all source are stored in SRCS-y
diff --git a/lib/librte_sched/meson.build b/lib/librte_sched/meson.build
index 8e989e5f6..59d43c6d8 100644
--- a/lib/librte_sched/meson.build
+++ b/lib/librte_sched/meson.build
@@ -1,7 +1,7 @@ 
 # SPDX-License-Identifier: BSD-3-Clause
 # Copyright(c) 2017 Intel Corporation
 
-version = 2
+version = 3
 sources = files('rte_sched.c', 'rte_red.c', 'rte_approx.c')
 headers = files('rte_sched.h', 'rte_sched_common.h',
 		'rte_red.h', 'rte_approx.h')
diff --git a/lib/librte_sched/rte_sched.c b/lib/librte_sched/rte_sched.c
index bc06bc3f4..f7c218ef0 100644
--- a/lib/librte_sched/rte_sched.c
+++ b/lib/librte_sched/rte_sched.c
@@ -37,6 +37,8 @@ 
 
 #define RTE_SCHED_TB_RATE_CONFIG_ERR          (1e-7)
 #define RTE_SCHED_WRR_SHIFT                   3
+#define RTE_SCHED_TRAFFIC_CLASS_BE            (RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE - 1)
+#define RTE_SCHED_MAX_QUEUES_PER_TC           RTE_SCHED_BE_QUEUES_PER_PIPE
 #define RTE_SCHED_GRINDER_PCACHE_SIZE         (64 / RTE_SCHED_QUEUES_PER_PIPE)
 #define RTE_SCHED_PIPE_INVALID                UINT32_MAX
 #define RTE_SCHED_BMP_POS_INVALID             UINT32_MAX
@@ -84,8 +86,8 @@  struct rte_sched_pipe_profile {
 	uint32_t tc_credits_per_period[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
 	uint8_t tc_ov_weight;
 
-	/* Pipe queues */
-	uint8_t  wrr_cost[RTE_SCHED_QUEUES_PER_PIPE];
+	/* Pipe best-effort traffic class queues */
+	uint8_t  wrr_cost[RTE_SCHED_BE_QUEUES_PER_PIPE];
 };
 
 struct rte_sched_pipe {
@@ -101,7 +103,7 @@  struct rte_sched_pipe {
 	uint32_t tc_credits[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
 
 	/* Weighted Round Robin (WRR) */
-	uint8_t wrr_tokens[RTE_SCHED_QUEUES_PER_PIPE];
+	uint8_t wrr_tokens[RTE_SCHED_BE_QUEUES_PER_PIPE];
 
 	/* TC oversubscription */
 	uint32_t tc_ov_credits;
@@ -153,16 +155,16 @@  struct rte_sched_grinder {
 	uint32_t tc_index;
 	struct rte_sched_queue *queue[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
 	struct rte_mbuf **qbase[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
-	uint32_t qindex[RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE];
+	uint32_t qindex[RTE_SCHED_MAX_QUEUES_PER_TC];
 	uint16_t qsize;
 	uint32_t qmask;
 	uint32_t qpos;
 	struct rte_mbuf *pkt;
 
 	/* WRR */
-	uint16_t wrr_tokens[RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS];
-	uint16_t wrr_mask[RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS];
-	uint8_t wrr_cost[RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS];
+	uint16_t wrr_tokens[RTE_SCHED_BE_QUEUES_PER_PIPE];
+	uint16_t wrr_mask[RTE_SCHED_BE_QUEUES_PER_PIPE];
+	uint8_t wrr_cost[RTE_SCHED_BE_QUEUES_PER_PIPE];
 };
 
 struct rte_sched_port {
@@ -483,7 +485,7 @@  rte_sched_port_log_pipe_profile(struct rte_sched_port *port, uint32_t i)
 		"    Token bucket: period = %u, credits per period = %u, size = %u\n"
 		"    Traffic classes: period = %u, credits per period = [%u, %u, %u, %u]\n"
 		"    Traffic class 3 oversubscription: weight = %hhu\n"
-		"    WRR cost: [%hhu, %hhu, %hhu, %hhu], [%hhu, %hhu, %hhu, %hhu], [%hhu, %hhu, %hhu, %hhu], [%hhu, %hhu, %hhu, %hhu]\n",
+		"    WRR cost: [%hhu, %hhu, %hhu, %hhu]\n",
 		i,
 
 		/* Token bucket */
@@ -502,10 +504,7 @@  rte_sched_port_log_pipe_profile(struct rte_sched_port *port, uint32_t i)
 		p->tc_ov_weight,
 
 		/* WRR */
-		p->wrr_cost[ 0], p->wrr_cost[ 1], p->wrr_cost[ 2], p->wrr_cost[ 3],
-		p->wrr_cost[ 4], p->wrr_cost[ 5], p->wrr_cost[ 6], p->wrr_cost[ 7],
-		p->wrr_cost[ 8], p->wrr_cost[ 9], p->wrr_cost[10], p->wrr_cost[11],
-		p->wrr_cost[12], p->wrr_cost[13], p->wrr_cost[14], p->wrr_cost[15]);
+		p->wrr_cost[0], p->wrr_cost[1], p->wrr_cost[2], p->wrr_cost[3]);
 }
 
 static inline uint64_t
@@ -523,6 +522,8 @@  rte_sched_pipe_profile_convert(struct rte_sched_pipe_params *src,
 	struct rte_sched_pipe_profile *dst,
 	uint32_t rate)
 {
+	uint32_t wrr_cost[RTE_SCHED_BE_QUEUES_PER_PIPE];
+	uint32_t lcd1, lcd2, lcd;
 	uint32_t i;
 
 	/* Token Bucket */
@@ -553,33 +554,25 @@  rte_sched_pipe_profile_convert(struct rte_sched_pipe_params *src,
 	dst->tc_ov_weight = src->tc_ov_weight;
 #endif
 
-	/* WRR */
-	for (i = 0; i < RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE; i++) {
-		uint32_t wrr_cost[RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS];
-		uint32_t lcd, lcd1, lcd2;
-		uint32_t qindex;
-
-		qindex = i * RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS;
-
-		wrr_cost[0] = src->wrr_weights[qindex];
-		wrr_cost[1] = src->wrr_weights[qindex + 1];
-		wrr_cost[2] = src->wrr_weights[qindex + 2];
-		wrr_cost[3] = src->wrr_weights[qindex + 3];
-
-		lcd1 = rte_get_lcd(wrr_cost[0], wrr_cost[1]);
-		lcd2 = rte_get_lcd(wrr_cost[2], wrr_cost[3]);
-		lcd = rte_get_lcd(lcd1, lcd2);
-
-		wrr_cost[0] = lcd / wrr_cost[0];
-		wrr_cost[1] = lcd / wrr_cost[1];
-		wrr_cost[2] = lcd / wrr_cost[2];
-		wrr_cost[3] = lcd / wrr_cost[3];
-
-		dst->wrr_cost[qindex] = (uint8_t) wrr_cost[0];
-		dst->wrr_cost[qindex + 1] = (uint8_t) wrr_cost[1];
-		dst->wrr_cost[qindex + 2] = (uint8_t) wrr_cost[2];
-		dst->wrr_cost[qindex + 3] = (uint8_t) wrr_cost[3];
-	}
+	/* WRR queues */
+	wrr_cost[0] = src->wrr_weights[0];
+	wrr_cost[1] = src->wrr_weights[1];
+	wrr_cost[2] = src->wrr_weights[2];
+	wrr_cost[3] = src->wrr_weights[3];
+
+	lcd1 = rte_get_lcd(wrr_cost[0], wrr_cost[1]);
+	lcd2 = rte_get_lcd(wrr_cost[2], wrr_cost[3]);
+	lcd = rte_get_lcd(lcd1, lcd2);
+
+	wrr_cost[0] = lcd / wrr_cost[0];
+	wrr_cost[1] = lcd / wrr_cost[1];
+	wrr_cost[2] = lcd / wrr_cost[2];
+	wrr_cost[3] = lcd / wrr_cost[3];
+
+	dst->wrr_cost[0] = (uint8_t) wrr_cost[0];
+	dst->wrr_cost[1] = (uint8_t) wrr_cost[1];
+	dst->wrr_cost[2] = (uint8_t) wrr_cost[2];
+	dst->wrr_cost[3] = (uint8_t) wrr_cost[3];
 }
 
 static void
@@ -1715,6 +1708,7 @@  grinder_schedule(struct rte_sched_port *port, uint32_t pos)
 	struct rte_sched_queue *queue = grinder->queue[grinder->qpos];
 	struct rte_mbuf *pkt = grinder->pkt;
 	uint32_t pkt_len = pkt->pkt_len + port->frame_overhead;
+	uint32_t be_tc_active;
 
 	if (!grinder_credits_check(port, pos))
 		return 0;
@@ -1725,13 +1719,18 @@  grinder_schedule(struct rte_sched_port *port, uint32_t pos)
 	/* Send packet */
 	port->pkts_out[port->n_pkts_out++] = pkt;
 	queue->qr++;
-	grinder->wrr_tokens[grinder->qpos] += pkt_len * grinder->wrr_cost[grinder->qpos];
+
+	be_tc_active = (grinder->tc_index == RTE_SCHED_TRAFFIC_CLASS_BE) ? ~0x0 : 0x0;
+	grinder->wrr_tokens[grinder->qpos] +=
+		(pkt_len * grinder->wrr_cost[grinder->qpos]) & be_tc_active;
+
 	if (queue->qr == queue->qw) {
 		uint32_t qindex = grinder->qindex[grinder->qpos];
 
 		rte_bitmap_clear(port->bmp, qindex);
 		grinder->qmask &= ~(1 << grinder->qpos);
-		grinder->wrr_mask[grinder->qpos] = 0;
+		if (be_tc_active)
+			grinder->wrr_mask[grinder->qpos] = 0;
 		rte_sched_port_set_queue_empty_timestamp(port, qindex);
 	}
 
@@ -1962,26 +1961,26 @@  grinder_wrr_load(struct rte_sched_port *port, uint32_t pos)
 	struct rte_sched_grinder *grinder = port->grinder + pos;
 	struct rte_sched_pipe *pipe = grinder->pipe;
 	struct rte_sched_pipe_profile *pipe_params = grinder->pipe_params;
-	uint32_t tc_index = grinder->tc_index;
 	uint32_t qmask = grinder->qmask;
-	uint32_t qindex;
-
-	qindex = tc_index * 4;
 
-	grinder->wrr_tokens[0] = ((uint16_t) pipe->wrr_tokens[qindex]) << RTE_SCHED_WRR_SHIFT;
-	grinder->wrr_tokens[1] = ((uint16_t) pipe->wrr_tokens[qindex + 1]) << RTE_SCHED_WRR_SHIFT;
-	grinder->wrr_tokens[2] = ((uint16_t) pipe->wrr_tokens[qindex + 2]) << RTE_SCHED_WRR_SHIFT;
-	grinder->wrr_tokens[3] = ((uint16_t) pipe->wrr_tokens[qindex + 3]) << RTE_SCHED_WRR_SHIFT;
+	grinder->wrr_tokens[0] =
+		((uint16_t) pipe->wrr_tokens[0]) << RTE_SCHED_WRR_SHIFT;
+	grinder->wrr_tokens[1] =
+		((uint16_t) pipe->wrr_tokens[1]) << RTE_SCHED_WRR_SHIFT;
+	grinder->wrr_tokens[2] =
+		((uint16_t) pipe->wrr_tokens[2]) << RTE_SCHED_WRR_SHIFT;
+	grinder->wrr_tokens[3] =
+		((uint16_t) pipe->wrr_tokens[3]) << RTE_SCHED_WRR_SHIFT;
 
 	grinder->wrr_mask[0] = (qmask & 0x1) * 0xFFFF;
 	grinder->wrr_mask[1] = ((qmask >> 1) & 0x1) * 0xFFFF;
 	grinder->wrr_mask[2] = ((qmask >> 2) & 0x1) * 0xFFFF;
 	grinder->wrr_mask[3] = ((qmask >> 3) & 0x1) * 0xFFFF;
 
-	grinder->wrr_cost[0] = pipe_params->wrr_cost[qindex];
-	grinder->wrr_cost[1] = pipe_params->wrr_cost[qindex + 1];
-	grinder->wrr_cost[2] = pipe_params->wrr_cost[qindex + 2];
-	grinder->wrr_cost[3] = pipe_params->wrr_cost[qindex + 3];
+	grinder->wrr_cost[0] = pipe_params->wrr_cost[0];
+	grinder->wrr_cost[1] = pipe_params->wrr_cost[1];
+	grinder->wrr_cost[2] = pipe_params->wrr_cost[2];
+	grinder->wrr_cost[3] = pipe_params->wrr_cost[3];
 }
 
 static inline void
@@ -1989,19 +1988,19 @@  grinder_wrr_store(struct rte_sched_port *port, uint32_t pos)
 {
 	struct rte_sched_grinder *grinder = port->grinder + pos;
 	struct rte_sched_pipe *pipe = grinder->pipe;
-	uint32_t tc_index = grinder->tc_index;
-	uint32_t qindex;
 
-	qindex = tc_index * 4;
-
-	pipe->wrr_tokens[qindex] = (grinder->wrr_tokens[0] & grinder->wrr_mask[0])
-		>> RTE_SCHED_WRR_SHIFT;
-	pipe->wrr_tokens[qindex + 1] = (grinder->wrr_tokens[1] & grinder->wrr_mask[1])
-		>> RTE_SCHED_WRR_SHIFT;
-	pipe->wrr_tokens[qindex + 2] = (grinder->wrr_tokens[2] & grinder->wrr_mask[2])
-		>> RTE_SCHED_WRR_SHIFT;
-	pipe->wrr_tokens[qindex + 3] = (grinder->wrr_tokens[3] & grinder->wrr_mask[3])
-		>> RTE_SCHED_WRR_SHIFT;
+	pipe->wrr_tokens[0] =
+			(grinder->wrr_tokens[0] & grinder->wrr_mask[0]) >>
+				RTE_SCHED_WRR_SHIFT;
+	pipe->wrr_tokens[1] =
+			(grinder->wrr_tokens[1] & grinder->wrr_mask[1]) >>
+				RTE_SCHED_WRR_SHIFT;
+	pipe->wrr_tokens[2] =
+			(grinder->wrr_tokens[2] & grinder->wrr_mask[2]) >>
+				RTE_SCHED_WRR_SHIFT;
+	pipe->wrr_tokens[3] =
+			(grinder->wrr_tokens[3] & grinder->wrr_mask[3]) >>
+				RTE_SCHED_WRR_SHIFT;
 }
 
 static inline void
@@ -2040,9 +2039,18 @@  static inline void
 grinder_prefetch_tc_queue_arrays(struct rte_sched_port *port, uint32_t pos)
 {
 	struct rte_sched_grinder *grinder = port->grinder + pos;
-	uint16_t qsize, qr[4];
+	uint16_t qsize, qr[RTE_SCHED_MAX_QUEUES_PER_TC];
 
 	qsize = grinder->qsize;
+	grinder->qpos = 0;
+
+	if (grinder->tc_index < RTE_SCHED_TRAFFIC_CLASS_BE) {
+		qr[0] = grinder->queue[0]->qr & (qsize - 1);
+
+		rte_prefetch0(grinder->qbase[0] + qr[0]);
+		return;
+	}
+
 	qr[0] = grinder->queue[0]->qr & (qsize - 1);
 	qr[1] = grinder->queue[1]->qr & (qsize - 1);
 	qr[2] = grinder->queue[2]->qr & (qsize - 1);
@@ -2118,18 +2126,24 @@  grinder_handle(struct rte_sched_port *port, uint32_t pos)
 
 	case e_GRINDER_READ_MBUF:
 	{
-		uint32_t result = 0;
+		uint32_t wrr_active, result = 0;
 
 		result = grinder_schedule(port, pos);
 
+		wrr_active = (grinder->tc_index == RTE_SCHED_TRAFFIC_CLASS_BE);
+
 		/* Look for next packet within the same TC */
 		if (result && grinder->qmask) {
-			grinder_wrr(port, pos);
+			if (wrr_active)
+				grinder_wrr(port, pos);
+
 			grinder_prefetch_mbuf(port, pos);
 
 			return 1;
 		}
-		grinder_wrr_store(port, pos);
+
+		if (wrr_active)
+			grinder_wrr_store(port, pos);
 
 		/* Look for another active TC within same pipe */
 		if (grinder_next_tc(port, pos)) {
diff --git a/lib/librte_sched/rte_sched.h b/lib/librte_sched/rte_sched.h
index d61dda9f5..f9947c4cd 100644
--- a/lib/librte_sched/rte_sched.h
+++ b/lib/librte_sched/rte_sched.h
@@ -66,6 +66,22 @@  extern "C" {
 #include "rte_red.h"
 #endif
 
+/** Maximum number of queues per pipe.
+ * Note that the multiple queues (power of 2) can only be assigned to
+ * lowest priority (best-effort) traffic class. Other higher priority traffic
+ * classes can only have one queue.
+ * Can not change.
+ *
+ * @see struct rte_sched_port_params
+ */
+#define RTE_SCHED_QUEUES_PER_PIPE    16
+
+/** Number of WRR queues for best-effort traffic class per pipe.
+ *
+ * @see struct rte_sched_pipe_params
+ */
+#define RTE_SCHED_BE_QUEUES_PER_PIPE    4
+
 /** Number of traffic classes per pipe (as well as subport).
  * Cannot be changed.
  */
@@ -74,10 +90,6 @@  extern "C" {
 /** Number of queues per pipe traffic class. Cannot be changed. */
 #define RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS    4
 
-/** Number of queues per pipe. */
-#define RTE_SCHED_QUEUES_PER_PIPE             \
-	(RTE_SCHED_TRAFFIC_CLASSES_PER_PIPE *     \
-	RTE_SCHED_QUEUES_PER_TRAFFIC_CLASS)
 
 /** Maximum number of pipe profiles that can be defined per port.
  * Compile-time configurable.
@@ -165,7 +177,7 @@  struct rte_sched_pipe_params {
 #endif
 
 	/* Pipe queues */
-	uint8_t  wrr_weights[RTE_SCHED_QUEUES_PER_PIPE]; /**< WRR weights */
+	uint8_t  wrr_weights[RTE_SCHED_BE_QUEUES_PER_PIPE]; /**< WRR weights */
 };
 
 /** Queue statistics */