[net-queue PATCH 0/5] i40e: stat counter updates and additions

@@ -1382,8 +1382,6 @@ static void i40e_reuse_rx_page(struct i40e_ring *rx_ring,
 	new_buff->page_offset	= old_buff->page_offset;
 	new_buff->pagecnt_bias	= old_buff->pagecnt_bias;
 
-	rx_ring->rx_stats.page_reuse_count++;
-
 	/* clear contents of buffer_info */
 	old_buff->page = NULL;
 }

@@ -853,6 +853,7 @@ struct i40e_vsi {
 	u64 tx_force_wb;
 	u64 rx_buf_failed;
 	u64 rx_page_failed;
+	u64 rx_page_reuse;
 
 	/* These are containers of ring pointers, allocated at run-time */
 	struct i40e_ring **rx_rings;

@@ -295,6 +295,7 @@ static const struct i40e_stats i40e_gstrings_misc_stats[] = {
 	I40E_VSI_STAT("tx_busy", tx_busy),
 	I40E_VSI_STAT("rx_alloc_fail", rx_buf_failed),
 	I40E_VSI_STAT("rx_pg_alloc_fail", rx_page_failed),
+	I40E_VSI_STAT("rx_cache_reuse", rx_page_reuse),
 };
 
 /* These PF_STATs might look like duplicates of some NETDEV_STATs,

@@ -812,7 +812,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	struct i40e_eth_stats *es;     /* device's eth stats */
 	u64 tx_restart, tx_busy;
 	struct i40e_ring *p;
-	u64 rx_page, rx_buf;
+	u64 rx_page, rx_buf, rx_reuse;
 	u64 bytes, packets;
 	unsigned int start;
 	u64 tx_linearize;

@@ -838,6 +838,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	tx_restart = tx_busy = tx_linearize = tx_force_wb = 0;
 	rx_page = 0;
 	rx_buf = 0;
+	rx_reuse = 0;
 	rcu_read_lock();
 	for (q = 0; q < vsi->num_queue_pairs; q++) {
 		/* locate Tx ring */

@@ -871,6 +872,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 		rx_p += packets;
 		rx_buf += p->rx_stats.alloc_buff_failed;
 		rx_page += p->rx_stats.alloc_page_failed;
+		rx_reuse += p->rx_stats.page_reuse_count;
 
 		if (i40e_enabled_xdp_vsi(vsi)) {
 			/* locate XDP ring */

@@ -898,6 +900,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	vsi->tx_force_wb = tx_force_wb;
 	vsi->rx_page_failed = rx_page;
 	vsi->rx_buf_failed = rx_buf;
+	vsi->rx_page_reuse = rx_reuse;
 
 	ns->rx_packets = rx_p;
 	ns->rx_bytes = rx_b;

@@ -854,6 +854,7 @@ struct i40e_vsi {
 	u64 rx_buf_failed;
 	u64 rx_page_failed;
 	u64 rx_page_reuse;
+	u64 rx_page_alloc;
 
 	/* These are containers of ring pointers, allocated at run-time */
 	struct i40e_ring **rx_rings;

@@ -296,6 +296,7 @@ static const struct i40e_stats i40e_gstrings_misc_stats[] = {
 	I40E_VSI_STAT("rx_alloc_fail", rx_buf_failed),
 	I40E_VSI_STAT("rx_pg_alloc_fail", rx_page_failed),
 	I40E_VSI_STAT("rx_cache_reuse", rx_page_reuse),
+	I40E_VSI_STAT("rx_cache_alloc", rx_page_alloc),
 };
 
 /* These PF_STATs might look like duplicates of some NETDEV_STATs,

@@ -812,7 +812,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	struct i40e_eth_stats *es;     /* device's eth stats */
 	u64 tx_restart, tx_busy;
 	struct i40e_ring *p;
-	u64 rx_page, rx_buf, rx_reuse;
+	u64 rx_page, rx_buf, rx_reuse, rx_alloc;
 	u64 bytes, packets;
 	unsigned int start;
 	u64 tx_linearize;

@@ -839,6 +839,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	rx_page = 0;
 	rx_buf = 0;
 	rx_reuse = 0;
+	rx_reuse = 0;
 	rcu_read_lock();
 	for (q = 0; q < vsi->num_queue_pairs; q++) {
 		/* locate Tx ring */

@@ -873,6 +874,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 		rx_buf += p->rx_stats.alloc_buff_failed;
 		rx_page += p->rx_stats.alloc_page_failed;
 		rx_reuse += p->rx_stats.page_reuse_count;
+		rx_alloc += p->rx_stats.page_alloc_count;
 
 		if (i40e_enabled_xdp_vsi(vsi)) {
 			/* locate XDP ring */

@@ -901,6 +903,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	vsi->rx_page_failed = rx_page;
 	vsi->rx_buf_failed = rx_buf;
 	vsi->rx_page_reuse = rx_reuse;
+	vsi->rx_page_alloc = rx_alloc;
 
 	ns->rx_packets = rx_p;
 	ns->rx_bytes = rx_b;

@@ -1671,6 +1671,8 @@ static bool i40e_alloc_mapped_page(struct i40e_ring *rx_ring,
 	if (unlikely(!page)) {
 		rx_ring->rx_stats.alloc_page_failed++;
 		return false;
+	} else {
+		rx_ring->rx_stats.page_alloc_count++;
 	}
 
 	/* map page for use */

@@ -299,6 +299,7 @@ struct i40e_rx_queue_stats {
 	u64 alloc_buff_failed;
 	u64 page_reuse_count;
 	u64 realloc_count;
+	u64 page_alloc_count;
 };
 
 enum i40e_ring_state_t {

@@ -856,6 +856,7 @@ struct i40e_vsi {
 	u64 rx_page_reuse;
 	u64 rx_page_alloc;
 	u64 rx_page_waive;
+	u64 rx_page_busy;
 
 	/* These are containers of ring pointers, allocated at run-time */
 	struct i40e_ring **rx_rings;

@@ -298,6 +298,7 @@ static const struct i40e_stats i40e_gstrings_misc_stats[] = {
 	I40E_VSI_STAT("rx_cache_reuse", rx_page_reuse),
 	I40E_VSI_STAT("rx_cache_alloc", rx_page_alloc),
 	I40E_VSI_STAT("rx_cache_waive", rx_page_waive),
+	I40E_VSI_STAT("rx_cache_busy", rx_page_busy),
 };
 
 /* These PF_STATs might look like duplicates of some NETDEV_STATs,

@@ -812,7 +812,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	struct i40e_eth_stats *es;     /* device's eth stats */
 	u64 tx_restart, tx_busy;
 	struct i40e_ring *p;
-	u64 rx_page, rx_buf, rx_reuse, rx_alloc, rx_waive;
+	u64 rx_page, rx_buf, rx_reuse, rx_alloc, rx_waive, rx_busy;
 	u64 bytes, packets;
 	unsigned int start;
 	u64 tx_linearize;

@@ -841,6 +841,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	rx_reuse = 0;
 	rx_reuse = 0;
 	rx_waive = 0;
+	rx_busy = 0;
 	rcu_read_lock();
 	for (q = 0; q < vsi->num_queue_pairs; q++) {
 		/* locate Tx ring */

@@ -877,6 +878,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 		rx_reuse += p->rx_stats.page_reuse_count;
 		rx_alloc += p->rx_stats.page_alloc_count;
 		rx_waive += p->rx_stats.page_waive_count;
+		rx_busy += p->rx_stats.page_busy_count;
 
 		if (i40e_enabled_xdp_vsi(vsi)) {
 			/* locate XDP ring */

@@ -907,6 +909,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	vsi->rx_page_reuse = rx_reuse;
 	vsi->rx_page_alloc = rx_alloc;
 	vsi->rx_page_waive = rx_waive;
+	vsi->rx_page_busy = rx_busy;
 
 	ns->rx_packets = rx_p;
 	ns->rx_bytes = rx_b;

@@ -1990,8 +1990,8 @@ static bool i40e_cleanup_headers(struct i40e_ring *rx_ring, struct sk_buff *skb,
  * pointing to; otherwise, the DMA mapping needs to be destroyed and
  * page freed.
  *
- * rx_stats will be updated to indicate if the page was waived because it was
- * not reusable.
+ * rx_stats will be updated to indicate whether the page was waived
+ * or busy if it could not be reused.
  */
 static bool i40e_can_reuse_rx_page(struct i40e_rx_buffer *rx_buffer,
 				   struct i40e_rx_queue_stats *rx_stats,

@@ -2008,13 +2008,17 @@ static bool i40e_can_reuse_rx_page(struct i40e_rx_buffer *rx_buffer,
 
 #if (PAGE_SIZE < 8192)
 	/* if we are only owner of page we can reuse it */
-	if (unlikely((rx_buffer_pgcnt - pagecnt_bias) > 1))
+	if (unlikely((rx_buffer_pgcnt - pagecnt_bias) > 1)) {
+		rx_stats->page_busy_count++;
 		return false;
+	}
 #else
 #define I40E_LAST_OFFSET \
 	(SKB_WITH_OVERHEAD(PAGE_SIZE) - I40E_RXBUFFER_2048)
-	if (rx_buffer->page_offset > I40E_LAST_OFFSET)
+	if (rx_buffer->page_offset > I40E_LAST_OFFSET) {
+		rx_stats->page_busy_count++;
 		return false;
+	}
 #endif
 
 	/* If we have drained the page fragment pool we need to update

@@ -301,6 +301,7 @@ struct i40e_rx_queue_stats {
 	u64 realloc_count;
 	u64 page_alloc_count;
 	u64 page_waive_count;
+	u64 page_busy_count;
 };
 
 enum i40e_ring_state_t {

@@ -855,6 +855,7 @@ struct i40e_vsi {
 	u64 rx_page_failed;
 	u64 rx_page_reuse;
 	u64 rx_page_alloc;
+	u64 rx_page_waive;
 
 	/* These are containers of ring pointers, allocated at run-time */
 	struct i40e_ring **rx_rings;

@@ -297,6 +297,7 @@ static const struct i40e_stats i40e_gstrings_misc_stats[] = {
 	I40E_VSI_STAT("rx_pg_alloc_fail", rx_page_failed),
 	I40E_VSI_STAT("rx_cache_reuse", rx_page_reuse),
 	I40E_VSI_STAT("rx_cache_alloc", rx_page_alloc),
+	I40E_VSI_STAT("rx_cache_waive", rx_page_waive),
 };
 
 /* These PF_STATs might look like duplicates of some NETDEV_STATs,

@@ -812,7 +812,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	struct i40e_eth_stats *es;     /* device's eth stats */
 	u64 tx_restart, tx_busy;
 	struct i40e_ring *p;
-	u64 rx_page, rx_buf, rx_reuse, rx_alloc;
+	u64 rx_page, rx_buf, rx_reuse, rx_alloc, rx_waive;
 	u64 bytes, packets;
 	unsigned int start;
 	u64 tx_linearize;

@@ -840,6 +840,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	rx_buf = 0;
 	rx_reuse = 0;
 	rx_reuse = 0;
+	rx_waive = 0;
 	rcu_read_lock();
 	for (q = 0; q < vsi->num_queue_pairs; q++) {
 		/* locate Tx ring */

@@ -875,6 +876,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 		rx_page += p->rx_stats.alloc_page_failed;
 		rx_reuse += p->rx_stats.page_reuse_count;
 		rx_alloc += p->rx_stats.page_alloc_count;
+		rx_waive += p->rx_stats.page_waive_count;
 
 		if (i40e_enabled_xdp_vsi(vsi)) {
 			/* locate XDP ring */

@@ -904,6 +906,7 @@ static void i40e_update_vsi_stats(struct i40e_vsi *vsi)
 	vsi->rx_buf_failed = rx_buf;
 	vsi->rx_page_reuse = rx_reuse;
 	vsi->rx_page_alloc = rx_alloc;
+	vsi->rx_page_waive = rx_waive;
 
 	ns->rx_packets = rx_p;
 	ns->rx_bytes = rx_b;

@@ -1982,22 +1982,29 @@ static bool i40e_cleanup_headers(struct i40e_ring *rx_ring, struct sk_buff *skb,
 /**
  * i40e_can_reuse_rx_page - Determine if page can be reused for another Rx
  * @rx_buffer: buffer containing the page
+ * @rx_stats: rx stats structure for the rx ring
  * @rx_buffer_pgcnt: buffer page refcount pre xdp_do_redirect() call
  *
  * If page is reusable, we have a green light for calling i40e_reuse_rx_page,
  * which will assign the current buffer to the buffer that next_to_alloc is
  * pointing to; otherwise, the DMA mapping needs to be destroyed and
- * page freed
+ * page freed.
+ *
+ * rx_stats will be updated to indicate if the page was waived because it was
+ * not reusable.
  */
 static bool i40e_can_reuse_rx_page(struct i40e_rx_buffer *rx_buffer,
+				   struct i40e_rx_queue_stats *rx_stats,
 				   int rx_buffer_pgcnt)
 {
 	unsigned int pagecnt_bias = rx_buffer->pagecnt_bias;
 	struct page *page = rx_buffer->page;
 
 	/* Is any reuse possible? */
-	if (!dev_page_is_reusable(page))
+	if (!dev_page_is_reusable(page)) {
+		rx_stats->page_waive_count++;
 		return false;
+	}
 
 #if (PAGE_SIZE < 8192)
 	/* if we are only owner of page we can reuse it */

@@ -2237,7 +2244,7 @@ static void i40e_put_rx_buffer(struct i40e_ring *rx_ring,
 			       struct i40e_rx_buffer *rx_buffer,
 			       int rx_buffer_pgcnt)
 {
-	if (i40e_can_reuse_rx_page(rx_buffer, rx_buffer_pgcnt)) {
+	if (i40e_can_reuse_rx_page(rx_buffer, &rx_ring->rx_stats, rx_buffer_pgcnt)) {
 		/* hand second half of page back to the ring */
 		i40e_reuse_rx_page(rx_ring, rx_buffer);
 	} else {

@@ -300,6 +300,7 @@ struct i40e_rx_queue_stats {
 	u64 page_reuse_count;
 	u64 realloc_count;
 	u64 page_alloc_count;
+	u64 page_waive_count;
 };
 
 enum i40e_ring_state_t {