s390/qeth: overhaul ethtool statistics

Accumulate per-TX queue statistics, and increase their size to 64 bit.
Don't bother with enabling/disabling the statistics, the overhead is
negligible.

Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

6 files changed, 256 insertions, 216 deletions

diff --git a/drivers/s390/net/qeth_core.h b/drivers/s390/net/qeth_core.h
index 5e08f112db13..c0c46be0b251 100644
--- a/drivers/s390/net/qeth_core.h
+++ b/drivers/s390/net/qeth_core.h
@@ -34,6 +34,8 @@
 #include <asm/ccwgroup.h>
 #include <asm/sysinfo.h>
 
+#include <uapi/linux/if_link.h>
+
 #include "qeth_core_mpc.h"
 
 /**
@@ -112,35 +114,6 @@ static inline u32 qeth_get_device_id(struct ccw_device *cdev)
 #define CCW_DEVID(cdev)		(qeth_get_device_id(cdev))
 #define CARD_DEVID(card)	(CCW_DEVID(CARD_RDEV(card)))
 
-/**
- * card stuff
- */
-struct qeth_perf_stats {
-	unsigned int bufs_rec;
-	unsigned int bufs_sent;
-	unsigned int buf_elements_sent;
-
-	unsigned int skbs_sent_pack;
-	unsigned int bufs_sent_pack;
-
-	unsigned int sc_dp_p;
-	unsigned int sc_p_dp;
-	unsigned int large_send_bytes;
-	unsigned int large_send_cnt;
-	unsigned int sg_skbs_sent;
-	/* initial values when measuring starts */
-	unsigned long initial_rx_packets;
-	unsigned long initial_tx_packets;
-	/* inbound scatter gather data */
-	unsigned int sg_skbs_rx;
-	unsigned int sg_frags_rx;
-	unsigned int sg_alloc_page_rx;
-	unsigned int tx_csum;
-	unsigned int tx_lin;
-	unsigned int tx_linfail;
-	unsigned int rx_csum;
-};
-
 /* Routing stuff */
 struct qeth_routing_info {
 	enum qeth_routing_types type;
@@ -470,10 +443,54 @@ enum qeth_out_q_states {
        QETH_OUT_Q_LOCKED_FLUSH,
 };
 
+#define QETH_CARD_STAT_ADD(_c, _stat, _val)	((_c)->stats._stat += (_val))
+#define QETH_CARD_STAT_INC(_c, _stat)		QETH_CARD_STAT_ADD(_c, _stat, 1)
+
+#define QETH_TXQ_STAT_ADD(_q, _stat, _val)	((_q)->stats._stat += (_val))
+#define QETH_TXQ_STAT_INC(_q, _stat)		QETH_TXQ_STAT_ADD(_q, _stat, 1)
+
+struct qeth_card_stats {
+	u64 rx_bufs;
+	u64 rx_skb_csum;
+	u64 rx_sg_skbs;
+	u64 rx_sg_frags;
+	u64 rx_sg_alloc_page;
+
+	/* rtnl_link_stats64 */
+	u64 rx_packets;
+	u64 rx_bytes;
+	u64 rx_errors;
+	u64 rx_dropped;
+	u64 rx_multicast;
+	u64 tx_errors;
+};
+
+struct qeth_out_q_stats {
+	u64 bufs;
+	u64 bufs_pack;
+	u64 buf_elements;
+	u64 skbs_pack;
+	u64 skbs_sg;
+	u64 skbs_csum;
+	u64 skbs_tso;
+	u64 skbs_linearized;
+	u64 skbs_linearized_fail;
+	u64 tso_bytes;
+	u64 packing_mode_switch;
+
+	/* rtnl_link_stats64 */
+	u64 tx_packets;
+	u64 tx_bytes;
+	u64 tx_errors;
+	u64 tx_dropped;
+	u64 tx_carrier_errors;
+};
+
 struct qeth_qdio_out_q {
 	struct qdio_buffer *qdio_bufs[QDIO_MAX_BUFFERS_PER_Q];
 	struct qeth_qdio_out_buffer *bufs[QDIO_MAX_BUFFERS_PER_Q];
 	struct qdio_outbuf_state *bufstates; /* convenience pointer */
+	struct qeth_out_q_stats stats;
 	int queue_no;
 	struct qeth_card *card;
 	atomic_t state;
@@ -677,7 +694,6 @@ struct qeth_card_options {
 	struct qeth_vnicc_info vnicc; /* VNICC options */
 	int fake_broadcast;
 	enum qeth_discipline_id layer;
-	int performance_stats;
 	int rx_sg_cb;
 	enum qeth_ipa_isolation_modes isolation;
 	enum qeth_ipa_isolation_modes prev_isolation;
@@ -753,8 +769,7 @@ struct qeth_card {
 	struct qeth_channel data;
 
 	struct net_device *dev;
-	struct net_device_stats stats;
-
+	struct qeth_card_stats stats;
 	struct qeth_card_info info;
 	struct qeth_token token;
 	struct qeth_seqno seqno;
@@ -777,7 +792,6 @@ struct qeth_card {
 	struct list_head cmd_waiter_list;
 	/* QDIO buffer handling */
 	struct qeth_qdio_info qdio;
-	struct qeth_perf_stats perf_stats;
 	int read_or_write_problem;
 	struct qeth_osn_info osn_info;
 	struct qeth_discipline *discipline;
@@ -858,8 +872,7 @@ static inline void qeth_rx_csum(struct qeth_card *card, struct sk_buff *skb,
 	if ((card->dev->features & NETIF_F_RXCSUM) &&
 	    (flags & QETH_HDR_EXT_CSUM_TRANSP_REQ)) {
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
-		if (card->options.performance_stats)
-			card->perf_stats.rx_csum++;
+		QETH_CARD_STAT_INC(card, rx_skb_csum);
 	} else {
 		skb->ip_summed = CHECKSUM_NONE;
 	}
@@ -966,7 +979,6 @@ void qeth_clear_working_pool_list(struct qeth_card *);
 void qeth_clear_cmd_buffers(struct qeth_channel *);
 void qeth_clear_qdio_buffers(struct qeth_card *);
 void qeth_setadp_promisc_mode(struct qeth_card *);
-struct net_device_stats *qeth_get_stats(struct net_device *);
 int qeth_setadpparms_change_macaddr(struct qeth_card *);
 void qeth_tx_timeout(struct net_device *);
 void qeth_prepare_control_data(struct qeth_card *, int,
@@ -1002,14 +1014,16 @@ netdev_features_t qeth_fix_features(struct net_device *, netdev_features_t);
 netdev_features_t qeth_features_check(struct sk_buff *skb,
 				      struct net_device *dev,
 				      netdev_features_t features);
+void qeth_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats);
 int qeth_open(struct net_device *dev);
 int qeth_stop(struct net_device *dev);
 
 int qeth_vm_request_mac(struct qeth_card *card);
 int qeth_xmit(struct qeth_card *card, struct sk_buff *skb,
 	      struct qeth_qdio_out_q *queue, int ipv, int cast_type,
-	      void (*fill_header)(struct qeth_card *card, struct qeth_hdr *hdr,
-				  struct sk_buff *skb, int ipv, int cast_type,
+	      void (*fill_header)(struct qeth_qdio_out_q *queue,
+				  struct qeth_hdr *hdr, struct sk_buff *skb,
+				  int ipv, int cast_type,
 				  unsigned int data_len));
 
 /* exports for OSN */
diff --git a/drivers/s390/net/qeth_core_main.c b/drivers/s390/net/qeth_core_main.c
index 96e4876a4daa..6d93bb48b1d9 100644
--- a/drivers/s390/net/qeth_core_main.c
+++ b/drivers/s390/net/qeth_core_main.c
@@ -2703,8 +2703,7 @@ static struct qeth_buffer_pool_entry *qeth_find_free_buffer_pool_entry(
 			} else {
 				free_page((unsigned long)entry->elements[i]);
 				entry->elements[i] = page_address(page);
-				if (card->options.performance_stats)
-					card->perf_stats.sg_alloc_page_rx++;
+				QETH_CARD_STAT_INC(card, rx_sg_alloc_page);
 			}
 		}
 	}
@@ -3180,7 +3179,7 @@ static int qeth_check_qdio_errors(struct qeth_card *card,
 			       buf->element[14].sflags);
 		QETH_CARD_TEXT_(card, 2, " qerr=%X", qdio_error);
 		if ((buf->element[15].sflags) == 0x12) {
-			card->stats.rx_dropped++;
+			QETH_CARD_STAT_INC(card, rx_dropped);
 			return 0;
 		} else
 			return 1;
@@ -3322,8 +3321,7 @@ static void qeth_switch_to_packing_if_needed(struct qeth_qdio_out_q *queue)
 		    >= QETH_HIGH_WATERMARK_PACK){
 			/* switch non-PACKING -> PACKING */
 			QETH_CARD_TEXT(queue->card, 6, "np->pack");
-			if (queue->card->options.performance_stats)
-				queue->card->perf_stats.sc_dp_p++;
+			QETH_TXQ_STAT_INC(queue, packing_mode_switch);
 			queue->do_pack = 1;
 		}
 	}
@@ -3342,8 +3340,7 @@ static int qeth_switch_to_nonpacking_if_needed(struct qeth_qdio_out_q *queue)
 		    <= QETH_LOW_WATERMARK_PACK) {
 			/* switch PACKING -> non-PACKING */
 			QETH_CARD_TEXT(queue->card, 6, "pack->np");
-			if (queue->card->options.performance_stats)
-				queue->card->perf_stats.sc_p_dp++;
+			QETH_TXQ_STAT_INC(queue, packing_mode_switch);
 			queue->do_pack = 0;
 			return qeth_prep_flush_pack_buffer(queue);
 		}
@@ -3397,6 +3394,7 @@ static void qeth_flush_buffers(struct qeth_qdio_out_q *queue, int index,
 		}
 	}
 
+	QETH_TXQ_STAT_ADD(queue, bufs, count);
 	netif_trans_update(queue->card->dev);
 	qdio_flags = QDIO_FLAG_SYNC_OUTPUT;
 	if (atomic_read(&queue->set_pci_flags_count))
@@ -3405,7 +3403,7 @@ static void qeth_flush_buffers(struct qeth_qdio_out_q *queue, int index,
 	rc = do_QDIO(CARD_DDEV(queue->card), qdio_flags,
 		     queue->queue_no, index, count);
 	if (rc) {
-		queue->card->stats.tx_errors += count;
+		QETH_TXQ_STAT_ADD(queue, tx_errors, count);
 		/* ignore temporary SIGA errors without busy condition */
 		if (rc == -ENOBUFS)
 			return;
@@ -3420,8 +3418,6 @@ static void qeth_flush_buffers(struct qeth_qdio_out_q *queue, int index,
 		qeth_schedule_recovery(queue->card);
 		return;
 	}
-	if (queue->card->options.performance_stats)
-		queue->card->perf_stats.bufs_sent += count;
 }
 
 static void qeth_check_outbound_queue(struct qeth_qdio_out_q *queue)
@@ -3452,10 +3448,8 @@ static void qeth_check_outbound_queue(struct qeth_qdio_out_q *queue)
 			if (!flush_cnt &&
 			    !atomic_read(&queue->set_pci_flags_count))
 				flush_cnt += qeth_prep_flush_pack_buffer(queue);
-			if (queue->card->options.performance_stats &&
-			    q_was_packing)
-				queue->card->perf_stats.bufs_sent_pack +=
-					flush_cnt;
+			if (q_was_packing)
+				QETH_TXQ_STAT_ADD(queue, bufs_pack, flush_cnt);
 			if (flush_cnt)
 				qeth_flush_buffers(queue, index, flush_cnt);
 			atomic_set(&queue->state, QETH_OUT_Q_UNLOCKED);
@@ -3748,11 +3742,12 @@ EXPORT_SYMBOL_GPL(qeth_count_elements);
  * The number of needed buffer elements is returned in @elements.
  * Error to create the hdr is indicated by returning with < 0.
  */
-static int qeth_add_hw_header(struct qeth_card *card, struct sk_buff *skb,
-			      struct qeth_hdr **hdr, unsigned int hdr_len,
-			      unsigned int proto_len, unsigned int *elements)
+static int qeth_add_hw_header(struct qeth_qdio_out_q *queue,
+			      struct sk_buff *skb, struct qeth_hdr **hdr,
+			      unsigned int hdr_len, unsigned int proto_len,
+			      unsigned int *elements)
 {
-	const unsigned int max_elements = QETH_MAX_BUFFER_ELEMENTS(card);
+	const unsigned int max_elements = QETH_MAX_BUFFER_ELEMENTS(queue->card);
 	const unsigned int contiguous = proto_len ? proto_len : 1;
 	unsigned int __elements;
 	addr_t start, end;
@@ -3791,15 +3786,12 @@ check_layout:
 		}
 
 		rc = skb_linearize(skb);
-		if (card->options.performance_stats) {
-			if (rc)
-				card->perf_stats.tx_linfail++;
-			else
-				card->perf_stats.tx_lin++;
-		}
-		if (rc)
+		if (rc) {
+			QETH_TXQ_STAT_INC(queue, skbs_linearized_fail);
 			return rc;
+		}
 
+		QETH_TXQ_STAT_INC(queue, skbs_linearized);
 		/* Linearization changed the layout, re-evaluate: */
 		goto check_layout;
 	}
@@ -3923,9 +3915,8 @@ static int qeth_fill_buffer(struct qeth_qdio_out_q *queue,
 		QETH_CARD_TEXT(queue->card, 6, "fillbfnp");
 	} else {
 		QETH_CARD_TEXT(queue->card, 6, "fillbfpa");
-		if (queue->card->options.performance_stats)
-			queue->card->perf_stats.skbs_sent_pack++;
 
+		QETH_TXQ_STAT_INC(queue, skbs_pack);
 		/* If the buffer still has free elements, keep using it. */
 		if (buf->next_element_to_fill <
 		    QETH_MAX_BUFFER_ELEMENTS(queue->card))
@@ -4039,8 +4030,8 @@ int qeth_do_send_packet(struct qeth_card *card, struct qeth_qdio_out_q *queue,
 	}
 out:
 	/* at this point the queue is UNLOCKED again */
-	if (queue->card->options.performance_stats && do_pack)
-		queue->card->perf_stats.bufs_sent_pack += flush_count;
+	if (do_pack)
+		QETH_TXQ_STAT_ADD(queue, bufs_pack, flush_count);
 
 	return rc;
 }
@@ -4064,8 +4055,9 @@ static void qeth_fill_tso_ext(struct qeth_hdr_tso *hdr,
 
 int qeth_xmit(struct qeth_card *card, struct sk_buff *skb,
 	      struct qeth_qdio_out_q *queue, int ipv, int cast_type,
-	      void (*fill_header)(struct qeth_card *card, struct qeth_hdr *hdr,
-				  struct sk_buff *skb, int ipv, int cast_type,
+	      void (*fill_header)(struct qeth_qdio_out_q *queue,
+				  struct qeth_hdr *hdr, struct sk_buff *skb,
+				  int ipv, int cast_type,
 				  unsigned int data_len))
 {
 	unsigned int proto_len, hw_hdr_len;
@@ -4090,7 +4082,7 @@ int qeth_xmit(struct qeth_card *card, struct sk_buff *skb,
 	if (rc)
 		return rc;
 
-	push_len = qeth_add_hw_header(card, skb, &hdr, hw_hdr_len, proto_len,
+	push_len = qeth_add_hw_header(queue, skb, &hdr, hw_hdr_len, proto_len,
 				      &elements);
 	if (push_len < 0)
 		return push_len;
@@ -4100,7 +4092,7 @@ int qeth_xmit(struct qeth_card *card, struct sk_buff *skb,
 		data_offset = push_len + proto_len;
 	}
 	memset(hdr, 0, hw_hdr_len);
-	fill_header(card, hdr, skb, ipv, cast_type, frame_len);
+	fill_header(queue, hdr, skb, ipv, cast_type, frame_len);
 	if (is_tso)
 		qeth_fill_tso_ext((struct qeth_hdr_tso *) hdr,
 				  frame_len - proto_len, skb, proto_len);
@@ -4117,14 +4109,12 @@ int qeth_xmit(struct qeth_card *card, struct sk_buff *skb,
 	}
 
 	if (!rc) {
-		if (card->options.performance_stats) {
-			card->perf_stats.buf_elements_sent += elements;
-			if (is_sg)
-				card->perf_stats.sg_skbs_sent++;
-			if (is_tso) {
-				card->perf_stats.large_send_bytes += frame_len;
-				card->perf_stats.large_send_cnt++;
-			}
+		QETH_TXQ_STAT_ADD(queue, buf_elements, elements);
+		if (is_sg)
+			QETH_TXQ_STAT_INC(queue, skbs_sg);
+		if (is_tso) {
+			QETH_TXQ_STAT_INC(queue, skbs_tso);
+			QETH_TXQ_STAT_ADD(queue, tso_bytes, frame_len);
 		}
 	} else {
 		if (!push_len)
@@ -4183,18 +4173,6 @@ void qeth_setadp_promisc_mode(struct qeth_card *card)
 }
 EXPORT_SYMBOL_GPL(qeth_setadp_promisc_mode);
 
-struct net_device_stats *qeth_get_stats(struct net_device *dev)
-{
-	struct qeth_card *card;
-
-	card = dev->ml_priv;
-
-	QETH_CARD_TEXT(card, 5, "getstat");
-
-	return &card->stats;
-}
-EXPORT_SYMBOL_GPL(qeth_get_stats);
-
 static int qeth_setadpparms_change_macaddr_cb(struct qeth_card *card,
 		struct qeth_reply *reply, unsigned long data)
 {
@@ -4388,7 +4366,7 @@ void qeth_tx_timeout(struct net_device *dev)
 
 	card = dev->ml_priv;
 	QETH_CARD_TEXT(card, 4, "txtimeo");
-	card->stats.tx_errors++;
+	QETH_CARD_STAT_INC(card, tx_errors);
 	qeth_schedule_recovery(card);
 }
 EXPORT_SYMBOL_GPL(qeth_tx_timeout);
@@ -5246,7 +5224,7 @@ struct sk_buff *qeth_core_get_next_skb(struct qeth_card *card,
 				QETH_CARD_TEXT(card, 4, "unexeob");
 				QETH_CARD_HEX(card, 2, buffer, sizeof(void *));
 				dev_kfree_skb_any(skb);
-				card->stats.rx_errors++;
+				QETH_CARD_STAT_INC(card, rx_errors);
 				return NULL;
 			}
 			element++;
@@ -5258,16 +5236,17 @@ struct sk_buff *qeth_core_get_next_skb(struct qeth_card *card,
 	}
 	*__element = element;
 	*__offset = offset;
-	if (use_rx_sg && card->options.performance_stats) {
-		card->perf_stats.sg_skbs_rx++;
-		card->perf_stats.sg_frags_rx += skb_shinfo(skb)->nr_frags;
+	if (use_rx_sg) {
+		QETH_CARD_STAT_INC(card, rx_sg_skbs);
+		QETH_CARD_STAT_ADD(card, rx_sg_frags,
+				   skb_shinfo(skb)->nr_frags);
 	}
 	return skb;
 no_mem:
 	if (net_ratelimit()) {
 		QETH_CARD_TEXT(card, 2, "noskbmem");
 	}
-	card->stats.rx_dropped++;
+	QETH_CARD_STAT_INC(card, rx_dropped);
 	return NULL;
 }
 EXPORT_SYMBOL_GPL(qeth_core_get_next_skb);
@@ -5308,8 +5287,7 @@ int qeth_poll(struct napi_struct *napi, int budget)
 				done = 1;
 
 			if (done) {
-				if (card->options.performance_stats)
-					card->perf_stats.bufs_rec++;
+				QETH_CARD_STAT_INC(card, rx_bufs);
 				qeth_put_buffer_pool_entry(card,
 					buffer->pool_entry);
 				qeth_queue_input_buffer(card, card->rx.b_index);
@@ -6223,6 +6201,33 @@ netdev_features_t qeth_features_check(struct sk_buff *skb,
 }
 EXPORT_SYMBOL_GPL(qeth_features_check);
 
+void qeth_get_stats64(struct net_device *dev, struct rtnl_link_stats64 *stats)
+{
+	struct qeth_card *card = dev->ml_priv;
+	struct qeth_qdio_out_q *queue;
+	unsigned int i;
+
+	QETH_CARD_TEXT(card, 5, "getstat");
+
+	stats->rx_packets = card->stats.rx_packets;
+	stats->rx_bytes = card->stats.rx_bytes;
+	stats->rx_errors = card->stats.rx_errors;
+	stats->rx_dropped = card->stats.rx_dropped;
+	stats->multicast = card->stats.rx_multicast;
+	stats->tx_errors = card->stats.tx_errors;
+
+	for (i = 0; i < card->qdio.no_out_queues; i++) {
+		queue = card->qdio.out_qs[i];
+
+		stats->tx_packets += queue->stats.tx_packets;
+		stats->tx_bytes += queue->stats.tx_bytes;
+		stats->tx_errors += queue->stats.tx_errors;
+		stats->tx_dropped += queue->stats.tx_dropped;
+		stats->tx_carrier_errors += queue->stats.tx_carrier_errors;
+	}
+}
+EXPORT_SYMBOL_GPL(qeth_get_stats64);
+
 int qeth_open(struct net_device *dev)
 {
 	struct qeth_card *card = dev->ml_priv;
diff --git a/drivers/s390/net/qeth_core_sys.c b/drivers/s390/net/qeth_core_sys.c
index 485a335669ba..8b223cc2c19b 100644
--- a/drivers/s390/net/qeth_core_sys.c
+++ b/drivers/s390/net/qeth_core_sys.c
@@ -336,35 +336,36 @@ static ssize_t qeth_dev_performance_stats_show(struct device *dev,
 	if (!card)
 		return -EINVAL;
 
-	return sprintf(buf, "%i\n", card->options.performance_stats ? 1:0);
+	return sprintf(buf, "1\n");
 }
 
 static ssize_t qeth_dev_performance_stats_store(struct device *dev,
 		struct device_attribute *attr, const char *buf, size_t count)
 {
 	struct qeth_card *card = dev_get_drvdata(dev);
-	char *tmp;
-	int i, rc = 0;
+	struct qeth_qdio_out_q *queue;
+	unsigned int i;
+	bool reset;
+	int rc;
 
 	if (!card)
 		return -EINVAL;
 
-	mutex_lock(&card->conf_mutex);
-	i = simple_strtoul(buf, &tmp, 16);
-	if ((i == 0) || (i == 1)) {
-		if (i == card->options.performance_stats)
-			goto out;
-		card->options.performance_stats = i;
-		if (i == 0)
-			memset(&card->perf_stats, 0,
-				sizeof(struct qeth_perf_stats));
-		card->perf_stats.initial_rx_packets = card->stats.rx_packets;
-		card->perf_stats.initial_tx_packets = card->stats.tx_packets;
-	} else
-		rc = -EINVAL;
-out:
-	mutex_unlock(&card->conf_mutex);
-	return rc ? rc : count;
+	rc = kstrtobool(buf, &reset);
+	if (rc)
+		return rc;
+
+	if (reset) {
+		memset(&card->stats, 0, sizeof(card->stats));
+		for (i = 0; i < card->qdio.no_out_queues; i++) {
+			queue = card->qdio.out_qs[i];
+			if (!queue)
+				break;
+			memset(&queue->stats, 0, sizeof(queue->stats));
+		}
+	}
+
+	return count;
 }
 
 static DEVICE_ATTR(performance_stats, 0644, qeth_dev_performance_stats_show,
diff --git a/drivers/s390/net/qeth_ethtool.c b/drivers/s390/net/qeth_ethtool.c
index a275f31150b2..9ce5ef5fcf7a 100644
--- a/drivers/s390/net/qeth_ethtool.c
+++ b/drivers/s390/net/qeth_ethtool.c
@@ -9,43 +9,82 @@
 #include <linux/ethtool.h>
 #include "qeth_core.h"
 
-static struct {
-	const char str[ETH_GSTRING_LEN];
-} qeth_ethtool_stats_keys[] = {
-/*  0 */{"rx skbs"},
-	{"rx buffers"},
-	{"tx skbs"},
-	{"tx buffers"},
-	{"tx skbs no packing"},
-	{"tx buffers no packing"},
-	{"tx skbs packing"},
-	{"tx buffers packing"},
-	{"tx sg skbs"},
-	{"tx buffer elements"},
-/* 10 */{"rx sg skbs"},
-	{"rx sg frags"},
-	{"rx sg page allocs"},
-	{"tx large kbytes"},
-	{"tx large count"},
-	{"tx pk state ch n->p"},
-	{"tx pk state ch p->n"},
-	{"tx pk watermark low"},
-	{"tx pk watermark high"},
-	{"queue 0 buffer usage"},
-/* 20 */{"queue 1 buffer usage"},
-	{"queue 2 buffer usage"},
-	{"queue 3 buffer usage"},
-	{"tx csum"},
-	{"tx lin"},
-	{"tx linfail"},
-	{"rx csum"}
+
+#define QETH_TXQ_STAT(_name, _stat) { \
+	.name = _name, \
+	.offset = offsetof(struct qeth_out_q_stats, _stat) \
+}
+
+#define QETH_CARD_STAT(_name, _stat) { \
+	.name = _name, \
+	.offset = offsetof(struct qeth_card_stats, _stat) \
+}
+
+struct qeth_stats {
+	char name[ETH_GSTRING_LEN];
+	unsigned int offset;
+};
+
+static const struct qeth_stats txq_stats[] = {
+	QETH_TXQ_STAT("IO buffers", bufs),
+	QETH_TXQ_STAT("IO buffer elements", buf_elements),
+	QETH_TXQ_STAT("packed IO buffers", bufs_pack),
+	QETH_TXQ_STAT("skbs", tx_packets),
+	QETH_TXQ_STAT("packed skbs", skbs_pack),
+	QETH_TXQ_STAT("SG skbs", skbs_sg),
+	QETH_TXQ_STAT("HW csum skbs", skbs_csum),
+	QETH_TXQ_STAT("TSO skbs", skbs_tso),
+	QETH_TXQ_STAT("linearized skbs", skbs_linearized),
+	QETH_TXQ_STAT("linearized+error skbs", skbs_linearized_fail),
+	QETH_TXQ_STAT("TSO bytes", tso_bytes),
+	QETH_TXQ_STAT("Packing mode switches", packing_mode_switch),
+};
+
+static const struct qeth_stats card_stats[] = {
+	QETH_CARD_STAT("rx0 IO buffers", rx_bufs),
+	QETH_CARD_STAT("rx0 HW csum skbs", rx_skb_csum),
+	QETH_CARD_STAT("rx0 SG skbs", rx_sg_skbs),
+	QETH_CARD_STAT("rx0 SG page frags", rx_sg_frags),
+	QETH_CARD_STAT("rx0 SG page allocs", rx_sg_alloc_page),
 };
 
+#define TXQ_STATS_LEN	ARRAY_SIZE(txq_stats)
+#define CARD_STATS_LEN	ARRAY_SIZE(card_stats)
+
+static void qeth_add_stat_data(u64 **dst, void *src,
+			       const struct qeth_stats stats[],
+			       unsigned int size)
+{
+	unsigned int i;
+	char *stat;
+
+	for (i = 0; i < size; i++) {
+		stat = (char *)src + stats[i].offset;
+		**dst = *(u64 *)stat;
+		(*dst)++;
+	}
+}
+
+static void qeth_add_stat_strings(u8 **data, const char *prefix,
+				  const struct qeth_stats stats[],
+				  unsigned int size)
+{
+	unsigned int i;
+
+	for (i = 0; i < size; i++) {
+		snprintf(*data, ETH_GSTRING_LEN, "%s%s", prefix, stats[i].name);
+		*data += ETH_GSTRING_LEN;
+	}
+}
+
 static int qeth_get_sset_count(struct net_device *dev, int stringset)
 {
+	struct qeth_card *card = dev->ml_priv;
+
 	switch (stringset) {
 	case ETH_SS_STATS:
-		return (sizeof(qeth_ethtool_stats_keys) / ETH_GSTRING_LEN);
+		return CARD_STATS_LEN +
+		       card->qdio.no_out_queues * TXQ_STATS_LEN;
 	default:
 		return -EINVAL;
 	}
@@ -55,48 +94,29 @@ static void qeth_get_ethtool_stats(struct net_device *dev,
 				   struct ethtool_stats *stats, u64 *data)
 {
 	struct qeth_card *card = dev->ml_priv;
+	unsigned int i;
 
-	data[0] = card->stats.rx_packets -
-				card->perf_stats.initial_rx_packets;
-	data[1] = card->perf_stats.bufs_rec;
-	data[2] = card->stats.tx_packets -
-				card->perf_stats.initial_tx_packets;
-	data[3] = card->perf_stats.bufs_sent;
-	data[4] = card->stats.tx_packets - card->perf_stats.initial_tx_packets
-			- card->perf_stats.skbs_sent_pack;
-	data[5] = card->perf_stats.bufs_sent - card->perf_stats.bufs_sent_pack;
-	data[6] = card->perf_stats.skbs_sent_pack;
-	data[7] = card->perf_stats.bufs_sent_pack;
-	data[8] = card->perf_stats.sg_skbs_sent;
-	data[9] = card->perf_stats.buf_elements_sent;
-	data[10] = card->perf_stats.sg_skbs_rx;
-	data[11] = card->perf_stats.sg_frags_rx;
-	data[12] = card->perf_stats.sg_alloc_page_rx;
-	data[13] = (card->perf_stats.large_send_bytes >> 10);
-	data[14] = card->perf_stats.large_send_cnt;
-	data[15] = card->perf_stats.sc_dp_p;
-	data[16] = card->perf_stats.sc_p_dp;
-	data[17] = QETH_LOW_WATERMARK_PACK;
-	data[18] = QETH_HIGH_WATERMARK_PACK;
-	data[19] = atomic_read(&card->qdio.out_qs[0]->used_buffers);
-	data[20] = (card->qdio.no_out_queues > 1) ?
-			atomic_read(&card->qdio.out_qs[1]->used_buffers) : 0;
-	data[21] = (card->qdio.no_out_queues > 2) ?
-			atomic_read(&card->qdio.out_qs[2]->used_buffers) : 0;
-	data[22] = (card->qdio.no_out_queues > 3) ?
-			atomic_read(&card->qdio.out_qs[3]->used_buffers) : 0;
-	data[23] = card->perf_stats.tx_csum;
-	data[24] = card->perf_stats.tx_lin;
-	data[25] = card->perf_stats.tx_linfail;
-	data[26] = card->perf_stats.rx_csum;
+	qeth_add_stat_data(&data, &card->stats, card_stats, CARD_STATS_LEN);
+	for (i = 0; i < card->qdio.no_out_queues; i++)
+		qeth_add_stat_data(&data, &card->qdio.out_qs[i]->stats,
+				   txq_stats, TXQ_STATS_LEN);
 }
 
 static void qeth_get_strings(struct net_device *dev, u32 stringset, u8 *data)
 {
+	struct qeth_card *card = dev->ml_priv;
+	char prefix[ETH_GSTRING_LEN] = "";
+	unsigned int i;
+
 	switch (stringset) {
 	case ETH_SS_STATS:
-		memcpy(data, &qeth_ethtool_stats_keys,
-			sizeof(qeth_ethtool_stats_keys));
+		qeth_add_stat_strings(&data, prefix, card_stats,
+				      CARD_STATS_LEN);
+		for (i = 0; i < card->qdio.no_out_queues; i++) {
+			snprintf(prefix, ETH_GSTRING_LEN, "tx%u ", i);
+			qeth_add_stat_strings(&data, prefix, txq_stats,
+					      TXQ_STATS_LEN);
+		}
 		break;
 	default:
 		WARN_ON(1);
diff --git a/drivers/s390/net/qeth_l2_main.c b/drivers/s390/net/qeth_l2_main.c
index dbcba77fd94c..b1280a155953 100644
--- a/drivers/s390/net/qeth_l2_main.c
+++ b/drivers/s390/net/qeth_l2_main.c
@@ -174,9 +174,9 @@ static int qeth_l2_get_cast_type(struct qeth_card *card, struct sk_buff *skb)
 	return RTN_UNICAST;
 }
 
-static void qeth_l2_fill_header(struct qeth_card *card, struct qeth_hdr *hdr,
-				struct sk_buff *skb, int ipv, int cast_type,
-				unsigned int data_len)
+static void qeth_l2_fill_header(struct qeth_qdio_out_q *queue,
+				struct qeth_hdr *hdr, struct sk_buff *skb,
+				int ipv, int cast_type, unsigned int data_len)
 {
 	struct vlan_ethhdr *veth = vlan_eth_hdr(skb);
 
@@ -188,8 +188,7 @@ static void qeth_l2_fill_header(struct qeth_card *card, struct qeth_hdr *hdr,
 		hdr->hdr.l2.id = QETH_HEADER_TYPE_LAYER2;
 		if (skb->ip_summed == CHECKSUM_PARTIAL) {
 			qeth_tx_csum(skb, &hdr->hdr.l2.flags[1], ipv);
-			if (card->options.performance_stats)
-				card->perf_stats.tx_csum++;
+			QETH_TXQ_STAT_INC(queue, skbs_csum);
 		}
 	}
 
@@ -369,8 +368,8 @@ static int qeth_l2_process_inbound_buffer(struct qeth_card *card,
 		}
 		work_done++;
 		budget--;
-		card->stats.rx_packets++;
-		card->stats.rx_bytes += len;
+		QETH_CARD_STAT_INC(card, rx_packets);
+		QETH_CARD_STAT_ADD(card, rx_bytes, len);
 	}
 	return work_done;
 }
@@ -626,12 +625,13 @@ static netdev_tx_t qeth_l2_hard_start_xmit(struct sk_buff *skb,
 	int tx_bytes = skb->len;
 	int rc;
 
+	queue = qeth_get_tx_queue(card, skb, ipv, cast_type);
+
 	if (card->state != CARD_STATE_UP) {
-		card->stats.tx_carrier_errors++;
+		QETH_TXQ_STAT_INC(queue, tx_carrier_errors);
 		goto tx_drop;
 	}
 
-	queue = qeth_get_tx_queue(card, skb, ipv, cast_type);
 	netif_stop_queue(dev);
 
 	if (IS_OSN(card))
@@ -641,8 +641,8 @@ static netdev_tx_t qeth_l2_hard_start_xmit(struct sk_buff *skb,
 			       qeth_l2_fill_header);
 
 	if (!rc) {
-		card->stats.tx_packets++;
-		card->stats.tx_bytes += tx_bytes;
+		QETH_TXQ_STAT_INC(queue, tx_packets);
+		QETH_TXQ_STAT_ADD(queue, tx_bytes, tx_bytes);
 		netif_wake_queue(dev);
 		return NETDEV_TX_OK;
 	} else if (rc == -EBUSY) {
@@ -650,8 +650,8 @@ static netdev_tx_t qeth_l2_hard_start_xmit(struct sk_buff *skb,
 	} /* else fall through */
 
 tx_drop:
-	card->stats.tx_dropped++;
-	card->stats.tx_errors++;
+	QETH_TXQ_STAT_INC(queue, tx_dropped);
+	QETH_TXQ_STAT_INC(queue, tx_errors);
 	dev_kfree_skb_any(skb);
 	netif_wake_queue(dev);
 	return NETDEV_TX_OK;
@@ -699,7 +699,7 @@ static void qeth_l2_remove_device(struct ccwgroup_device *cgdev)
 static const struct net_device_ops qeth_l2_netdev_ops = {
 	.ndo_open		= qeth_open,
 	.ndo_stop		= qeth_stop,
-	.ndo_get_stats		= qeth_get_stats,
+	.ndo_get_stats64	= qeth_get_stats64,
 	.ndo_start_xmit		= qeth_l2_hard_start_xmit,
 	.ndo_features_check	= qeth_features_check,
 	.ndo_validate_addr	= qeth_l2_validate_addr,
diff --git a/drivers/s390/net/qeth_l3_main.c b/drivers/s390/net/qeth_l3_main.c
index 40c7d53f5512..07c3149e228c 100644
--- a/drivers/s390/net/qeth_l3_main.c
+++ b/drivers/s390/net/qeth_l3_main.c
@@ -1315,12 +1315,11 @@ static void qeth_l3_rebuild_skb(struct qeth_card *card, struct sk_buff *skb,
 				ip_eth_mc_map(ip_hdr(skb)->daddr, tg_addr);
 			else
 				ipv6_eth_mc_map(&ipv6_hdr(skb)->daddr, tg_addr);
-
-			card->stats.multicast++;
+			QETH_CARD_STAT_INC(card, rx_multicast);
 			break;
 		case QETH_CAST_BROADCAST:
 			ether_addr_copy(tg_addr, card->dev->broadcast);
-			card->stats.multicast++;
+			QETH_CARD_STAT_INC(card, rx_multicast);
 			break;
 		default:
 			if (card->options.sniffer)
@@ -1401,8 +1400,8 @@ static int qeth_l3_process_inbound_buffer(struct qeth_card *card,
 		}
 		work_done++;
 		budget--;
-		card->stats.rx_packets++;
-		card->stats.rx_bytes += len;
+		QETH_CARD_STAT_INC(card, rx_packets);
+		QETH_CARD_STAT_ADD(card, rx_bytes, len);
 	}
 	return work_done;
 }
@@ -1945,12 +1944,13 @@ static u8 qeth_l3_cast_type_to_flag(int cast_type)
 	return QETH_CAST_UNICAST;
 }
 
-static void qeth_l3_fill_header(struct qeth_card *card, struct qeth_hdr *hdr,
-				struct sk_buff *skb, int ipv, int cast_type,
-				unsigned int data_len)
+static void qeth_l3_fill_header(struct qeth_qdio_out_q *queue,
+				struct qeth_hdr *hdr, struct sk_buff *skb,
+				int ipv, int cast_type, unsigned int data_len)
 {
 	struct qeth_hdr_layer3 *l3_hdr = &hdr->hdr.l3;
 	struct vlan_ethhdr *veth = vlan_eth_hdr(skb);
+	struct qeth_card *card = queue->card;
 
 	hdr->hdr.l3.length = data_len;
 
@@ -1972,8 +1972,7 @@ static void qeth_l3_fill_header(struct qeth_card *card, struct qeth_hdr *hdr,
 			/* some HW requires combined L3+L4 csum offload: */
 			if (ipv == 4)
 				hdr->hdr.l3.ext_flags |= QETH_HDR_EXT_CSUM_HDR_REQ;
-			if (card->options.performance_stats)
-				card->perf_stats.tx_csum++;
+			QETH_TXQ_STAT_INC(queue, skbs_csum);
 		}
 	}
 
@@ -2074,6 +2073,8 @@ static netdev_tx_t qeth_l3_hard_start_xmit(struct sk_buff *skb,
 	int tx_bytes = skb->len;
 	int rc;
 
+	queue = qeth_get_tx_queue(card, skb, ipv, cast_type);
+
 	if (IS_IQD(card)) {
 		if (card->options.sniffer)
 			goto tx_drop;
@@ -2084,14 +2085,13 @@ static netdev_tx_t qeth_l3_hard_start_xmit(struct sk_buff *skb,
 	}
 
 	if (card->state != CARD_STATE_UP) {
-		card->stats.tx_carrier_errors++;
+		QETH_TXQ_STAT_INC(queue, tx_carrier_errors);
 		goto tx_drop;
 	}
 
 	if (cast_type == RTN_BROADCAST && !card->info.broadcast_capable)
 		goto tx_drop;
 
-	queue = qeth_get_tx_queue(card, skb, ipv, cast_type);
 	netif_stop_queue(dev);
 
 	if (ipv == 4 || IS_IQD(card))
@@ -2101,8 +2101,8 @@ static netdev_tx_t qeth_l3_hard_start_xmit(struct sk_buff *skb,
 			       qeth_l3_fill_header);
 
 	if (!rc) {
-		card->stats.tx_packets++;
-		card->stats.tx_bytes += tx_bytes;
+		QETH_TXQ_STAT_INC(queue, tx_packets);
+		QETH_TXQ_STAT_ADD(queue, tx_bytes, tx_bytes);
 		netif_wake_queue(dev);
 		return NETDEV_TX_OK;
 	} else if (rc == -EBUSY) {
@@ -2110,8 +2110,8 @@ static netdev_tx_t qeth_l3_hard_start_xmit(struct sk_buff *skb,
 	} /* else fall through */
 
 tx_drop:
-	card->stats.tx_dropped++;
-	card->stats.tx_errors++;
+	QETH_TXQ_STAT_INC(queue, tx_dropped);
+	QETH_TXQ_STAT_INC(queue, tx_errors);
 	dev_kfree_skb_any(skb);
 	netif_wake_queue(dev);
 	return NETDEV_TX_OK;
@@ -2153,7 +2153,7 @@ static netdev_features_t qeth_l3_osa_features_check(struct sk_buff *skb,
 static const struct net_device_ops qeth_l3_netdev_ops = {
 	.ndo_open		= qeth_open,
 	.ndo_stop		= qeth_stop,
-	.ndo_get_stats		= qeth_get_stats,
+	.ndo_get_stats64	= qeth_get_stats64,
 	.ndo_start_xmit		= qeth_l3_hard_start_xmit,
 	.ndo_validate_addr	= eth_validate_addr,
 	.ndo_set_rx_mode	= qeth_l3_set_rx_mode,
@@ -2168,7 +2168,7 @@ static const struct net_device_ops qeth_l3_netdev_ops = {
 static const struct net_device_ops qeth_l3_osa_netdev_ops = {
 	.ndo_open		= qeth_open,
 	.ndo_stop		= qeth_stop,
-	.ndo_get_stats		= qeth_get_stats,
+	.ndo_get_stats64	= qeth_get_stats64,
 	.ndo_start_xmit		= qeth_l3_hard_start_xmit,
 	.ndo_features_check	= qeth_l3_osa_features_check,
 	.ndo_validate_addr	= eth_validate_addr,