net: stmmac: protect updates of 64-bit statistics counters

As explained by a comment in <linux/u64_stats_sync.h>, write side of struct
u64_stats_sync must ensure mutual exclusion, or one seqcount update could
be lost on 32-bit platforms, thus blocking readers forever. Such lockups
have been observed in real world after stmmac_xmit() on one CPU raced with
stmmac_napi_poll_tx() on another CPU.

To fix the issue without introducing a new lock, split the statics into
three parts:

1. fields updated only under the tx queue lock,
2. fields updated only during NAPI poll,
3. fields updated only from interrupt context,

Updates to fields in the first two groups are already serialized through
other locks. It is sufficient to split the existing struct u64_stats_sync
so that each group has its own.

Note that tx_set_ic_bit is updated from both contexts. Split this counter
so that each context gets its own, and calculate their sum to get the total
value in stmmac_get_ethtool_stats().

For the third group, multiple interrupts may be processed by different CPUs
at the same time, but interrupts on the same CPU will not nest. Move fields
from this group to a newly created per-cpu struct stmmac_pcpu_stats.

Fixes: 133466c3bbe1 ("net: stmmac: use per-queue 64 bit statistics where necessary")
Link: https://lore.kernel.org/netdev/Za173PhviYg-1qIn@torres.zugschlus.de/t/
Cc: stable@vger.kernel.org
Signed-off-by: Petr Tesarik <petr@tesarici.cz>
Reviewed-by: Jisheng Zhang <jszhang@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 66 insertions, 67 deletions

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index 25519952f754..75d029704503 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -2482,7 +2482,6 @@ static bool stmmac_xdp_xmit_zc(struct stmmac_priv *priv, u32 queue, u32 budget)
 	struct xdp_desc xdp_desc;
 	bool work_done = true;
 	u32 tx_set_ic_bit = 0;
-	unsigned long flags;
 
 	/* Avoids TX time-out as we are sharing with slow path */
 	txq_trans_cond_update(nq);
@@ -2566,9 +2565,9 @@ static bool stmmac_xdp_xmit_zc(struct stmmac_priv *priv, u32 queue, u32 budget)
 		tx_q->cur_tx = STMMAC_GET_ENTRY(tx_q->cur_tx, priv->dma_conf.dma_tx_size);
 		entry = tx_q->cur_tx;
 	}
-	flags = u64_stats_update_begin_irqsave(&txq_stats->syncp);
-	txq_stats->tx_set_ic_bit += tx_set_ic_bit;
-	u64_stats_update_end_irqrestore(&txq_stats->syncp, flags);
+	u64_stats_update_begin(&txq_stats->napi_syncp);
+	u64_stats_add(&txq_stats->napi.tx_set_ic_bit, tx_set_ic_bit);
+	u64_stats_update_end(&txq_stats->napi_syncp);
 
 	if (tx_desc) {
 		stmmac_flush_tx_descriptors(priv, queue);
@@ -2616,7 +2615,6 @@ static int stmmac_tx_clean(struct stmmac_priv *priv, int budget, u32 queue,
 	unsigned int bytes_compl = 0, pkts_compl = 0;
 	unsigned int entry, xmits = 0, count = 0;
 	u32 tx_packets = 0, tx_errors = 0;
-	unsigned long flags;
 
 	__netif_tx_lock_bh(netdev_get_tx_queue(priv->dev, queue));
 
@@ -2782,11 +2780,11 @@ static int stmmac_tx_clean(struct stmmac_priv *priv, int budget, u32 queue,
 	if (tx_q->dirty_tx != tx_q->cur_tx)
 		*pending_packets = true;
 
-	flags = u64_stats_update_begin_irqsave(&txq_stats->syncp);
-	txq_stats->tx_packets += tx_packets;
-	txq_stats->tx_pkt_n += tx_packets;
-	txq_stats->tx_clean++;
-	u64_stats_update_end_irqrestore(&txq_stats->syncp, flags);
+	u64_stats_update_begin(&txq_stats->napi_syncp);
+	u64_stats_add(&txq_stats->napi.tx_packets, tx_packets);
+	u64_stats_add(&txq_stats->napi.tx_pkt_n, tx_packets);
+	u64_stats_inc(&txq_stats->napi.tx_clean);
+	u64_stats_update_end(&txq_stats->napi_syncp);
 
 	priv->xstats.tx_errors += tx_errors;
 
@@ -4213,7 +4211,6 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev)
 	struct stmmac_tx_queue *tx_q;
 	bool has_vlan, set_ic;
 	u8 proto_hdr_len, hdr;
-	unsigned long flags;
 	u32 pay_len, mss;
 	dma_addr_t des;
 	int i;
@@ -4378,13 +4375,13 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev)
 		netif_tx_stop_queue(netdev_get_tx_queue(priv->dev, queue));
 	}
 
-	flags = u64_stats_update_begin_irqsave(&txq_stats->syncp);
-	txq_stats->tx_bytes += skb->len;
-	txq_stats->tx_tso_frames++;
-	txq_stats->tx_tso_nfrags += nfrags;
+	u64_stats_update_begin(&txq_stats->q_syncp);
+	u64_stats_add(&txq_stats->q.tx_bytes, skb->len);
+	u64_stats_inc(&txq_stats->q.tx_tso_frames);
+	u64_stats_add(&txq_stats->q.tx_tso_nfrags, nfrags);
 	if (set_ic)
-		txq_stats->tx_set_ic_bit++;
-	u64_stats_update_end_irqrestore(&txq_stats->syncp, flags);
+		u64_stats_inc(&txq_stats->q.tx_set_ic_bit);
+	u64_stats_update_end(&txq_stats->q_syncp);
 
 	if (priv->sarc_type)
 		stmmac_set_desc_sarc(priv, first, priv->sarc_type);
@@ -4483,7 +4480,6 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
 	struct stmmac_tx_queue *tx_q;
 	bool has_vlan, set_ic;
 	int entry, first_tx;
-	unsigned long flags;
 	dma_addr_t des;
 
 	tx_q = &priv->dma_conf.tx_queue[queue];
@@ -4653,11 +4649,11 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
 		netif_tx_stop_queue(netdev_get_tx_queue(priv->dev, queue));
 	}
 
-	flags = u64_stats_update_begin_irqsave(&txq_stats->syncp);
-	txq_stats->tx_bytes += skb->len;
+	u64_stats_update_begin(&txq_stats->q_syncp);
+	u64_stats_add(&txq_stats->q.tx_bytes, skb->len);
 	if (set_ic)
-		txq_stats->tx_set_ic_bit++;
-	u64_stats_update_end_irqrestore(&txq_stats->syncp, flags);
+		u64_stats_inc(&txq_stats->q.tx_set_ic_bit);
+	u64_stats_update_end(&txq_stats->q_syncp);
 
 	if (priv->sarc_type)
 		stmmac_set_desc_sarc(priv, first, priv->sarc_type);
@@ -4921,12 +4917,11 @@ static int stmmac_xdp_xmit_xdpf(struct stmmac_priv *priv, int queue,
 		set_ic = false;
 
 	if (set_ic) {
-		unsigned long flags;
 		tx_q->tx_count_frames = 0;
 		stmmac_set_tx_ic(priv, tx_desc);
-		flags = u64_stats_update_begin_irqsave(&txq_stats->syncp);
-		txq_stats->tx_set_ic_bit++;
-		u64_stats_update_end_irqrestore(&txq_stats->syncp, flags);
+		u64_stats_update_begin(&txq_stats->q_syncp);
+		u64_stats_inc(&txq_stats->q.tx_set_ic_bit);
+		u64_stats_update_end(&txq_stats->q_syncp);
 	}
 
 	stmmac_enable_dma_transmission(priv, priv->ioaddr);
@@ -5076,7 +5071,6 @@ static void stmmac_dispatch_skb_zc(struct stmmac_priv *priv, u32 queue,
 	unsigned int len = xdp->data_end - xdp->data;
 	enum pkt_hash_types hash_type;
 	int coe = priv->hw->rx_csum;
-	unsigned long flags;
 	struct sk_buff *skb;
 	u32 hash;
 
@@ -5106,10 +5100,10 @@ static void stmmac_dispatch_skb_zc(struct stmmac_priv *priv, u32 queue,
 	skb_record_rx_queue(skb, queue);
 	napi_gro_receive(&ch->rxtx_napi, skb);
 
-	flags = u64_stats_update_begin_irqsave(&rxq_stats->syncp);
-	rxq_stats->rx_pkt_n++;
-	rxq_stats->rx_bytes += len;
-	u64_stats_update_end_irqrestore(&rxq_stats->syncp, flags);
+	u64_stats_update_begin(&rxq_stats->napi_syncp);
+	u64_stats_inc(&rxq_stats->napi.rx_pkt_n);
+	u64_stats_add(&rxq_stats->napi.rx_bytes, len);
+	u64_stats_update_end(&rxq_stats->napi_syncp);
 }
 
 static bool stmmac_rx_refill_zc(struct stmmac_priv *priv, u32 queue, u32 budget)
@@ -5191,7 +5185,6 @@ static int stmmac_rx_zc(struct stmmac_priv *priv, int limit, u32 queue)
 	unsigned int desc_size;
 	struct bpf_prog *prog;
 	bool failure = false;
-	unsigned long flags;
 	int xdp_status = 0;
 	int status = 0;
 
@@ -5346,9 +5339,9 @@ read_again:
 
 	stmmac_finalize_xdp_rx(priv, xdp_status);
 
-	flags = u64_stats_update_begin_irqsave(&rxq_stats->syncp);
-	rxq_stats->rx_pkt_n += count;
-	u64_stats_update_end_irqrestore(&rxq_stats->syncp, flags);
+	u64_stats_update_begin(&rxq_stats->napi_syncp);
+	u64_stats_add(&rxq_stats->napi.rx_pkt_n, count);
+	u64_stats_update_end(&rxq_stats->napi_syncp);
 
 	priv->xstats.rx_dropped += rx_dropped;
 	priv->xstats.rx_errors += rx_errors;
@@ -5386,7 +5379,6 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
 	unsigned int desc_size;
 	struct sk_buff *skb = NULL;
 	struct stmmac_xdp_buff ctx;
-	unsigned long flags;
 	int xdp_status = 0;
 	int buf_sz;
 
@@ -5646,11 +5638,11 @@ drain_data:
 
 	stmmac_rx_refill(priv, queue);
 
-	flags = u64_stats_update_begin_irqsave(&rxq_stats->syncp);
-	rxq_stats->rx_packets += rx_packets;
-	rxq_stats->rx_bytes += rx_bytes;
-	rxq_stats->rx_pkt_n += count;
-	u64_stats_update_end_irqrestore(&rxq_stats->syncp, flags);
+	u64_stats_update_begin(&rxq_stats->napi_syncp);
+	u64_stats_add(&rxq_stats->napi.rx_packets, rx_packets);
+	u64_stats_add(&rxq_stats->napi.rx_bytes, rx_bytes);
+	u64_stats_add(&rxq_stats->napi.rx_pkt_n, count);
+	u64_stats_update_end(&rxq_stats->napi_syncp);
 
 	priv->xstats.rx_dropped += rx_dropped;
 	priv->xstats.rx_errors += rx_errors;
@@ -5665,13 +5657,12 @@ static int stmmac_napi_poll_rx(struct napi_struct *napi, int budget)
 	struct stmmac_priv *priv = ch->priv_data;
 	struct stmmac_rxq_stats *rxq_stats;
 	u32 chan = ch->index;
-	unsigned long flags;
 	int work_done;
 
 	rxq_stats = &priv->xstats.rxq_stats[chan];
-	flags = u64_stats_update_begin_irqsave(&rxq_stats->syncp);
-	rxq_stats->napi_poll++;
-	u64_stats_update_end_irqrestore(&rxq_stats->syncp, flags);
+	u64_stats_update_begin(&rxq_stats->napi_syncp);
+	u64_stats_inc(&rxq_stats->napi.poll);
+	u64_stats_update_end(&rxq_stats->napi_syncp);
 
 	work_done = stmmac_rx(priv, budget, chan);
 	if (work_done < budget && napi_complete_done(napi, work_done)) {
@@ -5693,13 +5684,12 @@ static int stmmac_napi_poll_tx(struct napi_struct *napi, int budget)
 	struct stmmac_txq_stats *txq_stats;
 	bool pending_packets = false;
 	u32 chan = ch->index;
-	unsigned long flags;
 	int work_done;
 
 	txq_stats = &priv->xstats.txq_stats[chan];
-	flags = u64_stats_update_begin_irqsave(&txq_stats->syncp);
-	txq_stats->napi_poll++;
-	u64_stats_update_end_irqrestore(&txq_stats->syncp, flags);
+	u64_stats_update_begin(&txq_stats->napi_syncp);
+	u64_stats_inc(&txq_stats->napi.poll);
+	u64_stats_update_end(&txq_stats->napi_syncp);
 
 	work_done = stmmac_tx_clean(priv, budget, chan, &pending_packets);
 	work_done = min(work_done, budget);
@@ -5729,17 +5719,16 @@ static int stmmac_napi_poll_rxtx(struct napi_struct *napi, int budget)
 	struct stmmac_rxq_stats *rxq_stats;
 	struct stmmac_txq_stats *txq_stats;
 	u32 chan = ch->index;
-	unsigned long flags;
 
 	rxq_stats = &priv->xstats.rxq_stats[chan];
-	flags = u64_stats_update_begin_irqsave(&rxq_stats->syncp);
-	rxq_stats->napi_poll++;
-	u64_stats_update_end_irqrestore(&rxq_stats->syncp, flags);
+	u64_stats_update_begin(&rxq_stats->napi_syncp);
+	u64_stats_inc(&rxq_stats->napi.poll);
+	u64_stats_update_end(&rxq_stats->napi_syncp);
 
 	txq_stats = &priv->xstats.txq_stats[chan];
-	flags = u64_stats_update_begin_irqsave(&txq_stats->syncp);
-	txq_stats->napi_poll++;
-	u64_stats_update_end_irqrestore(&txq_stats->syncp, flags);
+	u64_stats_update_begin(&txq_stats->napi_syncp);
+	u64_stats_inc(&txq_stats->napi.poll);
+	u64_stats_update_end(&txq_stats->napi_syncp);
 
 	tx_done = stmmac_tx_clean(priv, budget, chan, &tx_pending_packets);
 	tx_done = min(tx_done, budget);
@@ -7065,10 +7054,13 @@ static void stmmac_get_stats64(struct net_device *dev, struct rtnl_link_stats64
 		u64 tx_bytes;
 
 		do {
-			start = u64_stats_fetch_begin(&txq_stats->syncp);
-			tx_packets = txq_stats->tx_packets;
-			tx_bytes   = txq_stats->tx_bytes;
-		} while (u64_stats_fetch_retry(&txq_stats->syncp, start));
+			start = u64_stats_fetch_begin(&txq_stats->q_syncp);
+			tx_bytes   = u64_stats_read(&txq_stats->q.tx_bytes);
+		} while (u64_stats_fetch_retry(&txq_stats->q_syncp, start));
+		do {
+			start = u64_stats_fetch_begin(&txq_stats->napi_syncp);
+			tx_packets = u64_stats_read(&txq_stats->napi.tx_packets);
+		} while (u64_stats_fetch_retry(&txq_stats->napi_syncp, start));
 
 		stats->tx_packets += tx_packets;
 		stats->tx_bytes += tx_bytes;
@@ -7080,10 +7072,10 @@ static void stmmac_get_stats64(struct net_device *dev, struct rtnl_link_stats64
 		u64 rx_bytes;
 
 		do {
-			start = u64_stats_fetch_begin(&rxq_stats->syncp);
-			rx_packets = rxq_stats->rx_packets;
-			rx_bytes   = rxq_stats->rx_bytes;
-		} while (u64_stats_fetch_retry(&rxq_stats->syncp, start));
+			start = u64_stats_fetch_begin(&rxq_stats->napi_syncp);
+			rx_packets = u64_stats_read(&rxq_stats->napi.rx_packets);
+			rx_bytes   = u64_stats_read(&rxq_stats->napi.rx_bytes);
+		} while (u64_stats_fetch_retry(&rxq_stats->napi_syncp, start));
 
 		stats->rx_packets += rx_packets;
 		stats->rx_bytes += rx_bytes;
@@ -7477,9 +7469,16 @@ int stmmac_dvr_probe(struct device *device,
 	priv->dev = ndev;
 
 	for (i = 0; i < MTL_MAX_RX_QUEUES; i++)
-		u64_stats_init(&priv->xstats.rxq_stats[i].syncp);
-	for (i = 0; i < MTL_MAX_TX_QUEUES; i++)
-		u64_stats_init(&priv->xstats.txq_stats[i].syncp);
+		u64_stats_init(&priv->xstats.rxq_stats[i].napi_syncp);
+	for (i = 0; i < MTL_MAX_TX_QUEUES; i++) {
+		u64_stats_init(&priv->xstats.txq_stats[i].q_syncp);
+		u64_stats_init(&priv->xstats.txq_stats[i].napi_syncp);
+	}
+
+	priv->xstats.pcpu_stats =
+		devm_netdev_alloc_pcpu_stats(device, struct stmmac_pcpu_stats);
+	if (!priv->xstats.pcpu_stats)
+		return -ENOMEM;
 
 	stmmac_set_ethtool_ops(ndev);
 	priv->pause = pause;