cxgb4: fetch stats for offloaded tc flower flows

Add support to retrieve stats from hardware for offloaded tc flower
flows.  Also, poll for the stats of offloaded flows via timer callback.

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/chelsio/cxgb4/cxgb4.h

@@ -908,6 +908,7 @@ struct adapter {
 
 	/* TC flower offload */
 	DECLARE_HASHTABLE(flower_anymatch_tbl, 9);
+	struct timer_list flower_stats_timer;
 };
 
 /* Support for "sched-class" command to allow a TX Scheduling Class to be

drivers/net/ethernet/chelsio/cxgb4/cxgb4_filter.c

@@ -148,6 +148,82 @@ static int get_filter_steerq(struct net_device *dev,
 	return iq;
 }
 
+static int get_filter_count(struct adapter *adapter, unsigned int fidx,
+			    u64 *pkts, u64 *bytes)
+{
+	unsigned int tcb_base, tcbaddr;
+	unsigned int word_offset;
+	struct filter_entry *f;
+	__be64 be64_byte_count;
+	int ret;
+
+	tcb_base = t4_read_reg(adapter, TP_CMM_TCB_BASE_A);
+	if ((fidx != (adapter->tids.nftids + adapter->tids.nsftids - 1)) &&
+	    fidx >= adapter->tids.nftids)
+		return -E2BIG;
+
+	f = &adapter->tids.ftid_tab[fidx];
+	if (!f->valid)
+		return -EINVAL;
+
+	tcbaddr = tcb_base + f->tid * TCB_SIZE;
+
+	spin_lock(&adapter->win0_lock);
+	if (is_t4(adapter->params.chip)) {
+		__be64 be64_count;
+
+		/* T4 doesn't maintain byte counts in hw */
+		*bytes = 0;
+
+		/* Get pkts */
+		word_offset = 4;
+		ret = t4_memory_rw(adapter, MEMWIN_NIC, MEM_EDC0,
+				   tcbaddr + (word_offset * sizeof(__be32)),
+				   sizeof(be64_count),
+				   (__be32 *)&be64_count,
+				   T4_MEMORY_READ);
+		if (ret < 0)
+			goto out;
+		*pkts = be64_to_cpu(be64_count);
+	} else {
+		__be32 be32_count;
+
+		/* Get bytes */
+		word_offset = 4;
+		ret = t4_memory_rw(adapter, MEMWIN_NIC, MEM_EDC0,
+				   tcbaddr + (word_offset * sizeof(__be32)),
+				   sizeof(be64_byte_count),
+				   &be64_byte_count,
+				   T4_MEMORY_READ);
+		if (ret < 0)
+			goto out;
+		*bytes = be64_to_cpu(be64_byte_count);
+
+		/* Get pkts */
+		word_offset = 6;
+		ret = t4_memory_rw(adapter, MEMWIN_NIC, MEM_EDC0,
+				   tcbaddr + (word_offset * sizeof(__be32)),
+				   sizeof(be32_count),
+				   &be32_count,
+				   T4_MEMORY_READ);
+		if (ret < 0)
+			goto out;
+		*pkts = (u64)be32_to_cpu(be32_count);
+	}
+
+out:
+	spin_unlock(&adapter->win0_lock);
+	return ret;
+}
+
+int cxgb4_get_filter_counters(struct net_device *dev, unsigned int fidx,
+			      u64 *hitcnt, u64 *bytecnt)
+{
+	struct adapter *adapter = netdev2adap(dev);
+
+	return get_filter_count(adapter, fidx, hitcnt, bytecnt);
+}
+
 int cxgb4_get_free_ftid(struct net_device *dev, int family)
 {
 	struct adapter *adap = netdev2adap(dev);

drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c

@@ -4637,6 +4637,7 @@ static void free_some_resources(struct adapter *adapter)
 	kvfree(adapter->l2t);
 	t4_cleanup_sched(adapter);
 	kvfree(adapter->tids.tid_tab);
+	cxgb4_cleanup_tc_flower(adapter);
 	cxgb4_cleanup_tc_u32(adapter);
 	kfree(adapter->sge.egr_map);
 	kfree(adapter->sge.ingr_map);

drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c

@@ -39,9 +39,12 @@
 #include "cxgb4.h"
 #include "cxgb4_tc_flower.h"
 
+#define STATS_CHECK_PERIOD (HZ / 2)
+
 static struct ch_tc_flower_entry *allocate_flower_entry(void)
 {
 	struct ch_tc_flower_entry *new = kzalloc(sizeof(*new), GFP_KERNEL);
+	spin_lock_init(&new->lock);
 	return new;
 }
 
@@ -363,13 +366,87 @@ err:
 	return ret;
 }
 
+void ch_flower_stats_cb(unsigned long data)
+{
+	struct adapter *adap = (struct adapter *)data;
+	struct ch_tc_flower_entry *flower_entry;
+	struct ch_tc_flower_stats *ofld_stats;
+	unsigned int i;
+	u64 packets;
+	u64 bytes;
+	int ret;
+
+	rcu_read_lock();
+	hash_for_each_rcu(adap->flower_anymatch_tbl, i, flower_entry, link) {
+		ret = cxgb4_get_filter_counters(adap->port[0],
+						flower_entry->filter_id,
+						&packets, &bytes);
+		if (!ret) {
+			spin_lock(&flower_entry->lock);
+			ofld_stats = &flower_entry->stats;
+
+			if (ofld_stats->prev_packet_count != packets) {
+				ofld_stats->prev_packet_count = packets;
+				ofld_stats->last_used = jiffies;
+			}
+			spin_unlock(&flower_entry->lock);
+		}
+	}
+	rcu_read_unlock();
+	mod_timer(&adap->flower_stats_timer, jiffies + STATS_CHECK_PERIOD);
+}
+
 int cxgb4_tc_flower_stats(struct net_device *dev,
 			  struct tc_cls_flower_offload *cls)
 {
-	return -EOPNOTSUPP;
+	struct adapter *adap = netdev2adap(dev);
+	struct ch_tc_flower_stats *ofld_stats;
+	struct ch_tc_flower_entry *ch_flower;
+	u64 packets;
+	u64 bytes;
+	int ret;
+
+	ch_flower = ch_flower_lookup(adap, cls->cookie);
+	if (!ch_flower) {
+		ret = -ENOENT;
+		goto err;
+	}
+
+	ret = cxgb4_get_filter_counters(dev, ch_flower->filter_id,
+					&packets, &bytes);
+	if (ret < 0)
+		goto err;
+
+	spin_lock_bh(&ch_flower->lock);
+	ofld_stats = &ch_flower->stats;
+	if (ofld_stats->packet_count != packets) {
+		if (ofld_stats->prev_packet_count != packets)
+			ofld_stats->last_used = jiffies;
+		tcf_exts_stats_update(cls->exts, bytes - ofld_stats->byte_count,
+				      packets - ofld_stats->packet_count,
+				      ofld_stats->last_used);
+
+		ofld_stats->packet_count = packets;
+		ofld_stats->byte_count = bytes;
+		ofld_stats->prev_packet_count = packets;
+	}
+	spin_unlock_bh(&ch_flower->lock);
+	return 0;
+
+err:
+	return ret;
 }
 
 void cxgb4_init_tc_flower(struct adapter *adap)
 {
 	hash_init(adap->flower_anymatch_tbl);
+	setup_timer(&adap->flower_stats_timer, ch_flower_stats_cb,
+		    (unsigned long)adap);
+	mod_timer(&adap->flower_stats_timer, jiffies + STATS_CHECK_PERIOD);
+}
+
+void cxgb4_cleanup_tc_flower(struct adapter *adap)
+{
+	if (adap->flower_stats_timer.function)
+		del_timer_sync(&adap->flower_stats_timer);
 }

drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.h

@@ -38,6 +38,7 @@
 #include <net/pkt_cls.h>
 
 struct ch_tc_flower_stats {
+	u64 prev_packet_count;
 	u64 packet_count;
 	u64 byte_count;
 	u64 last_used;
@@ -49,6 +50,7 @@ struct ch_tc_flower_entry {
 	unsigned long tc_flower_cookie;
 	struct hlist_node link;
 	struct rcu_head rcu;
+	spinlock_t lock; /* lock for stats */
 	u32 filter_id;
 };
 
@@ -60,4 +62,5 @@ int cxgb4_tc_flower_stats(struct net_device *dev,
 			  struct tc_cls_flower_offload *cls);
 
 void cxgb4_init_tc_flower(struct adapter *adap);
+void cxgb4_cleanup_tc_flower(struct adapter *adap);
 #endif /* __CXGB4_TC_FLOWER_H */

drivers/net/ethernet/chelsio/cxgb4/cxgb4_uld.h

@@ -221,6 +221,8 @@ int __cxgb4_del_filter(struct net_device *dev, int filter_id,
 int cxgb4_set_filter(struct net_device *dev, int filter_id,
 		     struct ch_filter_specification *fs);
 int cxgb4_del_filter(struct net_device *dev, int filter_id);
+int cxgb4_get_filter_counters(struct net_device *dev, unsigned int fidx,
+			      u64 *hitcnt, u64 *bytecnt);
 
 static inline void set_wr_txq(struct sk_buff *skb, int prio, int queue)
 {