sfc: Split MAC stats DMA initiation and completion

From: Steve Hodgson <shodgson@solarflare.com>

Currently we initiate MAC stats DMA and busy-wait for completion when
stats are requested.  We can improve on this with a periodic timer to
initiate and poll for stats, and opportunistically poll when stats are
requested.

Since efx_nic::stats_disable_count and efx_stats_{disable,enable}()
are Falcon-specific, rename them and move them accordingly.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/sfc/efx.c

@@ -637,6 +637,7 @@ void __efx_reconfigure_port(struct efx_nic *efx)
 		netif_addr_unlock_bh(efx->net_dev);
 	}
 
+	falcon_stop_nic_stats(efx);
 	falcon_deconfigure_mac_wrapper(efx);
 
 	/* Reconfigure the PHY, disabling transmit in mac level loopback. */
@@ -651,6 +652,8 @@ void __efx_reconfigure_port(struct efx_nic *efx)
 
 	efx->mac_op->reconfigure(efx);
 
+	falcon_start_nic_stats(efx);
+
 	/* Inform kernel of loss/gain of carrier */
 	efx_link_status_changed(efx);
 	return;
@@ -749,7 +752,6 @@ static int efx_init_port(struct efx_nic *efx)
 	efx->mac_op->reconfigure(efx);
 
 	efx->port_initialized = true;
-	efx_stats_enable(efx);
 
 	mutex_unlock(&efx->mac_lock);
 	return 0;
@@ -802,7 +804,6 @@ static void efx_fini_port(struct efx_nic *efx)
 	if (!efx->port_initialized)
 		return;
 
-	efx_stats_disable(efx);
 	efx->phy_op->fini(efx);
 	efx->port_initialized = false;
 
@@ -1158,6 +1159,8 @@ static void efx_start_all(struct efx_nic *efx)
 	if (efx->state == STATE_RUNNING)
 		queue_delayed_work(efx->workqueue, &efx->monitor_work,
 				   efx_monitor_interval);
+
+	falcon_start_nic_stats(efx);
 }
 
 /* Flush all delayed work. Should only be called when no more delayed work
@@ -1195,6 +1198,8 @@ static void efx_stop_all(struct efx_nic *efx)
 	if (!efx->port_enabled)
 		return;
 
+	falcon_stop_nic_stats(efx);
+
 	/* Disable interrupts and wait for ISR to complete */
 	falcon_disable_interrupts(efx);
 	if (efx->legacy_irq)
@@ -1438,20 +1443,6 @@ static int efx_net_stop(struct net_device *net_dev)
 	return 0;
 }
 
-void efx_stats_disable(struct efx_nic *efx)
-{
-	spin_lock(&efx->stats_lock);
-	++efx->stats_disable_count;
-	spin_unlock(&efx->stats_lock);
-}
-
-void efx_stats_enable(struct efx_nic *efx)
-{
-	spin_lock(&efx->stats_lock);
-	--efx->stats_disable_count;
-	spin_unlock(&efx->stats_lock);
-}
-
 /* Context: process, dev_base_lock or RTNL held, non-blocking. */
 static struct net_device_stats *efx_net_stats(struct net_device *net_dev)
 {
@@ -1459,17 +1450,9 @@ static struct net_device_stats *efx_net_stats(struct net_device *net_dev)
 	struct efx_mac_stats *mac_stats = &efx->mac_stats;
 	struct net_device_stats *stats = &net_dev->stats;
 
-	/* Update stats if possible, but do not wait if another thread
-	 * is updating them or if MAC stats fetches are temporarily
-	 * disabled; slightly stale stats are acceptable.
-	 */
-	if (!spin_trylock(&efx->stats_lock))
-		return stats;
-	if (!efx->stats_disable_count) {
-		efx->mac_op->update_stats(efx);
-		falcon_update_nic_stats(efx);
-	}
-	spin_unlock(&efx->stats_lock);
+	spin_lock_bh(&efx->stats_lock);
+	falcon_update_nic_stats(efx);
+	spin_unlock_bh(&efx->stats_lock);
 
 	stats->rx_packets = mac_stats->rx_packets;
 	stats->tx_packets = mac_stats->tx_packets;
@@ -1726,7 +1709,6 @@ void efx_reset_down(struct efx_nic *efx, enum reset_type method,
 {
 	EFX_ASSERT_RESET_SERIALISED(efx);
 
-	efx_stats_disable(efx);
 	efx_stop_all(efx);
 	mutex_lock(&efx->mac_lock);
 	mutex_lock(&efx->spi_lock);
@@ -1776,10 +1758,8 @@ int efx_reset_up(struct efx_nic *efx, enum reset_type method,
 	mutex_unlock(&efx->spi_lock);
 	mutex_unlock(&efx->mac_lock);
 
-	if (ok) {
+	if (ok)
 		efx_start_all(efx);
-		efx_stats_enable(efx);
-	}
 	return rc;
 }
 
@@ -1977,7 +1957,6 @@ static int efx_init_struct(struct efx_nic *efx, struct efx_nic_type *type,
 	efx->rx_checksum_enabled = true;
 	spin_lock_init(&efx->netif_stop_lock);
 	spin_lock_init(&efx->stats_lock);
-	efx->stats_disable_count = 1;
 	mutex_init(&efx->mac_lock);
 	efx->mac_op = &efx_dummy_mac_operations;
 	efx->phy_op = &efx_dummy_phy_operations;
@@ -2219,9 +2198,8 @@ static int __devinit efx_pci_probe(struct pci_dev *pci_dev,
 		goto fail4;
 	}
 
-	/* Switch to the running state before we expose the device to
-	 * the OS.  This is to ensure that the initial gathering of
-	 * MAC stats succeeds. */
+	/* Switch to the running state before we expose the device to the OS,
+	 * so that dev_open()|efx_start_all() will actually start the device */
 	efx->state = STATE_RUNNING;
 
 	rc = efx_register_netdev(efx);

drivers/net/sfc/efx.h

@@ -60,8 +60,6 @@ extern void efx_process_channel_now(struct efx_channel *channel);
 #define EFX_EVQ_MASK (EFX_EVQ_SIZE - 1)
 
 /* Ports */
-extern void efx_stats_disable(struct efx_nic *efx);
-extern void efx_stats_enable(struct efx_nic *efx);
 extern void efx_reconfigure_port(struct efx_nic *efx);
 extern void __efx_reconfigure_port(struct efx_nic *efx);
 

drivers/net/sfc/falcon.c

@@ -36,8 +36,6 @@
  **************************************************************************
  */
 
-static int disable_dma_stats;
-
 /* This is set to 16 for a good reason.  In summary, if larger than
  * 16, the descriptor cache holds more than a default socket
  * buffer's worth of packets (for UDP we can only have at most one
@@ -1890,7 +1888,7 @@ static int falcon_reset_macs(struct efx_nic *efx)
 
 	/* MAC stats will fail whilst the TX fifo is draining. Serialise
 	 * the drain sequence with the statistics fetch */
-	efx_stats_disable(efx);
+	falcon_stop_nic_stats(efx);
 
 	efx_reado(efx, &reg, FR_AB_MAC_CTRL);
 	EFX_SET_OWORD_FIELD(reg, FRF_BB_TXFIFO_DRAIN_EN, 1);
@@ -1920,13 +1918,13 @@ static int falcon_reset_macs(struct efx_nic *efx)
 		udelay(10);
 	}
 
-	efx_stats_enable(efx);
-
 	/* If we've reset the EM block and the link is up, then
 	 * we'll have to kick the XAUI link so the PHY can recover */
 	if (efx->link_state.up && EFX_IS10G(efx) && EFX_WORKAROUND_5147(efx))
 		falcon_reset_xaui(efx);
 
+	falcon_start_nic_stats(efx);
+
 	return 0;
 }
 
@@ -2010,25 +2008,19 @@ void falcon_reconfigure_mac_wrapper(struct efx_nic *efx)
 	efx_writeo(efx, &reg, FR_AZ_RX_CFG);
 }
 
-int falcon_dma_stats(struct efx_nic *efx, unsigned int done_offset)
+static void falcon_stats_request(struct efx_nic *efx)
 {
+	struct falcon_nic_data *nic_data = efx->nic_data;
 	efx_oword_t reg;
-	u32 *dma_done;
-	int i;
 
-	if (disable_dma_stats)
-		return 0;
+	WARN_ON(nic_data->stats_pending);
+	WARN_ON(nic_data->stats_disable_count);
 
-	/* Statistics fetch will fail if the MAC is in TX drain */
-	if (falcon_rev(efx) >= FALCON_REV_B0) {
-		efx_oword_t temp;
-		efx_reado(efx, &temp, FR_AB_MAC_CTRL);
-		if (EFX_OWORD_FIELD(temp, FRF_BB_TXFIFO_DRAIN_EN))
-			return 0;
-	}
+	if (nic_data->stats_dma_done == NULL)
+		return;	/* no mac selected */
 
-	dma_done = (efx->stats_buffer.addr + done_offset);
-	*dma_done = FALCON_STATS_NOT_DONE;
+	*nic_data->stats_dma_done = FALCON_STATS_NOT_DONE;
+	nic_data->stats_pending = true;
 	wmb(); /* ensure done flag is clear */
 
 	/* Initiate DMA transfer of stats */
@@ -2038,17 +2030,37 @@ int falcon_dma_stats(struct efx_nic *efx, unsigned int done_offset)
 			     efx->stats_buffer.dma_addr);
 	efx_writeo(efx, &reg, FR_AB_MAC_STAT_DMA);
 
-	/* Wait for transfer to complete */
-	for (i = 0; i < 400; i++) {
-		if (*(volatile u32 *)dma_done == FALCON_STATS_DONE) {
-			rmb(); /* Ensure the stats are valid. */
-			return 0;
-		}
-		udelay(10);
+	mod_timer(&nic_data->stats_timer, round_jiffies_up(jiffies + HZ / 2));
+}
+
+static void falcon_stats_complete(struct efx_nic *efx)
+{
+	struct falcon_nic_data *nic_data = efx->nic_data;
+
+	if (!nic_data->stats_pending)
+		return;
+
+	nic_data->stats_pending = 0;
+	if (*nic_data->stats_dma_done == FALCON_STATS_DONE) {
+		rmb(); /* read the done flag before the stats */
+		efx->mac_op->update_stats(efx);
+	} else {
+		EFX_ERR(efx, "timed out waiting for statistics\n");
 	}
+}
 
-	EFX_ERR(efx, "timed out waiting for statistics\n");
-	return -ETIMEDOUT;
+static void falcon_stats_timer_func(unsigned long context)
+{
+	struct efx_nic *efx = (struct efx_nic *)context;
+	struct falcon_nic_data *nic_data = efx->nic_data;
+
+	spin_lock(&efx->stats_lock);
+
+	falcon_stats_complete(efx);
+	if (nic_data->stats_disable_count == 0)
+		falcon_stats_request(efx);
+
+	spin_unlock(&efx->stats_lock);
 }
 
 /**************************************************************************
@@ -2206,10 +2218,12 @@ static void falcon_clock_mac(struct efx_nic *efx)
 int falcon_switch_mac(struct efx_nic *efx)
 {
 	struct efx_mac_operations *old_mac_op = efx->mac_op;
+	struct falcon_nic_data *nic_data = efx->nic_data;
+	unsigned int stats_done_offset;
 	int rc = 0;
 
 	/* Don't try to fetch MAC stats while we're switching MACs */
-	efx_stats_disable(efx);
+	falcon_stop_nic_stats(efx);
 
 	/* Internal loopbacks override the phy speed setting */
 	if (efx->loopback_mode == LOOPBACK_GMAC) {
@@ -2224,6 +2238,12 @@ int falcon_switch_mac(struct efx_nic *efx)
 	efx->mac_op = (EFX_IS10G(efx) ?
 		       &falcon_xmac_operations : &falcon_gmac_operations);
 
+	if (EFX_IS10G(efx))
+		stats_done_offset = XgDmaDone_offset;
+	else
+		stats_done_offset = GDmaDone_offset;
+	nic_data->stats_dma_done = efx->stats_buffer.addr + stats_done_offset;
+
 	if (old_mac_op == efx->mac_op)
 		goto out;
 
@@ -2235,7 +2255,7 @@ int falcon_switch_mac(struct efx_nic *efx)
 
 	rc = falcon_reset_macs(efx);
 out:
-	efx_stats_enable(efx);
+	falcon_start_nic_stats(efx);
 	return rc;
 }
 
@@ -2900,6 +2920,10 @@ int falcon_probe_nic(struct efx_nic *efx)
 		goto fail6;
 	}
 
+	nic_data->stats_disable_count = 1;
+	setup_timer(&nic_data->stats_timer, &falcon_stats_timer_func,
+		    (unsigned long)efx);
+
 	return 0;
 
  fail6:
@@ -3125,11 +3149,58 @@ void falcon_remove_nic(struct efx_nic *efx)
 
 void falcon_update_nic_stats(struct efx_nic *efx)
 {
+	struct falcon_nic_data *nic_data = efx->nic_data;
 	efx_oword_t cnt;
 
+	if (nic_data->stats_disable_count)
+		return;
+
 	efx_reado(efx, &cnt, FR_AZ_RX_NODESC_DROP);
 	efx->n_rx_nodesc_drop_cnt +=
 		EFX_OWORD_FIELD(cnt, FRF_AB_RX_NODESC_DROP_CNT);
+
+	if (nic_data->stats_pending &&
+	    *nic_data->stats_dma_done == FALCON_STATS_DONE) {
+		nic_data->stats_pending = false;
+		rmb(); /* read the done flag before the stats */
+		efx->mac_op->update_stats(efx);
+	}
+}
+
+void falcon_start_nic_stats(struct efx_nic *efx)
+{
+	struct falcon_nic_data *nic_data = efx->nic_data;
+
+	spin_lock_bh(&efx->stats_lock);
+	if (--nic_data->stats_disable_count == 0)
+		falcon_stats_request(efx);
+	spin_unlock_bh(&efx->stats_lock);
+}
+
+void falcon_stop_nic_stats(struct efx_nic *efx)
+{
+	struct falcon_nic_data *nic_data = efx->nic_data;
+	int i;
+
+	might_sleep();
+
+	spin_lock_bh(&efx->stats_lock);
+	++nic_data->stats_disable_count;
+	spin_unlock_bh(&efx->stats_lock);
+
+	del_timer_sync(&nic_data->stats_timer);
+
+	/* Wait enough time for the most recent transfer to
+	 * complete. */
+	for (i = 0; i < 4 && nic_data->stats_pending; i++) {
+		if (*nic_data->stats_dma_done == FALCON_STATS_DONE)
+			break;
+		msleep(1);
+	}
+
+	spin_lock_bh(&efx->stats_lock);
+	falcon_stats_complete(efx);
+	spin_unlock_bh(&efx->stats_lock);
 }
 
 /**************************************************************************

drivers/net/sfc/falcon.h

@@ -75,10 +75,18 @@ struct falcon_board {
  * struct falcon_nic_data - Falcon NIC state
  * @pci_dev2: Secondary function of Falcon A
  * @board: Board state and functions
+ * @stats_disable_count: Nest count for disabling statistics fetches
+ * @stats_pending: Is there a pending DMA of MAC statistics.
+ * @stats_timer: A timer for regularly fetching MAC statistics.
+ * @stats_dma_done: Pointer to the flag which indicates DMA completion.
  */
 struct falcon_nic_data {
 	struct pci_dev *pci_dev2;
 	struct falcon_board board;
+	unsigned int stats_disable_count;
+	bool stats_pending;
+	struct timer_list stats_timer;
+	u32 *stats_dma_done;
 };
 
 static inline struct falcon_board *falcon_board(struct efx_nic *efx)
@@ -128,8 +136,6 @@ extern void falcon_remove_port(struct efx_nic *efx);
 /* MAC/PHY */
 extern int falcon_switch_mac(struct efx_nic *efx);
 extern bool falcon_xaui_link_ok(struct efx_nic *efx);
-extern int falcon_dma_stats(struct efx_nic *efx,
-			    unsigned int done_offset);
 extern void falcon_drain_tx_fifo(struct efx_nic *efx);
 extern void falcon_deconfigure_mac_wrapper(struct efx_nic *efx);
 extern void falcon_reconfigure_mac_wrapper(struct efx_nic *efx);
@@ -154,6 +160,8 @@ extern int falcon_flush_queues(struct efx_nic *efx);
 extern int falcon_reset_hw(struct efx_nic *efx, enum reset_type method);
 extern void falcon_remove_nic(struct efx_nic *efx);
 extern void falcon_update_nic_stats(struct efx_nic *efx);
+extern void falcon_start_nic_stats(struct efx_nic *efx);
+extern void falcon_stop_nic_stats(struct efx_nic *efx);
 extern void falcon_set_multicast_hash(struct efx_nic *efx);
 extern int falcon_reset_xaui(struct efx_nic *efx);
 

drivers/net/sfc/falcon_boards.c

@@ -347,14 +347,14 @@ static ssize_t set_phy_flash_cfg(struct device *dev,
 		 * MAC stats accordingly. */
 		efx->phy_mode = new_mode;
 		if (new_mode & PHY_MODE_SPECIAL)
-			efx_stats_disable(efx);
+			falcon_stop_nic_stats(efx);
 		if (falcon_board(efx)->type->id == FALCON_BOARD_SFE4001)
 			err = sfe4001_poweron(efx);
 		else
 			err = sfn4111t_reset(efx);
 		efx_reconfigure_port(efx);
 		if (!(new_mode & PHY_MODE_SPECIAL))
-			efx_stats_enable(efx);
+			falcon_start_nic_stats(efx);
 	}
 	rtnl_unlock();
 
@@ -441,7 +441,7 @@ static int sfe4001_init(struct efx_nic *efx)
 	if (efx->phy_mode & PHY_MODE_SPECIAL) {
 		/* PHY won't generate a 156.25 MHz clock and MAC stats fetch
 		 * will fail. */
-		efx_stats_disable(efx);
+		falcon_stop_nic_stats(efx);
 	}
 	rc = sfe4001_poweron(efx);
 	if (rc)
@@ -504,7 +504,7 @@ static void sfn4111t_init_phy(struct efx_nic *efx)
 			return;
 
 		efx->phy_mode = PHY_MODE_SPECIAL;
-		efx_stats_disable(efx);
+		falcon_stop_nic_stats(efx);
 	}
 
 	sfn4111t_reset(efx);
@@ -531,7 +531,7 @@ static int sfn4111t_init(struct efx_nic *efx)
 	if (efx->phy_mode & PHY_MODE_SPECIAL)
 		/* PHY may not generate a 156.25 MHz clock and MAC
 		 * stats fetch will fail. */
-		efx_stats_disable(efx);
+		falcon_stop_nic_stats(efx);
 
 	return 0;
 

drivers/net/sfc/falcon_gmac.c

@@ -130,11 +130,6 @@ static void falcon_update_stats_gmac(struct efx_nic *efx)
 	struct efx_mac_stats *mac_stats = &efx->mac_stats;
 	unsigned long old_rx_pause, old_tx_pause;
 	unsigned long new_rx_pause, new_tx_pause;
-	int rc;
-
-	rc = falcon_dma_stats(efx, GDmaDone_offset);
-	if (rc)
-		return;
 
 	/* Pause frames are erroneously counted as errors (SFC bug 3269) */
 	old_rx_pause = mac_stats->rx_pause;

drivers/net/sfc/falcon_xmac.c

@@ -60,9 +60,13 @@ static void falcon_setup_xaui(struct efx_nic *efx)
 
 int falcon_reset_xaui(struct efx_nic *efx)
 {
+	struct falcon_nic_data *nic_data = efx->nic_data;
 	efx_oword_t reg;
 	int count;
 
+	/* Don't fetch MAC statistics over an XMAC reset */
+	WARN_ON(nic_data->stats_disable_count == 0);
+
 	/* Start reset sequence */
 	EFX_POPULATE_OWORD_1(reg, FRF_AB_XX_RST_XX_EN, 1);
 	efx_writeo(efx, &reg, FR_AB_XX_PWR_RST);
@@ -250,6 +254,8 @@ static void falcon_check_xaui_link_up(struct efx_nic *efx, int tries)
 		/* XAUI link is expected to be down */
 		return;
 
+	falcon_stop_nic_stats(efx);
+
 	while (!efx->mac_up && tries) {
 		EFX_LOG(efx, "bashing xaui\n");
 		falcon_reset_xaui(efx);
@@ -258,6 +264,8 @@ static void falcon_check_xaui_link_up(struct efx_nic *efx, int tries)
 		efx->mac_up = falcon_xaui_link_ok(efx);
 		--tries;
 	}
+
+	falcon_start_nic_stats(efx);
 }
 
 static void falcon_reconfigure_xmac(struct efx_nic *efx)
@@ -276,11 +284,6 @@ static void falcon_reconfigure_xmac(struct efx_nic *efx)
 static void falcon_update_stats_xmac(struct efx_nic *efx)
 {
 	struct efx_mac_stats *mac_stats = &efx->mac_stats;
-	int rc;
-
-	rc = falcon_dma_stats(efx, XgDmaDone_offset);
-	if (rc)
-		return;
 
 	/* Update MAC stats from DMAed values */
 	FALCON_STAT(efx, XgRxOctets, rx_bytes);

drivers/net/sfc/net_driver.h

@@ -717,7 +717,6 @@ union efx_multicast_hash {
  *	&struct net_device_stats.
  * @stats_buffer: DMA buffer for statistics
  * @stats_lock: Statistics update lock. Serialises statistics fetches
- * @stats_disable_count: Nest count for disabling statistics fetches
  * @mac_op: MAC interface
  * @mac_address: Permanent MAC address
  * @phy_type: PHY type
@@ -799,7 +798,6 @@ struct efx_nic {
 	struct efx_mac_stats mac_stats;
 	struct efx_buffer stats_buffer;
 	spinlock_t stats_lock;
-	unsigned int stats_disable_count;
 
 	struct efx_mac_operations *mac_op;
 	unsigned char mac_address[ETH_ALEN];

drivers/net/sfc/tenxpress.c

@@ -374,7 +374,7 @@ static int tenxpress_special_reset(struct efx_nic *efx)
 	/* The XGMAC clock is driven from the SFC7101/SFT9001 312MHz clock, so
 	 * a special software reset can glitch the XGMAC sufficiently for stats
 	 * requests to fail. */
-	efx_stats_disable(efx);
+	falcon_stop_nic_stats(efx);
 
 	/* Initiate reset */
 	reg = efx_mdio_read(efx, MDIO_MMD_PMAPMD, PMA_PMD_XCONTROL_REG);
@@ -396,7 +396,7 @@ static int tenxpress_special_reset(struct efx_nic *efx)
 	/* Wait for the XGXS state machine to churn */
 	mdelay(10);
 out:
-	efx_stats_enable(efx);
+	falcon_start_nic_stats(efx);
 	return rc;
 }