netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Ingo Molnar <mingo@elte.hu>
To: linux-kernel@vger.kernel.org
Cc: netdev@vger.kernel.org, "David S. Miller" <davem@davemloft.net>,
	"Rafael J. Wysocki" <rjw@sisk.pl>,
	Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [bug] stuck localhost TCP connections, v2.6.26-rc3+
Date: Thu, 29 May 2008 10:45:24 +0200	[thread overview]
Message-ID: <20080529084524.GA24892@elte.hu> (raw)
In-Reply-To: <20080526115628.GA31316@elte.hu>


* Ingo Molnar <mingo@elte.hu> wrote:

> in an overnight -tip testruns that is based on recent -git i got two 
> stuck TCP connections:
> 
> Active Internet connections (w/o servers)
> Proto Recv-Q Send-Q Local Address               Foreign Address             State      
> tcp        0 174592 10.0.1.14:58015             10.0.1.14:3632              ESTABLISHED 
> tcp    72134      0 10.0.1.14:3632              10.0.1.14:58015             ESTABLISHED 

update: in the past 5 days of -tip testing i've gathered about 10 
randconfig kernel configs that all produced such failures.

Since the bug itself is very elusive (it takes up to 50 boot + 
kernel-rebuild-via-distccc iterations to trigger) bisection was still 
not an option - but with 10 configs statistical analysis of the configs 
is now possible.

I made a histogram of all kernel options present in those configs, and 
one networking related kernel option stood out:

      5 CONFIG_TCP_CONG_ADVANCED=y
      6 CONFIG_INET_TCP_DIAG=y
      6 CONFIG_TCP_MD5SIG=y
      9 CONFIG_TCP_CONG_CUBIC=y

that code is called in the bootlogs:

> [   13.279410] calling  cubictcp_register+0x0/0x80
> [   13.279412] TCP cubic registered

the likelyhood of CONFIG_TCP_CONG_CUBIC=y being enabled in my randconfig 
runs is 75%. The likelyhood of CONFIG_TCP_CONG_CUBIC=y being enabled in 
10 configs in a row is 0.75^10, or 5.6%. So statistical analysis can say 
it with a 95% confidence that the presence of this option correlates to 
the hung sockets.

i have started testing this theory now, via the patch below, which turns 
off TCP_CONG_CUBIC. It will take about 50 bootups on the affected 
testsystems to confirm. (it will take a couple of hours today as not all 
testsystems show these hung socket symptoms)

distributions enable TCP_CONG_CUBIC by default:

  $ grep CUBIC /boot/config-2.6.24.7-92.fc8
  CONFIG_TCP_CONG_CUBIC=y
  CONFIG_DEFAULT_CUBIC=y

which would explain why Arjan and Peter triggered similar hangs as well.

	Ingo

---------------------->
Subject: qa: no TCP_CONG_CUBIC
From: Ingo Molnar <mingo@elte.hu>
Date: Thu May 29 09:45:51 CEST 2008

---
 net/ipv4/Kconfig |    4 ++++
 1 file changed, 4 insertions(+)

Index: tip/net/ipv4/Kconfig
===================================================================
--- tip.orig/net/ipv4/Kconfig
+++ tip/net/ipv4/Kconfig
@@ -454,6 +454,8 @@ config TCP_CONG_BIC
 config TCP_CONG_CUBIC
 	tristate "CUBIC TCP"
 	default y
+	depends on BROKEN_BOOT_ALLOWED
+	select BROKEN_BOOT
 	---help---
 	This is version 2.0 of BIC-TCP which uses a cubic growth function
 	among other techniques.
@@ -608,6 +610,8 @@ endif
 config TCP_CONG_CUBIC
 	tristate
 	depends on !TCP_CONG_ADVANCED
+	depends on BROKEN_BOOT_ALLOWED
+	select BROKEN_BOOT
 	default y
 
 config DEFAULT_TCP_CONG

  parent reply	other threads:[~2008-05-29  8:45 UTC|newest]

Thread overview: 96+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-05-26 11:56 [bug] stuck localhost TCP connections, v2.6.26-rc3+ Ingo Molnar
2008-05-26 13:28 ` Ilpo Järvinen
2008-05-26 13:59   ` Ingo Molnar
2008-05-26 14:12     ` Ingo Molnar
2008-05-26 14:17       ` Ingo Molnar
2008-05-26 14:29         ` Ingo Molnar
2008-05-26 14:43         ` Ilpo Järvinen
2008-05-26 14:58       ` Ilpo Järvinen
2008-05-26 16:23         ` Ingo Molnar
2008-05-26 16:32           ` Ilpo Järvinen
2008-05-26 16:54             ` Ingo Molnar
2008-05-26 17:08               ` Ilpo Järvinen
2008-05-26 18:12                 ` Ingo Molnar
2008-05-26 20:41                   ` Ingo Molnar
2008-05-26 21:20                     ` Ilpo Järvinen
2008-05-30 16:23   ` Ray Lee
2008-05-26 16:24 ` Arjan van de Ven
2008-05-28  9:27 ` Peter Zijlstra
2008-05-31 14:25   ` Håkon Løvdal
2008-05-31 16:09     ` Ilpo Järvinen
2008-05-31 17:22       ` Ilpo Järvinen
2008-05-31 17:58       ` Håkon Løvdal
2008-05-31 18:37         ` Ilpo Järvinen
2008-05-31 20:25           ` Håkon Løvdal
2008-05-31 21:39             ` Ilpo Järvinen
2008-05-31 21:45               ` Håkon Løvdal
2008-06-04  0:10               ` Håkon Løvdal
2008-06-04 11:14                 ` Ilpo Järvinen
2008-06-04 14:00                   ` Håkon Løvdal
2008-06-04 15:09                     ` Ilpo Järvinen
2008-06-06  9:32                       ` Håkon Løvdal
2008-06-09 19:24                         ` Ilpo Järvinen
2008-06-10 23:26                           ` Håkon Løvdal
2008-06-11 13:39                             ` Ilpo Järvinen
2008-06-19  0:30                               ` Håkon Løvdal
2008-05-29  8:45 ` Ingo Molnar [this message]
2008-05-29 11:14   ` Ilpo Järvinen
2008-05-29 11:22     ` Ingo Molnar
2008-05-29 13:05       ` Evgeniy Polyakov
2008-05-29 13:43         ` Ingo Molnar
2008-05-29 13:08       ` Ingo Molnar
2008-05-29 13:48         ` Ilpo Järvinen
2008-05-30 11:09         ` Ingo Molnar
2008-05-30 21:12           ` Ilpo Järvinen
2008-05-30 18:18       ` Ingo Molnar
2008-05-31  6:09         ` Ingo Molnar
2008-05-31 11:46           ` Ilpo Järvinen
2008-05-31 12:18             ` Ilpo Järvinen
2008-05-31 12:54               ` Ingo Molnar
2008-05-31 12:58                 ` Ilpo Järvinen
2008-05-31 16:35                   ` Ingo Molnar
2008-05-31 22:46                     ` Patrick McManus
2008-06-01  5:51                       ` Ilpo Järvinen
2008-06-01  6:04                       ` Eric Dumazet
2008-06-02  9:23                         ` Ingo Molnar
2008-06-03  9:40                     ` [fixed] [patch] " Ingo Molnar
2008-06-03 14:41                       ` Patrick McManus
2008-06-03 21:46                       ` Ilpo Järvinen
2008-06-03 22:01                         ` Ilpo Järvinen
2008-06-03 22:03                           ` David Miller
2008-06-03 22:10                             ` Ilpo Järvinen
2008-06-03 23:22                             ` Ilpo Järvinen
2008-06-03 23:54                               ` Joe Perches
2008-06-04  6:25                                 ` Ilpo Järvinen
2008-06-04  2:54                               ` Patrick McManus
2008-06-04  6:42                                 ` Ilpo Järvinen
2008-06-05 14:22                               ` Ingo Molnar
2008-06-05 18:00                                 ` Ilpo Järvinen
2008-06-05 21:13                                   ` Ilpo Järvinen
2008-06-05 23:29                                     ` Patrick McManus
2008-06-06 10:03                                       ` Ilpo Järvinen
2008-06-06 17:11                                         ` Patrick McManus
2008-06-06 17:33                                           ` Ingo Molnar
2008-06-06 18:19                                             ` Ilpo Järvinen
2008-06-06 18:39                                               ` Ingo Molnar
2008-06-06 19:49                                                 ` Ilpo Järvinen
2008-06-06 20:08                                                 ` Patrick McManus
2008-06-06 21:12                                                   ` Ilpo Järvinen
2008-06-06 21:23                                                     ` Arjan van de Ven
2008-06-06 21:28                                                       ` Ilpo Järvinen
2008-06-10 22:49                                                   ` David Miller
2008-06-06 18:25                                           ` Ilpo Järvinen
2008-06-10 22:32                               ` David Miller
2008-06-11 13:10                                 ` Patrick McManus
2008-06-11 15:13                                 ` Ilpo Järvinen
2008-06-04  7:23                         ` Ingo Molnar
2008-06-04 18:24                           ` David Miller
2008-06-04 20:56                             ` Ilpo Järvinen
2008-06-04 21:55                               ` David Miller
  -- strict thread matches above, loose matches on Subject: below --
2008-05-30 16:31 Ray Lee
2008-05-30 21:11 ` Ilpo Järvinen
2008-05-31  6:03   ` Evgeniy Polyakov
2008-05-31 10:05     ` Ilpo Järvinen
2008-06-02  6:19       ` Herbert Xu
2008-06-02 11:53         ` Ilpo Järvinen
2008-06-02 14:08           ` Herbert Xu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20080529084524.GA24892@elte.hu \
    --to=mingo@elte.hu \
    --cc=akpm@linux-foundation.org \
    --cc=davem@davemloft.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=rjw@sisk.pl \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).