sfc: per-queue stats

From: Edward Cree <[email protected]>

The n_rx_tobe_disc and n_rx_mcast_mismatch counters are a legacy
from farch, and are never written in EF10 or EF100 code. Remove them from the struct and from ethtool -S output, saving a bit of memory and avoiding user confusion.

Signed-off-by Edward Cree <[email protected]>

---

From: Edward Cree <[email protected]>

Just RX and TX packet counts for now. We do not have per-queue
byte counts, which causes us to fail stats.pkt_byte_sum selftest with "Drivers should always report basic keys" error.

Per-queue counts are since the last time the queue was inited
(typically by efx_start_datapath(), on ifup or reconfiguration); device-wide total (efx_get_base_stats()) is since driver probe. This is not the same lifetime as rtnl_link_stats64, which uses firmware stats which count since FW (re)booted; this can cause a "Qstats are lower" or "RTNL stats are lower" failure in stats.pkt_byte_sum selftest.

Move the increment of rx_queue->rx_packets to match the semantics
specified for netdev per-queue stats, i.e. just before handing the packet to XDP (if present) or the netstack (through GRO). This will affect the existing ethtool -S output which also reports these counters.

Quoted 5 lines · On 8/28/2024 6:45 AM, [email protected] wrote:

From: Edward Cree <[email protected]>

Just RX and TX packet counts for now. We do not have per-queue
byte counts, which causes us to fail stats.pkt_byte_sum selftest with "Drivers should always report basic keys" error.

Seems like the self tests here should be fixed for this?

drivers/net/ethernet/sfc/efx.c static const struct net_device_ops efx_netdev_ops = { · line 659

Quoted 8 lines

drivers/net/ethernet/sfc/efx.c +8 -0

+	/* If a TX channel has XDP TXQs, the stats for these will be counted
+	 * under the channel rather than in base stats.  Unclear whether this
+	 * is correct behaviour, but we can't reliably exclude XDP TXes from
+	 * these stats anyway because in EFX_XDP_TX_QUEUES_BORROWED we use
+	 * the same TXQ as the core.
+	 */
+	efx_for_each_channel_tx_queue(tx_queue, channel)
+		stats->packets += tx_queue->tx_packets - tx_queue->old_tx_packets;

We haven't had to deal with shared host/XDP queues in the other drivers, yet. But talking to Olek about his stats work it sounds like he's planning to add support for reporting XDP queues. At which point it will be relatively intuitive - if XDP queues are listed - they count XDP packets, if not, and XDP_TX happens
- the queues must be shared and so are the counters.

IOW let's not count dedicated XDP queues here at all, if we can. XDP traffic on shared queues can get added in.

IOW let's not count dedicated XDP queues here at all, if we can.

Quoted 1 line · Jakub Kicinski

IOW let's not count dedicated XDP queues here at all, if we can.

But they should still go in base_stats, like other not-core
queues, right?

From: Edward Cree <[email protected]>

This counter is the main difference between the old and new locations
of the rx_packets increment (the other is scatter errors which produce a WARN_ON). It previously was not reported anywhere; add it to ethtool -S output to ensure users still have this information.

Signed-off-by Edward Cree <[email protected]>

---

 drivers/net/ethernet/sfc/ethtool_common.c | 1 +
 1 file changed, 1 insertion(+)

drivers/net/ethernet/sfc/ethtool_common.c +1 -0

index a8baeacd83c0..ae32e08540fa 100644
@@ -83,6 +83,7 @@ static const struct efx_sw_stat_desc efx_sw_stat_desc[] = {
 	EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_outer_tcp_udp_chksum_err),
 	EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_eth_crc_err),
 	EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_frm_trunc),
+	EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_overlength),
 	EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_merge_events),
 	EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_merge_packets),
 	EFX_ETHTOOL_UINT_CHANNEL_STAT(rx_xdp_drops),

This counter is the main difference between the old and new locations
of the rx_packets increment (the other is scatter errors which produce a WARN_ON). It previously was not reported anywhere; add it to ethtool -S output to ensure users still have this information.

Quoted 4 lines · On Wed, 28 Aug 2024 14:45:12 +0100 [email protected] wrote:

This counter is the main difference between the old and new locations
of the rx_packets increment (the other is scatter errors which produce a WARN_ON). It previously was not reported anywhere; add it to ethtool -S output to ensure users still have this information.

What is it tho? Not IEEE 802.3 30.3.1.1.25 aFrameTooLongErrors ?

The description makes sense in context with the whole series but doesn't quite work for me if I think about viewing it without context. Perhaps a little more clarification about the rx_packets behavioral change?

From: Edward Cree <[email protected]>

Signed-off-by Edward Cree <[email protected]>

---

 drivers/net/ethernet/sfc/efx.c          | 15 +++++++++++++--
 drivers/net/ethernet/sfc/efx_channels.c |  4 ++++
 drivers/net/ethernet/sfc/efx_channels.h |  7 +++++++
 drivers/net/ethernet/sfc/net_driver.h   |  7 +++++++
 4 files changed, 31 insertions(+), 2 deletions(-)

drivers/net/ethernet/sfc/efx.c +13 -2

index e4656efce969..8b46d143b6c7 100644
@@ -638,6 +638,10 @@ static void efx_get_queue_stats_rx(struct net_device *net_dev, int idx,
 	rx_queue = efx_channel_get_rx_queue(channel);
 	/* Count only packets since last time datapath was started */
 	stats->packets = rx_queue->rx_packets - rx_queue->old_rx_packets;
+	stats->hw_drops = efx_get_queue_stat_rx_hw_drops(channel) -
+			  channel->old_n_rx_hw_drops;
+	stats->hw_drop_overruns = channel->n_rx_nodesc_trunc -
+				  channel->old_n_rx_hw_drop_overruns;
 }
 
 static void efx_get_queue_stats_tx(struct net_device *net_dev, int idx,

@@ -669,6 +673,8 @@ static void efx_get_base_stats(struct net_device *net_dev,
 	struct efx_channel *channel;
 
 	rx->packets = 0;
+	rx->hw_drops = 0;
+	rx->hw_drop_overruns = 0;
 	tx->packets = 0;
 
 	/* Count all packets on non-core queues, and packets before last

@@ -676,10 +682,15 @@ static void efx_get_base_stats(struct net_device *net_dev,
 	 */
 	efx_for_each_channel(channel, efx) {
 		rx_queue = efx_channel_get_rx_queue(channel);
-		if (channel->channel >= net_dev->real_num_rx_queues)
+		if (channel->channel >= net_dev->real_num_rx_queues) {
 			rx->packets += rx_queue->rx_packets;
-		else
+			rx->hw_drops += efx_get_queue_stat_rx_hw_drops(channel);
+			rx->hw_drop_overruns += channel->n_rx_nodesc_trunc;
+		} else {
 			rx->packets += rx_queue->old_rx_packets;
+			rx->hw_drops += channel->old_n_rx_hw_drops;
+			rx->hw_drop_overruns += channel->old_n_rx_hw_drop_overruns;
+		}
 		efx_for_each_channel_tx_queue(tx_queue, channel) {
 			if (channel->channel < efx->tx_channel_offset ||
 			    channel->channel >= efx->tx_channel_offset +

drivers/net/ethernet/sfc/efx_channels.c +4 -0

index c9e17a8208a9..90b9986ceaa3 100644
@@ -1100,6 +1100,10 @@ void efx_start_channels(struct efx_nic *efx)
 			atomic_inc(&efx->active_queues);
 		}
 
+		/* reset per-queue stats */
+		channel->old_n_rx_hw_drops = efx_get_queue_stat_rx_hw_drops(channel);
+		channel->old_n_rx_hw_drop_overruns = channel->n_rx_nodesc_trunc;
+
 		efx_for_each_channel_rx_queue(rx_queue, channel) {
 			efx_init_rx_queue(rx_queue);
 			atomic_inc(&efx->active_queues);

drivers/net/ethernet/sfc/efx_channels.h +7 -0

index b3b5e18a69cc..cccbc7d66e77 100644
@@ -43,6 +43,13 @@ struct efx_channel *efx_copy_channel(const struct efx_channel *old_channel);
 void efx_start_channels(struct efx_nic *efx);
 void efx_stop_channels(struct efx_nic *efx);
 
+static inline u64 efx_get_queue_stat_rx_hw_drops(struct efx_channel *channel)
+{
+	return channel->n_rx_eth_crc_err + channel->n_rx_frm_trunc +
+	       channel->n_rx_overlength + channel->n_rx_nodesc_trunc +
+	       channel->n_rx_mport_bad;
+}
+
 void efx_init_napi_channel(struct efx_channel *channel);
 void efx_init_napi(struct efx_nic *efx);
 void efx_fini_napi_channel(struct efx_channel *channel);

drivers/net/ethernet/sfc/net_driver.h +7 -0

index cc96716d8dbe..25701f37aa40 100644
@@ -472,6 +472,10 @@ enum efx_sync_events_state {
  * @n_rx_xdp_redirect: Count of RX packets redirected to a different NIC by XDP
  * @n_rx_mport_bad: Count of RX packets dropped because their ingress mport was
  *	not recognised
+ * @old_n_rx_hw_drops: Count of all RX packets dropped for any reason as of last
+ *	efx_start_channels()
+ * @old_n_rx_hw_drop_overruns: Value of @n_rx_nodesc_trunc as of last
+ *	efx_start_channels()
  * @rx_pkt_n_frags: Number of fragments in next packet to be delivered by
  *	__efx_rx_packet(), or zero if there is none
  * @rx_pkt_index: Ring index of first buffer for next packet to be delivered

@@ -534,6 +538,9 @@ struct efx_channel {
 	unsigned int n_rx_xdp_redirect;
 	unsigned int n_rx_mport_bad;
 
+	unsigned int old_n_rx_hw_drops;
+	unsigned int old_n_rx_hw_drop_overruns;
+
 	unsigned int rx_pkt_n_frags;
 	unsigned int rx_pkt_index;
 

From: Edward Cree <[email protected]>

Signed-off-by Edward Cree <[email protected]>

---

 drivers/net/ethernet/sfc/efx.c        | 20 +++++++++++++++++---
 drivers/net/ethernet/sfc/net_driver.h |  4 ++++
 drivers/net/ethernet/sfc/tx_common.c  |  2 ++
 3 files changed, 23 insertions(+), 3 deletions(-)

drivers/net/ethernet/sfc/efx.c +17 -3

index 8b46d143b6c7..bf06fbcdcbff 100644
@@ -653,14 +653,21 @@ static void efx_get_queue_stats_tx(struct net_device *net_dev, int idx,
 
 	channel = efx_get_tx_channel(efx, idx);
 	stats->packets = 0;
+	stats->hw_gso_packets = 0;
+	stats->hw_gso_wire_packets = 0;
 	/* If a TX channel has XDP TXQs, the stats for these will be counted
 	 * under the channel rather than in base stats.  Unclear whether this
 	 * is correct behaviour, but we can't reliably exclude XDP TXes from
 	 * these stats anyway because in EFX_XDP_TX_QUEUES_BORROWED we use
 	 * the same TXQ as the core.
 	 */
-	efx_for_each_channel_tx_queue(tx_queue, channel)
+	efx_for_each_channel_tx_queue(tx_queue, channel) {
 		stats->packets += tx_queue->tx_packets - tx_queue->old_tx_packets;
+		stats->hw_gso_packets += tx_queue->tso_bursts -
+					 tx_queue->old_tso_bursts;
+		stats->hw_gso_wire_packets += tx_queue->tso_packets -
+					      tx_queue->old_tso_packets;
+	}
 }
 
 static void efx_get_base_stats(struct net_device *net_dev,

@@ -676,6 +683,8 @@ static void efx_get_base_stats(struct net_device *net_dev,
 	rx->hw_drops = 0;
 	rx->hw_drop_overruns = 0;
 	tx->packets = 0;
+	tx->hw_gso_packets = 0;
+	tx->hw_gso_wire_packets = 0;
 
 	/* Count all packets on non-core queues, and packets before last
 	 * datapath start on core queues.

@@ -694,10 +703,15 @@ static void efx_get_base_stats(struct net_device *net_dev,
 		efx_for_each_channel_tx_queue(tx_queue, channel) {
 			if (channel->channel < efx->tx_channel_offset ||
 			    channel->channel >= efx->tx_channel_offset +
-						net_dev->real_num_tx_queues)
+						net_dev->real_num_tx_queues) {
 				tx->packets += tx_queue->tx_packets;
-			else
+				tx->hw_gso_packets += tx_queue->tso_bursts;
+				tx->hw_gso_wire_packets += tx_queue->tso_packets;
+			} else {
 				tx->packets += tx_queue->old_tx_packets;
+				tx->hw_gso_packets += tx_queue->old_tso_bursts;
+				tx->hw_gso_wire_packets += tx_queue->old_tso_packets;
+			}
 		}
 	}
 }

drivers/net/ethernet/sfc/net_driver.h +4 -0

index 25701f37aa40..2cf2935a713c 100644
@@ -234,6 +234,8 @@ struct efx_tx_buffer {
  * @notify_count: Count of notified descriptors to the NIC
  * @tx_packets: Number of packets sent since this struct was created
  * @old_tx_packets: Value of @tx_packets as of last efx_init_tx_queue()
+ * @old_tso_bursts: Value of @tso_bursts as of last efx_init_tx_queue()
+ * @old_tso_packets: Value of @tso_packets as of last efx_init_tx_queue()
  * @empty_read_count: If the completion path has seen the queue as empty
  *	and the transmission path has not yet checked this, the value of
  *	@read_count bitwise-added to %EFX_EMPTY_COUNT_VALID; otherwise 0.

@@ -284,6 +286,8 @@ struct efx_tx_queue {
 	/* Statistics to supplement MAC stats */
 	unsigned long tx_packets;
 	unsigned long old_tx_packets;
+	unsigned int old_tso_bursts;
+	unsigned int old_tso_packets;
 
 	/* Members shared between paths and sometimes updated */
 	unsigned int empty_read_count ____cacheline_aligned_in_smp;

drivers/net/ethernet/sfc/tx_common.c +2 -0

index f1694900e0f0..cd0857131aa8 100644
@@ -87,6 +87,8 @@ void efx_init_tx_queue(struct efx_tx_queue *tx_queue)
 	tx_queue->completed_timestamp_minor = 0;
 
 	tx_queue->old_tx_packets = tx_queue->tx_packets;
+	tx_queue->old_tso_bursts = tx_queue->tso_bursts;
+	tx_queue->old_tso_packets = tx_queue->tso_packets;
 
 	tx_queue->xdp_tx = efx_channel_is_xdp_tx(tx_queue->channel);
 	tx_queue->tso_version = 0;

From: Edward Cree <[email protected]>

While this does add overhead to the fast path, it should be minimal
as the cacheline should already be held for write from updating the queue's [tr]x_packets stat.

Signed-off-by Edward Cree <[email protected]>

---

 drivers/net/ethernet/sfc/ef100_rx.c   |  1 +
 drivers/net/ethernet/sfc/ef100_tx.c   |  1 +
 drivers/net/ethernet/sfc/efx.c        |  9 +++++++++
 drivers/net/ethernet/sfc/net_driver.h | 10 ++++++++++
 drivers/net/ethernet/sfc/rx.c         |  1 +
 drivers/net/ethernet/sfc/rx_common.c  |  1 +
 drivers/net/ethernet/sfc/tx.c         |  2 ++
 drivers/net/ethernet/sfc/tx_common.c  |  1 +
 8 files changed, 26 insertions(+)

drivers/net/ethernet/sfc/ef100_rx.c +1 -0

index 992151775cb8..44dc75feb162 100644
@@ -135,6 +135,7 @@ void __ef100_rx_packet(struct efx_channel *channel)
 	}
 
 	++rx_queue->rx_packets;
+	rx_queue->rx_bytes += rx_buf->len;
 
 	efx_rx_packet_gro(channel, rx_buf, channel->rx_pkt_n_frags, eh, csum);
 	goto out;

drivers/net/ethernet/sfc/ef100_tx.c +1 -0

index e6b6be549581..a7e30289e231 100644
@@ -493,6 +493,7 @@ int __ef100_enqueue_skb(struct efx_tx_queue *tx_queue, struct sk_buff *skb,
 	} else {
 		tx_queue->tx_packets++;
 	}
+	tx_queue->tx_bytes += skb->len;
 	return 0;
 
 err:

drivers/net/ethernet/sfc/efx.c +9 -0

index bf06fbcdcbff..b3fbffed4e61 100644
@@ -638,6 +638,7 @@ static void efx_get_queue_stats_rx(struct net_device *net_dev, int idx,
 	rx_queue = efx_channel_get_rx_queue(channel);
 	/* Count only packets since last time datapath was started */
 	stats->packets = rx_queue->rx_packets - rx_queue->old_rx_packets;
+	stats->bytes = rx_queue->rx_bytes - rx_queue->old_rx_bytes;
 	stats->hw_drops = efx_get_queue_stat_rx_hw_drops(channel) -
 			  channel->old_n_rx_hw_drops;
 	stats->hw_drop_overruns = channel->n_rx_nodesc_trunc -

@@ -653,6 +654,7 @@ static void efx_get_queue_stats_tx(struct net_device *net_dev, int idx,
 
 	channel = efx_get_tx_channel(efx, idx);
 	stats->packets = 0;
+	stats->bytes = 0;
 	stats->hw_gso_packets = 0;
 	stats->hw_gso_wire_packets = 0;
 	/* If a TX channel has XDP TXQs, the stats for these will be counted

@@ -663,6 +665,7 @@ static void efx_get_queue_stats_tx(struct net_device *net_dev, int idx,
 	 */
 	efx_for_each_channel_tx_queue(tx_queue, channel) {
 		stats->packets += tx_queue->tx_packets - tx_queue->old_tx_packets;
+		stats->bytes += tx_queue->tx_bytes - tx_queue->old_tx_bytes;
 		stats->hw_gso_packets += tx_queue->tso_bursts -
 					 tx_queue->old_tso_bursts;
 		stats->hw_gso_wire_packets += tx_queue->tso_packets -

@@ -680,9 +683,11 @@ static void efx_get_base_stats(struct net_device *net_dev,
 	struct efx_channel *channel;
 
 	rx->packets = 0;
+	rx->bytes = 0;
 	rx->hw_drops = 0;
 	rx->hw_drop_overruns = 0;
 	tx->packets = 0;
+	tx->bytes = 0;
 	tx->hw_gso_packets = 0;
 	tx->hw_gso_wire_packets = 0;
 

@@ -693,10 +698,12 @@ static void efx_get_base_stats(struct net_device *net_dev,
 		rx_queue = efx_channel_get_rx_queue(channel);
 		if (channel->channel >= net_dev->real_num_rx_queues) {
 			rx->packets += rx_queue->rx_packets;
+			rx->bytes += rx_queue->rx_bytes;
 			rx->hw_drops += efx_get_queue_stat_rx_hw_drops(channel);
 			rx->hw_drop_overruns += channel->n_rx_nodesc_trunc;
 		} else {
 			rx->packets += rx_queue->old_rx_packets;
+			rx->bytes += rx_queue->old_rx_bytes;
 			rx->hw_drops += channel->old_n_rx_hw_drops;
 			rx->hw_drop_overruns += channel->old_n_rx_hw_drop_overruns;
 		}

@@ -705,10 +712,12 @@ static void efx_get_base_stats(struct net_device *net_dev,
 			    channel->channel >= efx->tx_channel_offset +
 						net_dev->real_num_tx_queues) {
 				tx->packets += tx_queue->tx_packets;
+				tx->bytes += tx_queue->tx_bytes;
 				tx->hw_gso_packets += tx_queue->tso_bursts;
 				tx->hw_gso_wire_packets += tx_queue->tso_packets;
 			} else {
 				tx->packets += tx_queue->old_tx_packets;
+				tx->bytes += tx_queue->old_tx_bytes;
 				tx->hw_gso_packets += tx_queue->old_tso_bursts;
 				tx->hw_gso_wire_packets += tx_queue->old_tso_packets;
 			}

drivers/net/ethernet/sfc/net_driver.h +10 -0

index 2cf2935a713c..147052c1e25a 100644
@@ -233,7 +233,11 @@ struct efx_tx_buffer {
  * @cb_packets: Number of times the TX copybreak feature has been used
  * @notify_count: Count of notified descriptors to the NIC
  * @tx_packets: Number of packets sent since this struct was created
+ * @tx_bytes: Number of bytes sent since this struct was created.  For TSO,
+ *	counts the superframe size, not the sizes of generated frames on the
+ *	wire (i.e. the headers are only counted once)
  * @old_tx_packets: Value of @tx_packets as of last efx_init_tx_queue()
+ * @old_tx_bytes: Value of @tx_bytes as of last efx_init_tx_queue()
  * @old_tso_bursts: Value of @tso_bursts as of last efx_init_tx_queue()
  * @old_tso_packets: Value of @tso_packets as of last efx_init_tx_queue()
  * @empty_read_count: If the completion path has seen the queue as empty

@@ -285,7 +289,9 @@ struct efx_tx_queue {
 	unsigned int notify_count;
 	/* Statistics to supplement MAC stats */
 	unsigned long tx_packets;
+	unsigned long tx_bytes;
 	unsigned long old_tx_packets;
+	unsigned long old_tx_bytes;
 	unsigned int old_tso_bursts;
 	unsigned int old_tso_packets;
 

@@ -378,7 +384,9 @@ struct efx_rx_page_state {
  * @slow_fill: Timer used to defer efx_nic_generate_fill_event().
  * @grant_work: workitem used to grant credits to the MAE if @grant_credits
  * @rx_packets: Number of packets received since this struct was created
+ * @rx_bytes: Number of bytes received since this struct was created
  * @old_rx_packets: Value of @rx_packets as of last efx_init_rx_queue()
+ * @old_rx_bytes: Value of @rx_bytes as of last efx_init_rx_queue()
  * @xdp_rxq_info: XDP specific RX queue information.
  * @xdp_rxq_info_valid: Is xdp_rxq_info valid data?.
  */

@@ -415,7 +423,9 @@ struct efx_rx_queue {
 	struct work_struct grant_work;
 	/* Statistics to supplement MAC stats */
 	unsigned long rx_packets;
+	unsigned long rx_bytes;
 	unsigned long old_rx_packets;
+	unsigned long old_rx_bytes;
 	struct xdp_rxq_info xdp_rxq_info;
 	bool xdp_rxq_info_valid;
 };

drivers/net/ethernet/sfc/rx.c +1 -0

index f07495582125..ffca82207e47 100644
@@ -393,6 +393,7 @@ void __efx_rx_packet(struct efx_channel *channel)
 	}
 
 	rx_queue->rx_packets++;
+	rx_queue->rx_bytes += rx_buf->len;
 
 	if (!efx_do_xdp(efx, channel, rx_buf, &eh))
 		goto out;

drivers/net/ethernet/sfc/rx_common.c +1 -0

index bdb4325a7c2c..ab358fe13e1d 100644
@@ -242,6 +242,7 @@ void efx_init_rx_queue(struct efx_rx_queue *rx_queue)
 	rx_queue->page_recycle_full = 0;
 
 	rx_queue->old_rx_packets = rx_queue->rx_packets;
+	rx_queue->old_rx_bytes = rx_queue->rx_bytes;
 
 	/* Initialise limit fields */
 	max_fill = efx->rxq_entries - EFX_RXD_HEAD_ROOM;

drivers/net/ethernet/sfc/tx.c +2 -0

index fe2d476028e7..1aea19488a56 100644
@@ -394,6 +394,7 @@ netdev_tx_t __efx_enqueue_skb(struct efx_tx_queue *tx_queue, struct sk_buff *skb
 	} else {
 		tx_queue->tx_packets++;
 	}
+	tx_queue->tx_bytes += skb_len;
 
 	return NETDEV_TX_OK;
 

@@ -490,6 +491,7 @@ int efx_xdp_tx_buffers(struct efx_nic *efx, int n, struct xdp_frame **xdpfs,
 		tx_buffer->dma_offset = 0;
 		tx_buffer->unmap_len = len;
 		tx_queue->tx_packets++;
+		tx_queue->tx_bytes += len;
 	}
 
 	/* Pass mapped frames to hardware. */

drivers/net/ethernet/sfc/tx_common.c +1 -0

index cd0857131aa8..7ef2baa3439a 100644
@@ -87,6 +87,7 @@ void efx_init_tx_queue(struct efx_tx_queue *tx_queue)
 	tx_queue->completed_timestamp_minor = 0;
 
 	tx_queue->old_tx_packets = tx_queue->tx_packets;
+	tx_queue->old_tx_bytes = tx_queue->tx_bytes;
 	tx_queue->old_tso_bursts = tx_queue->tso_bursts;
 	tx_queue->old_tso_packets = tx_queue->tso_packets;
 

Ah, the broken self tests from the early commit get fixed here. Ok.