netdev - Re: 3.14 regression: huge latency in read/select on tun

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]

Message-ID: <1396456973.3989.41.camel@edumazet-glaptop2.roam.corp.google.com>
Date:	Wed, 02 Apr 2014 09:42:53 -0700
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Ortwin Glück <odi@....ch>
Cc:	linux-kernel@...r.kernel.org, netdev@...r.kernel.org
Subject: Re: 3.14 regression: huge latency in read/select on tun

On Wed, 2014-04-02 at 18:17 +0200, Ortwin Glück wrote:
> Hi,
> 
> Since 3.14 the openconnect VPN tunnel becomes unusable for me because
> packets appear on the tun device at a horribly low rate. 3.12 and 3.13
> do not exhibt the problem.
> 
> Here is an strace of openconnect trying to read from its fd
>   7 -> /dev/net/tun
> 
> 15:07:33.130640 read(7, 0x1e05e58, 1280) = -1 EAGAIN (Resource
> temporarily unavailable)
> ===> should return available data already
> 15:07:33.130745 select(8, [3 6 7], [], [6], {30, 0}) = 1 (in [7], left
> {29, 783272})
> ===> HUGE 217ms delay here
> 15:07:33.347681 read(6, 0x1dfc973, 5)   = -1 EAGAIN (Resource
> temporarily unavailable)
> 15:07:33.347806 read(7,
> "E\10\5\0b\343@\0@\6\17~\n\363X\236\n\271UE\222:\0\26\37O\7\342\315\21q\33"...,
> 1280) = 1280
> 
> 
> The send queue of the socket being routed via the tun device has a lot
> of outstanding data (here an scp/ssh upload):
> tcp        0  29788 local:46577  remote:22   ESTABLISHED
> 
> (IPs replaced for privacy)
> 
> toggling TCP autocorking has no influence.
> 
> Any ideas what could be the culprit?

Nothing else than autocorking comes to mind.

You could post a tcpdump maybe ?

I believe some tools exist to anonymize a pcap file.



--
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