netdev - Re: [PATCH] mlx4: give precise rx/tx bytes/packets counters

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 28 Nov 2016 14:57:43 -0800
From:   Eric Dumazet <eric.dumazet@...il.com>
To:     Saeed Mahameed <saeedm@....mellanox.co.il>
Cc:     David Miller <davem@...emloft.net>,
        Linux Netdev List <netdev@...r.kernel.org>,
        Tariq Toukan <tariqt@...lanox.com>
Subject: Re: [PATCH] mlx4: give precise rx/tx bytes/packets counters

On Mon, 2016-11-28 at 23:55 +0200, Saeed Mahameed wrote:

> I have nothing against this patch, I just wanted to point out that
> this patch is just fixing the symptom.

> We keep ignoring the root cause that dev_get_stats is called under a
> spin_lock which really ties our hands "us device drivers developers"
> and push us towards those fragile solutions like deferred work for
> caching statistics.

Well, we do not want to allow a driver to wait several ms to fetch
stats, or allowing the operation to fail is memory for the transaction
can not be allocated.

Otherwise, some monitoring tools could really have serious problems
under stress.

We really don't care of how often the 'not really hot counters' are
fetched from the NIC. I guess even once per second should be good
enough.

But these 4 counters (bytes/packets rx/tx), especially if they are
given by counters managed in the driver, should reflect the most current
value.