* [PATCH net-next v8 0/2] net: dsa: add stats64 support
@ 2021-01-11 10:46 Oleksij Rempel
2021-01-11 10:46 ` [PATCH net-next v8 1/2] net: dsa: add optional " Oleksij Rempel
` (2 more replies)
0 siblings, 3 replies; 4+ messages in thread
From: Oleksij Rempel @ 2021-01-11 10:46 UTC (permalink / raw)
To: Andrew Lunn, Vivien Didelot, Florian Fainelli, Vladimir Oltean,
David S. Miller, Jakub Kicinski, Russell King
Cc: Oleksij Rempel, Pengutronix Kernel Team, netdev, linux-kernel,
linux-mips
changes v8:
- stats.no_handler should not be assigned from HW stats
changes v7:
- move raw.filtered from rx_errors to rx_dropped counter
changes v6:
- move stats64 callback to ethtool section
- ar9331: diff. fixes
- ar9331: move stats calculation to the worker
- ar9331: extend rx/tx error counters
- use spin lock instead of u64_stats*
changes v5:
- read all stats in one regmap_bulk_read() request
- protect stats with u64_stats* helpers.
changes v4:
- do no read MIBs withing stats64 call
- change polling frequency to 0.3Hz
changes v3:
- fix wrong multiplication
- cancel port workers on remove
changes v2:
- use stats64 instead of get_ethtool_stats
- add worked to poll for the stats
Oleksij Rempel (2):
net: dsa: add optional stats64 support
net: dsa: qca: ar9331: export stats64
drivers/net/dsa/qca/ar9331.c | 163 ++++++++++++++++++++++++++++++++++-
include/net/dsa.h | 4 +-
net/dsa/slave.c | 14 ++-
3 files changed, 178 insertions(+), 3 deletions(-)
--
2.30.0
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH net-next v8 1/2] net: dsa: add optional stats64 support
2021-01-11 10:46 [PATCH net-next v8 0/2] net: dsa: add stats64 support Oleksij Rempel
@ 2021-01-11 10:46 ` Oleksij Rempel
2021-01-11 10:46 ` [PATCH net-next v8 2/2] net: dsa: qca: ar9331: export stats64 Oleksij Rempel
2021-01-13 4:30 ` [PATCH net-next v8 0/2] net: dsa: add stats64 support patchwork-bot+netdevbpf
2 siblings, 0 replies; 4+ messages in thread
From: Oleksij Rempel @ 2021-01-11 10:46 UTC (permalink / raw)
To: Andrew Lunn, Vivien Didelot, Florian Fainelli, Vladimir Oltean,
David S. Miller, Jakub Kicinski, Russell King
Cc: Oleksij Rempel, Pengutronix Kernel Team, netdev, linux-kernel,
linux-mips
Allow DSA drivers to export stats64
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
---
include/net/dsa.h | 4 +++-
net/dsa/slave.c | 14 +++++++++++++-
2 files changed, 16 insertions(+), 2 deletions(-)
diff --git a/include/net/dsa.h b/include/net/dsa.h
index 4e60d2610f20..c50abb295fb6 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -477,7 +477,7 @@ struct dsa_switch_ops {
void (*phylink_fixed_state)(struct dsa_switch *ds, int port,
struct phylink_link_state *state);
/*
- * ethtool hardware statistics.
+ * Port statistics counters.
*/
void (*get_strings)(struct dsa_switch *ds, int port,
u32 stringset, uint8_t *data);
@@ -486,6 +486,8 @@ struct dsa_switch_ops {
int (*get_sset_count)(struct dsa_switch *ds, int port, int sset);
void (*get_ethtool_phy_stats)(struct dsa_switch *ds,
int port, uint64_t *data);
+ void (*get_stats64)(struct dsa_switch *ds, int port,
+ struct rtnl_link_stats64 *s);
/*
* ethtool Wake-on-LAN
diff --git a/net/dsa/slave.c b/net/dsa/slave.c
index 4a0498bf6c65..1fb823d3c7d7 100644
--- a/net/dsa/slave.c
+++ b/net/dsa/slave.c
@@ -1602,6 +1602,18 @@ static struct devlink_port *dsa_slave_get_devlink_port(struct net_device *dev)
return dp->ds->devlink ? &dp->devlink_port : NULL;
}
+static void dsa_slave_get_stats64(struct net_device *dev,
+ struct rtnl_link_stats64 *s)
+{
+ struct dsa_port *dp = dsa_slave_to_port(dev);
+ struct dsa_switch *ds = dp->ds;
+
+ if (ds->ops->get_stats64)
+ ds->ops->get_stats64(ds, dp->index, s);
+ else
+ dev_get_tstats64(dev, s);
+}
+
static const struct net_device_ops dsa_slave_netdev_ops = {
.ndo_open = dsa_slave_open,
.ndo_stop = dsa_slave_close,
@@ -1621,7 +1633,7 @@ static const struct net_device_ops dsa_slave_netdev_ops = {
#endif
.ndo_get_phys_port_name = dsa_slave_get_phys_port_name,
.ndo_setup_tc = dsa_slave_setup_tc,
- .ndo_get_stats64 = dev_get_tstats64,
+ .ndo_get_stats64 = dsa_slave_get_stats64,
.ndo_get_port_parent_id = dsa_slave_get_port_parent_id,
.ndo_vlan_rx_add_vid = dsa_slave_vlan_rx_add_vid,
.ndo_vlan_rx_kill_vid = dsa_slave_vlan_rx_kill_vid,
--
2.30.0
^ permalink raw reply related [flat|nested] 4+ messages in thread
* [PATCH net-next v8 2/2] net: dsa: qca: ar9331: export stats64
2021-01-11 10:46 [PATCH net-next v8 0/2] net: dsa: add stats64 support Oleksij Rempel
2021-01-11 10:46 ` [PATCH net-next v8 1/2] net: dsa: add optional " Oleksij Rempel
@ 2021-01-11 10:46 ` Oleksij Rempel
2021-01-13 4:30 ` [PATCH net-next v8 0/2] net: dsa: add stats64 support patchwork-bot+netdevbpf
2 siblings, 0 replies; 4+ messages in thread
From: Oleksij Rempel @ 2021-01-11 10:46 UTC (permalink / raw)
To: Andrew Lunn, Vivien Didelot, Florian Fainelli, Vladimir Oltean,
David S. Miller, Jakub Kicinski, Russell King
Cc: Oleksij Rempel, Pengutronix Kernel Team, netdev, linux-kernel,
linux-mips
Add stats support for the ar9331 switch.
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
---
drivers/net/dsa/qca/ar9331.c | 163 ++++++++++++++++++++++++++++++++++-
1 file changed, 162 insertions(+), 1 deletion(-)
diff --git a/drivers/net/dsa/qca/ar9331.c b/drivers/net/dsa/qca/ar9331.c
index 4d49c5f2b790..1e3706054ae1 100644
--- a/drivers/net/dsa/qca/ar9331.c
+++ b/drivers/net/dsa/qca/ar9331.c
@@ -101,6 +101,9 @@
AR9331_SW_PORT_STATUS_RX_FLOW_EN | AR9331_SW_PORT_STATUS_TX_FLOW_EN | \
AR9331_SW_PORT_STATUS_SPEED_M)
+/* MIB registers */
+#define AR9331_MIB_COUNTER(x) (0x20000 + ((x) * 0x100))
+
/* Phy bypass mode
* ------------------------------------------------------------------------
* Bit: | 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |10 |11 |12 |13 |14 |15 |
@@ -154,6 +157,66 @@
#define AR9331_SW_MDIO_POLL_SLEEP_US 1
#define AR9331_SW_MDIO_POLL_TIMEOUT_US 20
+/* The interval should be small enough to avoid overflow of 32bit MIBs */
+/*
+ * FIXME: until we can read MIBs from stats64 call directly (i.e. sleep
+ * there), we have to poll stats more frequently then it is actually needed.
+ * For overflow protection, normally, 100 sec interval should have been OK.
+ */
+#define STATS_INTERVAL_JIFFIES (3 * HZ)
+
+struct ar9331_sw_stats_raw {
+ u32 rxbroad; /* 0x00 */
+ u32 rxpause; /* 0x04 */
+ u32 rxmulti; /* 0x08 */
+ u32 rxfcserr; /* 0x0c */
+ u32 rxalignerr; /* 0x10 */
+ u32 rxrunt; /* 0x14 */
+ u32 rxfragment; /* 0x18 */
+ u32 rx64byte; /* 0x1c */
+ u32 rx128byte; /* 0x20 */
+ u32 rx256byte; /* 0x24 */
+ u32 rx512byte; /* 0x28 */
+ u32 rx1024byte; /* 0x2c */
+ u32 rx1518byte; /* 0x30 */
+ u32 rxmaxbyte; /* 0x34 */
+ u32 rxtoolong; /* 0x38 */
+ u32 rxgoodbyte; /* 0x3c */
+ u32 rxgoodbyte_hi;
+ u32 rxbadbyte; /* 0x44 */
+ u32 rxbadbyte_hi;
+ u32 rxoverflow; /* 0x4c */
+ u32 filtered; /* 0x50 */
+ u32 txbroad; /* 0x54 */
+ u32 txpause; /* 0x58 */
+ u32 txmulti; /* 0x5c */
+ u32 txunderrun; /* 0x60 */
+ u32 tx64byte; /* 0x64 */
+ u32 tx128byte; /* 0x68 */
+ u32 tx256byte; /* 0x6c */
+ u32 tx512byte; /* 0x70 */
+ u32 tx1024byte; /* 0x74 */
+ u32 tx1518byte; /* 0x78 */
+ u32 txmaxbyte; /* 0x7c */
+ u32 txoversize; /* 0x80 */
+ u32 txbyte; /* 0x84 */
+ u32 txbyte_hi;
+ u32 txcollision; /* 0x8c */
+ u32 txabortcol; /* 0x90 */
+ u32 txmulticol; /* 0x94 */
+ u32 txsinglecol; /* 0x98 */
+ u32 txexcdefer; /* 0x9c */
+ u32 txdefer; /* 0xa0 */
+ u32 txlatecol; /* 0xa4 */
+};
+
+struct ar9331_sw_port {
+ int idx;
+ struct delayed_work mib_read;
+ struct rtnl_link_stats64 stats;
+ struct spinlock stats_lock;
+};
+
struct ar9331_sw_priv {
struct device *dev;
struct dsa_switch ds;
@@ -165,8 +228,17 @@ struct ar9331_sw_priv {
struct mii_bus *sbus; /* mdio slave */
struct regmap *regmap;
struct reset_control *sw_reset;
+ struct ar9331_sw_port port[AR9331_SW_PORTS];
};
+static struct ar9331_sw_priv *ar9331_sw_port_to_priv(struct ar9331_sw_port *port)
+{
+ struct ar9331_sw_port *p = port - port->idx;
+
+ return (struct ar9331_sw_priv *)((void *)p -
+ offsetof(struct ar9331_sw_priv, port));
+}
+
/* Warning: switch reset will reset last AR9331_SW_MDIO_PHY_MODE_PAGE request
* If some kind of optimization is used, the request should be repeated.
*/
@@ -424,6 +496,7 @@ static void ar9331_sw_phylink_mac_link_down(struct dsa_switch *ds, int port,
phy_interface_t interface)
{
struct ar9331_sw_priv *priv = (struct ar9331_sw_priv *)ds->priv;
+ struct ar9331_sw_port *p = &priv->port[port];
struct regmap *regmap = priv->regmap;
int ret;
@@ -431,6 +504,8 @@ static void ar9331_sw_phylink_mac_link_down(struct dsa_switch *ds, int port,
AR9331_SW_PORT_STATUS_MAC_MASK, 0);
if (ret)
dev_err_ratelimited(priv->dev, "%s: %i\n", __func__, ret);
+
+ cancel_delayed_work_sync(&p->mib_read);
}
static void ar9331_sw_phylink_mac_link_up(struct dsa_switch *ds, int port,
@@ -441,10 +516,13 @@ static void ar9331_sw_phylink_mac_link_up(struct dsa_switch *ds, int port,
bool tx_pause, bool rx_pause)
{
struct ar9331_sw_priv *priv = (struct ar9331_sw_priv *)ds->priv;
+ struct ar9331_sw_port *p = &priv->port[port];
struct regmap *regmap = priv->regmap;
u32 val;
int ret;
+ schedule_delayed_work(&p->mib_read, 0);
+
val = AR9331_SW_PORT_STATUS_MAC_MASK;
switch (speed) {
case SPEED_1000:
@@ -477,6 +555,73 @@ static void ar9331_sw_phylink_mac_link_up(struct dsa_switch *ds, int port,
dev_err_ratelimited(priv->dev, "%s: %i\n", __func__, ret);
}
+static void ar9331_read_stats(struct ar9331_sw_port *port)
+{
+ struct ar9331_sw_priv *priv = ar9331_sw_port_to_priv(port);
+ struct rtnl_link_stats64 *stats = &port->stats;
+ struct ar9331_sw_stats_raw raw;
+ int ret;
+
+ /* Do the slowest part first, to avoid needless locking for long time */
+ ret = regmap_bulk_read(priv->regmap, AR9331_MIB_COUNTER(port->idx),
+ &raw, sizeof(raw) / sizeof(u32));
+ if (ret) {
+ dev_err_ratelimited(priv->dev, "%s: %i\n", __func__, ret);
+ return;
+ }
+ /* All MIB counters are cleared automatically on read */
+
+ spin_lock(&port->stats_lock);
+
+ stats->rx_bytes += raw.rxgoodbyte;
+ stats->tx_bytes += raw.txbyte;
+
+ stats->rx_packets += raw.rx64byte + raw.rx128byte + raw.rx256byte +
+ raw.rx512byte + raw.rx1024byte + raw.rx1518byte + raw.rxmaxbyte;
+ stats->tx_packets += raw.tx64byte + raw.tx128byte + raw.tx256byte +
+ raw.tx512byte + raw.tx1024byte + raw.tx1518byte + raw.txmaxbyte;
+
+ stats->rx_length_errors += raw.rxrunt + raw.rxfragment + raw.rxtoolong;
+ stats->rx_crc_errors += raw.rxfcserr;
+ stats->rx_frame_errors += raw.rxalignerr;
+ stats->rx_missed_errors += raw.rxoverflow;
+ stats->rx_dropped += raw.filtered;
+ stats->rx_errors += raw.rxfcserr + raw.rxalignerr + raw.rxrunt +
+ raw.rxfragment + raw.rxoverflow + raw.rxtoolong;
+
+ stats->tx_window_errors += raw.txlatecol;
+ stats->tx_fifo_errors += raw.txunderrun;
+ stats->tx_aborted_errors += raw.txabortcol;
+ stats->tx_errors += raw.txoversize + raw.txabortcol + raw.txunderrun +
+ raw.txlatecol;
+
+ stats->multicast += raw.rxmulti;
+ stats->collisions += raw.txcollision;
+
+ spin_unlock(&port->stats_lock);
+}
+
+static void ar9331_do_stats_poll(struct work_struct *work)
+{
+ struct ar9331_sw_port *port = container_of(work, struct ar9331_sw_port,
+ mib_read.work);
+
+ ar9331_read_stats(port);
+
+ schedule_delayed_work(&port->mib_read, STATS_INTERVAL_JIFFIES);
+}
+
+static void ar9331_get_stats64(struct dsa_switch *ds, int port,
+ struct rtnl_link_stats64 *s)
+{
+ struct ar9331_sw_priv *priv = (struct ar9331_sw_priv *)ds->priv;
+ struct ar9331_sw_port *p = &priv->port[port];
+
+ spin_lock(&p->stats_lock);
+ memcpy(s, &p->stats, sizeof(*s));
+ spin_unlock(&p->stats_lock);
+}
+
static const struct dsa_switch_ops ar9331_sw_ops = {
.get_tag_protocol = ar9331_sw_get_tag_protocol,
.setup = ar9331_sw_setup,
@@ -485,6 +630,7 @@ static const struct dsa_switch_ops ar9331_sw_ops = {
.phylink_mac_config = ar9331_sw_phylink_mac_config,
.phylink_mac_link_down = ar9331_sw_phylink_mac_link_down,
.phylink_mac_link_up = ar9331_sw_phylink_mac_link_up,
+ .get_stats64 = ar9331_get_stats64,
};
static irqreturn_t ar9331_sw_irq(int irq, void *data)
@@ -796,7 +942,7 @@ static int ar9331_sw_probe(struct mdio_device *mdiodev)
{
struct ar9331_sw_priv *priv;
struct dsa_switch *ds;
- int ret;
+ int ret, i;
priv = devm_kzalloc(&mdiodev->dev, sizeof(*priv), GFP_KERNEL);
if (!priv)
@@ -831,6 +977,14 @@ static int ar9331_sw_probe(struct mdio_device *mdiodev)
ds->ops = &priv->ops;
dev_set_drvdata(&mdiodev->dev, priv);
+ for (i = 0; i < ARRAY_SIZE(priv->port); i++) {
+ struct ar9331_sw_port *port = &priv->port[i];
+
+ port->idx = i;
+ spin_lock_init(&port->stats_lock);
+ INIT_DELAYED_WORK(&port->mib_read, ar9331_do_stats_poll);
+ }
+
ret = dsa_register_switch(ds);
if (ret)
goto err_remove_irq;
@@ -846,6 +1000,13 @@ static int ar9331_sw_probe(struct mdio_device *mdiodev)
static void ar9331_sw_remove(struct mdio_device *mdiodev)
{
struct ar9331_sw_priv *priv = dev_get_drvdata(&mdiodev->dev);
+ unsigned int i;
+
+ for (i = 0; i < ARRAY_SIZE(priv->port); i++) {
+ struct ar9331_sw_port *port = &priv->port[i];
+
+ cancel_delayed_work_sync(&port->mib_read);
+ }
irq_domain_remove(priv->irqdomain);
mdiobus_unregister(priv->mbus);
--
2.30.0
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH net-next v8 0/2] net: dsa: add stats64 support
2021-01-11 10:46 [PATCH net-next v8 0/2] net: dsa: add stats64 support Oleksij Rempel
2021-01-11 10:46 ` [PATCH net-next v8 1/2] net: dsa: add optional " Oleksij Rempel
2021-01-11 10:46 ` [PATCH net-next v8 2/2] net: dsa: qca: ar9331: export stats64 Oleksij Rempel
@ 2021-01-13 4:30 ` patchwork-bot+netdevbpf
2 siblings, 0 replies; 4+ messages in thread
From: patchwork-bot+netdevbpf @ 2021-01-13 4:30 UTC (permalink / raw)
To: Oleksij Rempel
Cc: andrew, vivien.didelot, f.fainelli, olteanv, davem, kuba, linux,
kernel, netdev, linux-kernel, linux-mips
Hello:
This series was applied to netdev/net-next.git (refs/heads/master):
On Mon, 11 Jan 2021 11:46:56 +0100 you wrote:
> changes v8:
> - stats.no_handler should not be assigned from HW stats
>
> changes v7:
> - move raw.filtered from rx_errors to rx_dropped counter
>
> changes v6:
> - move stats64 callback to ethtool section
> - ar9331: diff. fixes
> - ar9331: move stats calculation to the worker
> - ar9331: extend rx/tx error counters
> - use spin lock instead of u64_stats*
>
> [...]
Here is the summary with links:
- [net-next,v8,1/2] net: dsa: add optional stats64 support
https://git.kernel.org/netdev/net-next/c/c2ec5f2ecf6a
- [net-next,v8,2/2] net: dsa: qca: ar9331: export stats64
https://git.kernel.org/netdev/net-next/c/bf9ce385932b
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2021-01-13 4:31 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-01-11 10:46 [PATCH net-next v8 0/2] net: dsa: add stats64 support Oleksij Rempel
2021-01-11 10:46 ` [PATCH net-next v8 1/2] net: dsa: add optional " Oleksij Rempel
2021-01-11 10:46 ` [PATCH net-next v8 2/2] net: dsa: qca: ar9331: export stats64 Oleksij Rempel
2021-01-13 4:30 ` [PATCH net-next v8 0/2] net: dsa: add stats64 support patchwork-bot+netdevbpf
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).