Re: [PATCH net-next] sock: don't enable netstamp for af_unix sockets

On Tue, Oct 27, 2015, at 14:19, Eric Dumazet wrote:

On Tue, 2015-10-27 at 12:15 +0100, Hannes Frederic Sowa wrote:

quoted

Also counter question: why is the netstamp code protected by a
static_key otherwise if not for trying to suppress the code path as
often as possible if not used? ;)

Any idea of why timestamping is asked on AF_UNIX in the first place ?

I guess syslog code want's to have more accurate timetstamps on when the
packet is send.

For messages sent/received on af_unix sockets, in which place timestamp
is taken ?

in unix_sendmsg on the sending unix socket (we check peer unix socket
for timestamp flag).

Is it at the time skb is cooked and stored in receive queue, or the time
it was dequeued ?

No, at time it is send by sendmsg on the sending socket.

In any case, is your patch changing af_unix behavior ? It is not clear
from your changelog...

No, af_unix logic does not pass this logic at all, so we don't need to
care about netstamp code. netstamp_needed is private to dev.c.

Bye,
Hannes