Cake - FQ_codel the next generation
 help / color / mirror / Atom feed
From: Pete Heist <pete@heistp.net>
To: "Toke Høiland-Jørgensen" <toke@redhat.com>
Cc: Cake List <cake@lists.bufferbloat.net>
Subject: Re: [Cake] lockup with multiple cake instances on 3.16.7
Date: Thu, 31 Jan 2019 08:15:49 +0100	[thread overview]
Message-ID: <9540B582-7B7C-4846-BA40-54419DF109D4@heistp.net> (raw)
In-Reply-To: <87zhrhiwfv.fsf@toke.dk>

Thanks much for the try, only that didn’t fix it yet. I also tried changing two more calls to sch_tree_lock/unlock in cake_change to the maybe versions (recognizing that those may have still been there on purpose), but I still see the lockup. Even commented all the lock calls out to make sure.

I noticed that tc-adv is without xtables support when I compile it (TC_CONFIG_NO_XT:=y) in case that makes any difference. No libxtables-dev package available on this distro.

Pete

> On Jan 30, 2019, at 11:59 PM, Toke Høiland-Jørgensen <toke@redhat.com> wrote:
> 
> Pete Heist <pete@heistp.net> writes:
> 
>> I’m seeing a lockup with cake on 3.16.7 with 64-bit Intel when adding
>> more than 4 cake instances. I’m using cake and tc-adv from head. Does
>> anyone have an idea what this could be?
> 
> Looks like it's a compatibility issue with old kernels. Could you please
> check if the latest commit in the github repo fixes this? :)
> 
> The explanation, copied from that commit:
> 
>> The qdisc stats dumping was changed in 4.8 (by commit
>> edb09eb17ed89eaa82a52dd306beac93e292b485) to be lockless and moving the
>> responsibility for taking the qdisc root lock into the qdisc itself. Cake
>> inherited this from FQ-CoDel, but it leads to deadlocks on old kernels
>> where the core stats dumping code still takes the root qdisc lock.
> 
> 
> -Toke


  reply	other threads:[~2019-01-31  7:15 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-01-30  8:58 Pete Heist
2019-01-30 22:59 ` Toke Høiland-Jørgensen
2019-01-31  7:15   ` Pete Heist [this message]
2019-01-31 14:53     ` Toke Høiland-Jørgensen
2019-01-31 20:25       ` Pete Heist
2019-01-31 23:09         ` Toke Høiland-Jørgensen
2019-01-31 23:18           ` Pete Heist
2019-01-31 23:10         ` Pete Heist
2019-01-31 23:18           ` Toke Høiland-Jørgensen
2019-01-31 23:23             ` Pete Heist
2019-01-31 23:31               ` Toke Høiland-Jørgensen

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

  List information: https://lists.bufferbloat.net/postorius/lists/cake.lists.bufferbloat.net/

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

  git send-email \
    --in-reply-to=9540B582-7B7C-4846-BA40-54419DF109D4@heistp.net \
    --to=pete@heistp.net \
    --cc=cake@lists.bufferbloat.net \
    --cc=toke@redhat.com \
    /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