diff mbox series

[v3,7/7,not,for,merge] netstats: example use of stats_fs API

Message ID 20200526110318.69006-8-eesposit@redhat.com (mailing list archive)
State New, archived
Headers show
Series Statsfs: a new ram-based file system for Linux kernel statistics | expand

Commit Message

Emanuele Giuseppe Esposito May 26, 2020, 11:03 a.m. UTC 
Apply stats_fs on the networking statistics subsystem.

Currently it only works with disabled network namespace
(CONFIG_NET_NS=n), because multiple namespaces will have the same
device name under the same root source that will cause a conflict in
stats_fs.

Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
---
 include/linux/netdevice.h |  2 ++
 net/Kconfig               |  1 +
 net/core/dev.c            | 66 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 69 insertions(+)

Comments

Andrew Lunn May 26, 2020, 2:16 p.m. UTC | #1 
On Tue, May 26, 2020 at 01:03:17PM +0200, Emanuele Giuseppe Esposito wrote:
> Apply stats_fs on the networking statistics subsystem.
> 
> Currently it only works with disabled network namespace
> (CONFIG_NET_NS=n), because multiple namespaces will have the same
> device name under the same root source that will cause a conflict in
> stats_fs.

Hi Emanuele

How do you atomically get and display a group of statistics?

If you look at how the netlink socket works, you will see code like:

                do {
                        start = u64_stats_fetch_begin_irq(&cpu_stats->syncp);
                        rx_packets = cpu_stats->rx_packets;
                        rx_bytes = cpu_stats->rx_bytes;
			....
                } while (u64_stats_fetch_retry_irq(&cpu_stats->syncp, start));

It will ensure that rx_packets and rx_bytes are consistent with each
other. If the value of the sequence counter changes while inside the
loop, the loop so repeated until it does not change.

In general, hardware counters in NICs are the same.  You tell it to
take a snapshot of the statistics counters, and then read them all
back, to give a consistent view across all the statistics.

I've not looked at this new code in detail, but it looks like you have
one file per statistic, and assume each statistic is independent of
every other statistic. This independence can limit how you use the
values, particularly when debugging. The netlink interface we use does
not have this limitation.

	     Andrew
Emanuele Giuseppe Esposito May 26, 2020, 3:45 p.m. UTC | #2 
Hi Andrew

> How do you atomically get and display a group of statistics?
> 
> If you look at how the netlink socket works, you will see code like:
> 
>                  do {
>                          start = u64_stats_fetch_begin_irq(&cpu_stats->syncp);
>                          rx_packets = cpu_stats->rx_packets;
>                          rx_bytes = cpu_stats->rx_bytes;
> 			....
>                  } while (u64_stats_fetch_retry_irq(&cpu_stats->syncp, start));
> 
> It will ensure that rx_packets and rx_bytes are consistent with each
> other. If the value of the sequence counter changes while inside the
> loop, the loop so repeated until it does not change.
> 
> In general, hardware counters in NICs are the same.  You tell it to
> take a snapshot of the statistics counters, and then read them all
> back, to give a consistent view across all the statistics.
> 
> I've not looked at this new code in detail, but it looks like you have
> one file per statistic, and assume each statistic is independent of
> every other statistic. This independence can limit how you use the
> values, particularly when debugging. The netlink interface we use does
> not have this limitation.

You're right, statistics are treated independently so what you describe 
is currently not supported.

In KVM the utilization is more qualitative, so there isn't such problem.
But as long as the interface is based on file access, the possibility of 
snapshotting might not be useful; however, it could still be considered 
to be added later together with the binary access.

Jonathan, how is your metricfs handling this case?

Thank you,
Emanuele
diff mbox series

Patch

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 130a668049ab..408c4e7b0e21 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -48,6 +48,7 @@ 
 #include <uapi/linux/if_bonding.h>
 #include <uapi/linux/pkt_cls.h>
 #include <linux/hashtable.h>
+#include <linux/stats_fs.h>
 
 struct netpoll_info;
 struct device;
@@ -2117,6 +2118,7 @@  struct net_device {
 	unsigned		wol_enabled:1;
 
 	struct list_head	net_notifier_list;
+	struct stats_fs_source	*stats_fs_src;
 
 #if IS_ENABLED(CONFIG_MACSEC)
 	/* MACsec management functions */
diff --git a/net/Kconfig b/net/Kconfig
index df8d8c9bd021..3441d5bb6107 100644
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -8,6 +8,7 @@  menuconfig NET
 	select NLATTR
 	select GENERIC_NET_UTILS
 	select BPF
+	select STATS_FS_API
 	---help---
 	  Unless you really know what you are doing, you should say Y here.
 	  The reason is that some programs need kernel networking support even
diff --git a/net/core/dev.c b/net/core/dev.c
index 522288177bbd..3db48cd1a097 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -142,6 +142,7 @@ 
 #include <linux/net_namespace.h>
 #include <linux/indirect_call_wrapper.h>
 #include <net/devlink.h>
+#include <linux/stats_fs.h>
 
 #include "net-sysfs.h"
 
@@ -150,6 +151,11 @@ 
 /* This should be increased if a protocol with a bigger head is added. */
 #define GRO_MAX_HEAD (MAX_HEADER + 128)
 
+#define NETDEV_STAT(str, m, ...)						\
+	{ str, offsetof(struct rtnl_link_stats64, m),				\
+	  &stats_fs_type_netdev_u64,						\
+	  STATS_FS_SUM, ## __VA_ARGS__ }
+
 static DEFINE_SPINLOCK(ptype_lock);
 static DEFINE_SPINLOCK(offload_lock);
 struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
@@ -196,6 +202,53 @@  static DEFINE_READ_MOSTLY_HASHTABLE(napi_hash, 8);
 
 static seqcount_t devnet_rename_seq;
 
+static uint64_t stats_fs_get_netdev_u64(struct stats_fs_value *val,
+					void *base)
+{
+	struct net_device *netdev = (struct net_device *)base;
+	struct rtnl_link_stats64 net_stats;
+
+	dev_get_stats(netdev, &net_stats);
+
+	return stats_fs_get_u64(val, &net_stats);
+}
+
+static struct stats_fs_type stats_fs_type_netdev_u64 = {
+	.get = stats_fs_get_netdev_u64,
+	.clear = NULL,
+	.sign = false
+};
+
+static struct stats_fs_source *netdev_root;
+
+static struct stats_fs_value stats_fs_netdev_entries[] = {
+	NETDEV_STAT("rx_packets", rx_packets),
+	NETDEV_STAT("tx_packets", tx_packets),
+	NETDEV_STAT("rx_bytes", rx_bytes),
+	NETDEV_STAT("tx_bytes", tx_bytes),
+	NETDEV_STAT("rx_errors", rx_errors),
+	NETDEV_STAT("tx_errors", tx_errors),
+	NETDEV_STAT("rx_dropped", rx_dropped),
+	NETDEV_STAT("tx_dropped", tx_dropped),
+	NETDEV_STAT("multicast", multicast),
+	NETDEV_STAT("collisions", collisions),
+	NETDEV_STAT("rx_length_errors", rx_length_errors),
+	NETDEV_STAT("rx_over_errors", rx_over_errors),
+	NETDEV_STAT("rx_crc_errors", rx_crc_errors),
+	NETDEV_STAT("rx_frame_errors", rx_frame_errors),
+	NETDEV_STAT("rx_fifo_errors", rx_fifo_errors),
+	NETDEV_STAT("rx_missed_errors", rx_missed_errors),
+	NETDEV_STAT("tx_aborted_errors", tx_aborted_errors),
+	NETDEV_STAT("tx_carrier_errors", tx_carrier_errors),
+	NETDEV_STAT("tx_fifo_errors", tx_fifo_errors),
+	NETDEV_STAT("tx_heartbeat_errors", tx_heartbeat_errors),
+	NETDEV_STAT("tx_window_errors", tx_window_errors),
+	NETDEV_STAT("rx_compressed", rx_compressed),
+	NETDEV_STAT("tx_compressed", tx_compressed),
+	NETDEV_STAT("rx_nohandler", rx_nohandler),
+	{ NULL }
+};
+
 static inline void dev_base_seq_inc(struct net *net)
 {
 	while (++net->dev_base_seq == 0)
@@ -8783,6 +8836,11 @@  static void rollback_registered_many(struct list_head *head)
 	ASSERT_RTNL();
 
 	list_for_each_entry_safe(dev, tmp, head, unreg_list) {
+		stats_fs_source_remove_subordinate(netdev_root,
+						   dev->stats_fs_src);
+		stats_fs_source_revoke(dev->stats_fs_src);
+		stats_fs_source_put(dev->stats_fs_src);
+
 		/* Some devices call without registering
 		 * for initialization unwind. Remove those
 		 * devices and proceed with the remaining.
@@ -9436,6 +9494,11 @@  int register_netdevice(struct net_device *dev)
 	    dev->rtnl_link_state == RTNL_LINK_INITIALIZED)
 		rtmsg_ifinfo(RTM_NEWLINK, dev, ~0U, GFP_KERNEL);
 
+	dev->stats_fs_src = stats_fs_source_create(0, dev->name);
+	stats_fs_source_add_subordinate(netdev_root, dev->stats_fs_src);
+	stats_fs_source_add_values(dev->stats_fs_src, stats_fs_netdev_entries,
+				   dev, 0);
+
 out:
 	return ret;
 
@@ -10500,6 +10563,9 @@  static int __init net_dev_init(void)
 	if (netdev_kobject_init())
 		goto out;
 
+	netdev_root = stats_fs_source_create(0, "net");
+	stats_fs_source_register(netdev_root);
+
 	INIT_LIST_HEAD(&ptype_all);
 	for (i = 0; i < PTYPE_HASH_SIZE; i++)
 		INIT_LIST_HEAD(&ptype_base[i]);