netdev - Re: Killing sk->sk_callback

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 for Android: free password hash cracker in your pocket

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

Date:	Wed, 18 Jun 2008 09:36:23 +0200
From:	Remi Denis-Courmont <rdenis@...phalempin.com>
To:	David Miller <davem@...emloft.net>
Cc:	pmullaney@...ell.com, herbert@...dor.apana.org.au,
	GHaskins.WAL-1.WALTHAM@...ell.com, chuck.lever@...cle.com,
	netdev@...r.kernel.org
Subject: Re: Killing sk->sk_callback_lock

    Hello,

On Tue, 17 Jun 2008 14:40:41 -0700 (PDT), David Miller
<davem@...emloft.net> wrote:
>> The task can go directly back into a wait. This will effectively yield 2
>> wake ups per udp request-response.
> 
> I made the mistake of assuming that a high performance threaded
> networking application would use non-blocking operations and
> select/poll/epoll, which is clearly not the case here.
> 
> It's blocking in a recv() and this is woken up by a write space
> extraneous wakeup.

With UDP, I have consistently gotten significantly (IIRC around 30%) better
results using plain recv()/send() on blocking sockets than
poll()/recv()/send() on non-blocking sockets, on Linux, on the fast path.
Of course, this assumes there is only one socket per thread. Namely, using
my Teredo IPv6 userland implementation.

You are surely in a better position than I can ever be as to explaining why
this is so, and how bad and wrong my approach may be. Nevertheless my guess
has been the system call overhead is simply higher when adding poll() to
recv() and send().

-- 
Rémi Denis-Courmont
http://www.remlab.net

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html