2014-03-27 23:25 GMT-07:00 Jiri Pirko <j...@resnulli.us>:
> Fri, Mar 28, 2014 at 05:35:00AM CET, de...@googlers.com wrote:
>>Tested:
>> grep . /sys/class/net/*/count_link_*
>> + ip link set dev X down/up
>>
>>Signed-off-by: David Decotigny <de...@googlers.com>
>>---
>> include/linux/netdevice.h | 4 ++++
>> net/core/net-sysfs.c | 18 ++++++++++++++++++
>> net/sched/sch_generic.c | 2 ++
>> 3 files changed, 24 insertions(+)
>>
>>diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
>>index 4b6d12c..cf52869 100644
>>--- a/include/linux/netdevice.h
>>+++ b/include/linux/netdevice.h
>>@@ -1315,6 +1315,10 @@ struct net_device {
>> * Do not use this in drivers.
>> */
>>
>>+ /* Stats to monitor link on/off, flapping */
>>+ atomic_t count_link_up;
>>+ atomic_t count_link_down;
>>+
>> #ifdef CONFIG_WIRELESS_EXT
>> /* List of functions to handle Wireless Extensions (instead of ioctl).
>> * See <net/iw_handler.h> for details. Jean II */
>>diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
>>index daed9a6..a65ac54 100644
>>--- a/net/core/net-sysfs.c
>>+++ b/net/core/net-sysfs.c
>>@@ -253,6 +253,22 @@ static ssize_t operstate_show(struct device *dev,
>> }
>> static DEVICE_ATTR_RO(operstate);
>>
>>+static ssize_t count_link_up_show(struct device *dev,
>>+ struct device_attribute *attr, char *buf)
>>+{
>>+ struct net_device *netdev = to_net_dev(dev);
>>+ return sprintf(buf, fmt_dec, atomic_read(&netdev->count_link_up));
>>+}
>>+
>>+static ssize_t count_link_down_show(struct device *dev,
>>+ struct device_attribute *attr, char *buf)
>>+{
>>+ struct net_device *netdev = to_net_dev(dev);
>>+ return sprintf(buf, fmt_dec, atomic_read(&netdev->count_link_down));
>>+}
>>+static DEVICE_ATTR_RO(count_link_up);
>>+static DEVICE_ATTR_RO(count_link_down);
>>+
>
>
> If this is exposed in sysfs, this needs to be exposed via RT netlink as
> well.
This makes me wonder if this really needs to be a kernel-maintained
pair of counters. Since this seems to be useful for monitoring e.g:
servers for link flapping, I would assume that some user-space
script/program does read these sysfs entries periodically to report a
healthy link. If that's the case, what prevents the same
script/program from listening to netlink events and count the link
UP/DOWN events from there and report that?
>
>
>> /* read-write attributes */
>>
>> static int change_mtu(struct net_device *net, unsigned long new_mtu)
>>@@ -386,6 +402,8 @@ static struct attribute *net_class_attrs[] = {
>> &dev_attr_duplex.attr,
>> &dev_attr_dormant.attr,
>> &dev_attr_operstate.attr,
>>+ &dev_attr_count_link_up.attr,
>>+ &dev_attr_count_link_down.attr,
>> &dev_attr_ifalias.attr,
>> &dev_attr_carrier.attr,
>> &dev_attr_mtu.attr,
>>diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
>>index e82e43b..2d06943 100644
>>--- a/net/sched/sch_generic.c
>>+++ b/net/sched/sch_generic.c
>>@@ -310,6 +310,7 @@ void netif_carrier_on(struct net_device *dev)
>> if (test_and_clear_bit(__LINK_STATE_NOCARRIER, &dev->state)) {
>> if (dev->reg_state == NETREG_UNINITIALIZED)
>> return;
>>+ atomic_inc(&dev->count_link_up);
>> linkwatch_fire_event(dev);
>> if (netif_running(dev))
>> __netdev_watchdog_up(dev);
>>@@ -328,6 +329,7 @@ void netif_carrier_off(struct net_device *dev)
>> if (!test_and_set_bit(__LINK_STATE_NOCARRIER, &dev->state)) {
>> if (dev->reg_state == NETREG_UNINITIALIZED)
>> return;
>>+ atomic_inc(&dev->count_link_down);
>> linkwatch_fire_event(dev);
>> }
>> }
>>--
>>1.9.1.423.g4596e3a
>>
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majord...@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
Florian
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
