From: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Date: Mon, 8 Oct 2018 18:57:35 -0700
Subject: nfp: flower: use stats array instead of storing stats per flow
Patch-mainline: v4.20-rc1
Git-commit: 7fade1077c5056a9a489110516143bbdf5a013c8
References: bsc#1109837

Make use of an array stats instead of storing stats per flow which
would require a hash lookup at critical times.

Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
 drivers/net/ethernet/netronome/nfp/flower/main.h     |    6 +-
 drivers/net/ethernet/netronome/nfp/flower/metadata.c |   56 ++++++++-----------
 drivers/net/ethernet/netronome/nfp/flower/offload.c  |   21 ++++---
 3 files changed, 41 insertions(+), 42 deletions(-)

--- a/drivers/net/ethernet/netronome/nfp/flower/main.h
+++ b/drivers/net/ethernet/netronome/nfp/flower/main.h
@@ -139,6 +139,8 @@ struct nfp_fl_lag {
  * @mask_ids:		List of free mask ids
  * @mask_table:		Hash table used to store masks
  * @flow_table:		Hash table used to store flower rules
+ * @stats:		Stored stats updates for flower rules
+ * @stats_lock:		Lock for flower rule stats updates
  * @cmsg_work:		Workqueue for control messages processing
  * @cmsg_skbs_high:	List of higher priority skbs for control message
  *			processing
@@ -172,6 +174,8 @@ struct nfp_flower_priv {
 	struct nfp_fl_mask_id mask_ids;
 	DECLARE_HASHTABLE(mask_table, NFP_FLOWER_MASK_HASH_BITS);
 	struct rhashtable flow_table;
+	struct nfp_fl_stats *stats;
+	spinlock_t stats_lock; /* lock stats */
 	struct work_struct cmsg_work;
 	struct sk_buff_head cmsg_skbs_high;
 	struct sk_buff_head cmsg_skbs_low;
@@ -229,8 +233,6 @@ struct nfp_fl_payload {
 	unsigned long tc_flower_cookie;
 	struct rhash_head fl_node;
 	struct rcu_head rcu;
-	spinlock_t lock; /* lock stats */
-	struct nfp_fl_stats stats;
 	__be32 nfp_tun_ipv4_addr;
 	struct net_device *ingress_dev;
 	char *unmasked_data;
--- a/drivers/net/ethernet/netronome/nfp/flower/metadata.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/metadata.c
@@ -119,42 +119,26 @@ nfp_flower_search_fl_table(struct nfp_ap
 				      nfp_flower_table_params);
 }
 
-static void
-nfp_flower_update_stats(struct nfp_app *app, struct nfp_fl_stats_frame *stats)
-{
-	struct nfp_fl_payload *nfp_flow;
-	unsigned long flower_cookie;
-
-	flower_cookie = be64_to_cpu(stats->stats_cookie);
-
-	rcu_read_lock();
-	nfp_flow = nfp_flower_search_fl_table(app, flower_cookie, NULL,
-					      stats->stats_con_id);
-	if (!nfp_flow)
-		goto exit_rcu_unlock;
-
-	spin_lock(&nfp_flow->lock);
-	nfp_flow->stats.pkts += be32_to_cpu(stats->pkt_count);
-	nfp_flow->stats.bytes += be64_to_cpu(stats->byte_count);
-	nfp_flow->stats.used = jiffies;
-	spin_unlock(&nfp_flow->lock);
-
-exit_rcu_unlock:
-	rcu_read_unlock();
-}
-
 void nfp_flower_rx_flow_stats(struct nfp_app *app, struct sk_buff *skb)
 {
 	unsigned int msg_len = nfp_flower_cmsg_get_data_len(skb);
-	struct nfp_fl_stats_frame *stats_frame;
+	struct nfp_flower_priv *priv = app->priv;
+	struct nfp_fl_stats_frame *stats;
 	unsigned char *msg;
+	u32 ctx_id;
 	int i;
 
 	msg = nfp_flower_cmsg_get_data(skb);
 
-	stats_frame = (struct nfp_fl_stats_frame *)msg;
-	for (i = 0; i < msg_len / sizeof(*stats_frame); i++)
-		nfp_flower_update_stats(app, stats_frame + i);
+	spin_lock(&priv->stats_lock);
+	for (i = 0; i < msg_len / sizeof(*stats); i++) {
+		stats = (struct nfp_fl_stats_frame *)msg + i;
+		ctx_id = be32_to_cpu(stats->stats_con_id);
+		priv->stats[ctx_id].pkts += be32_to_cpu(stats->pkt_count);
+		priv->stats[ctx_id].bytes += be64_to_cpu(stats->byte_count);
+		priv->stats[ctx_id].used = jiffies;
+	}
+	spin_unlock(&priv->stats_lock);
 }
 
 static int nfp_release_mask_id(struct nfp_app *app, u8 mask_id)
@@ -348,9 +332,9 @@ int nfp_compile_flow_metadata(struct nfp
 
 	/* Update flow payload with mask ids. */
 	nfp_flow->unmasked_data[NFP_FL_MASK_ID_LOCATION] = new_mask_id;
-	nfp_flow->stats.pkts = 0;
-	nfp_flow->stats.bytes = 0;
-	nfp_flow->stats.used = jiffies;
+	priv->stats[stats_cxt].pkts = 0;
+	priv->stats[stats_cxt].bytes = 0;
+	priv->stats[stats_cxt].used = jiffies;
 
 	check_entry = nfp_flower_search_fl_table(app, flow->cookie, netdev,
 						 NFP_FL_STATS_CTX_DONT_CARE);
@@ -468,8 +452,17 @@ int nfp_flower_metadata_init(struct nfp_
 
 	priv->stats_ids.init_unalloc = NFP_FL_REPEATED_HASH_MAX;
 
+	priv->stats = kvmalloc_array(NFP_FL_STATS_ENTRY_RS,
+				     sizeof(struct nfp_fl_stats), GFP_KERNEL);
+	if (!priv->stats)
+		goto err_free_ring_buf;
+
+	spin_lock_init(&priv->stats_lock);
+
 	return 0;
 
+err_free_ring_buf:
+	vfree(priv->stats_ids.free_list.buf);
 err_free_last_used:
 	kfree(priv->mask_ids.last_used);
 err_free_mask_id:
@@ -488,6 +481,7 @@ void nfp_flower_metadata_cleanup(struct
 
 	rhashtable_free_and_destroy(&priv->flow_table,
 				    nfp_check_rhashtable_empty, NULL);
+	kvfree(priv->stats);
 	kfree(priv->mask_ids.mask_id_free_list.buf);
 	kfree(priv->mask_ids.last_used);
 	vfree(priv->stats_ids.free_list.buf);
--- a/drivers/net/ethernet/netronome/nfp/flower/offload.c
+++ b/drivers/net/ethernet/netronome/nfp/flower/offload.c
@@ -428,8 +428,6 @@ nfp_flower_allocate_new(struct nfp_fl_ke
 
 	flow_pay->nfp_tun_ipv4_addr = 0;
 	flow_pay->meta.flags = 0;
-	spin_lock_init(&flow_pay->lock);
-
 	flow_pay->ingress_offload = !egress;
 
 	return flow_pay;
@@ -604,8 +602,10 @@ static int
 nfp_flower_get_stats(struct nfp_app *app, struct net_device *netdev,
 		     struct tc_cls_flower_offload *flow, bool egress)
 {
+	struct nfp_flower_priv *priv = app->priv;
 	struct nfp_fl_payload *nfp_flow;
 	struct net_device *ingr_dev;
+	u32 ctx_id;
 
 	ingr_dev = egress ? NULL : netdev;
 	nfp_flow = nfp_flower_search_fl_table(app, flow->cookie, ingr_dev,
@@ -616,13 +616,16 @@ nfp_flower_get_stats(struct nfp_app *app
 	if (nfp_flow->ingress_offload && egress)
 		return 0;
 
-	spin_lock_bh(&nfp_flow->lock);
-	tcf_exts_stats_update(flow->exts, nfp_flow->stats.bytes,
-			      nfp_flow->stats.pkts, nfp_flow->stats.used);
-
-	nfp_flow->stats.pkts = 0;
-	nfp_flow->stats.bytes = 0;
-	spin_unlock_bh(&nfp_flow->lock);
+	ctx_id = be32_to_cpu(nfp_flow->meta.host_ctx_id);
+
+	spin_lock_bh(&priv->stats_lock);
+	tcf_exts_stats_update(flow->exts, priv->stats[ctx_id].bytes,
+			      priv->stats[ctx_id].pkts,
+			      priv->stats[ctx_id].used);
+
+	priv->stats[ctx_id].pkts = 0;
+	priv->stats[ctx_id].bytes = 0;
+	spin_unlock_bh(&priv->stats_lock);
 
 	return 0;
 }