From mboxrd@z Thu Jan  1 00:00:00 1970
From: Andrei Vagin <avagin@virtuozzo.com>
Subject: Re: [PATCH net-next] net: core: Expose number of link up/down
 transitions
Date: Wed, 17 Jan 2018 16:23:09 -0800
Message-ID: <20180118002308.GA21605@outlook.office365.com>
References: <20180117230704.21949-1-f.fainelli@gmail.com>
 <20180117234941.GA19561@outlook.office365.com>
 <20180118000652.GD32299@lunn.ch>
Mime-Version: 1.0
Content-Type: text/plain; charset=koi8-r
Cc: Florian Fainelli <f.fainelli@gmail.com>, netdev@vger.kernel.org,
        edumazet@google.com, cphealy@gmail.com,
        David Decotigny <decot@googlers.com>,
        "David S. Miller" <davem@davemloft.net>,
        Jamal Hadi Salim <jhs@mojatatu.com>,
        Cong Wang <xiyou.wangcong@gmail.com>,
        Jiri Pirko <jiri@resnulli.us>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Nikolay Aleksandrov <nikolay@cumulusnetworks.com>,
        Alexei Starovoitov <ast@fb.com>,
        Roopa Prabhu <roopa@cumulusnetworks.com>,
        Mahesh Bandewar <maheshb@google.com>,
        Vlad Yasevich <vyasevich@gmail.com>,
        Jakub Kicinski <jakub.kicinski@netronome.com>,
        Jonas Bonn <jonas@southpole.se>,
        stephen hemminger <stephen@networkplumber.org>,
        Hans Liljestrand <ishkamiel@gmail.com>,
        "Reshetova, Elena" <elena.reshetova@intel.com>,
        Kiril
To: Andrew Lunn <andrew@lunn.ch>
Return-path: <linux-kernel-owner@vger.kernel.org>
Content-Disposition: inline
In-Reply-To: <20180118000652.GD32299@lunn.ch>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: netdev.vger.kernel.org

On Thu, Jan 18, 2018 at 01:06:52AM +0100, Andrew Lunn wrote:
> > What is the idea to have two separate counters? Can a delta between them
> > be a bigger than 1?
> 
> Yes, it can.
> 
> These counters are incremented in netif_carrier_on() /
> netif_carrier_off(). They are not always called in pairs and they can
> be called multiple times for the same event. The phylib will call them
> when it notices the PHY saying the link is down/up, and the MAC driver
> sometimes also calls them.

We check the __LINK_STATE_NOCARRIER bit before changing these counters,
so if we call netif_carrier_on() twice, the counter will be incrimented
only by one, will it not?

void netif_carrier_on(struct net_device *dev)
        if (test_and_clear_bit(__LINK_STATE_NOCARRIER, &dev->state)) {
                atomic_inc(&dev->carrier_changes);

...

void netif_carrier_off(struct net_device *dev)
        if (!test_and_set_bit(__LINK_STATE_NOCARRIER, &dev->state)) {
                atomic_inc(&dev->carrier_changes);


> 
> 	  Andrew