From: David Miller <davem@davemloft.net>
To: phil@nwl.cc
Cc: fw@strlen.de, netdev@vger.kernel.org, brouer@redhat.com,
cwang@twopensource.com, eric.dumazet@gmail.com
Subject: Re: [net-next PATCH 1/3] net: make default tx_queue_len configurable
Date: Wed, 29 Jul 2015 14:37:31 -0700 (PDT) [thread overview]
Message-ID: <20150729.143731.834219016906219771.davem@davemloft.net> (raw)
In-Reply-To: <20150729213428.GC15554@orbit.nwl.cc>
From: Phil Sutter <phil@nwl.cc>
Date: Wed, 29 Jul 2015 23:34:28 +0200
> On Wed, Jul 29, 2015 at 11:06:18PM +0200, Florian Westphal wrote:
>> Phil Sutter <phil@nwl.cc> wrote:
> [...]
>> > +config DEFAULT_TX_QUEUE_LEN
>> > + prompt "Default TX queue length (in packets)" if EXPERT
>> > + int
>> > + default 1000 # Ethernet wants good queues
>> > + help
>> > + Set the default value of tx_queue_len for newly created network
>> > + interfaces. It is used by queueing disciplines to determine how many
>> > + packets to keep in backlog before starting to drop new ones.
>> > +
>> > + The default value of 1000 packets is there for a very long time and
>> > + in combination with GSO way too big.
>> > +
>>
>> I can't see how this could be used in a meaningful way.
>>
>> No distro is going to touch this.
>>
>> I don't think sysctl value would help either.
>
> I just didn't want to introduce yet another magic value assignment. It's
> merely a #define with a little flexibility and a subtle note that the
> default should be changed attached.
Like others have mentioned, fix the _REAL_ issue.
Which is that there are devices (virtual or whatever) which don't want
a qdisc attached no matter what. Flag those devices as such and
adjust the qdisc attachment logic to check that new flag.
Anything is better than hacking the queue len.
next prev parent reply other threads:[~2015-07-29 21:37 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-07-29 20:51 [net-next PATCH 0/3] Backwards-compatible noqueue in virtual interfaces Phil Sutter
2015-07-29 20:51 ` [net-next PATCH 1/3] net: make default tx_queue_len configurable Phil Sutter
2015-07-29 20:51 ` [net-next PATCH 2/3] net: sched: set tx_queue_len to default when changing noqueue device's qdisc Phil Sutter
2015-07-29 20:51 ` [net-next PATCH 3/3] veth: don't assign a qdisc to veth Phil Sutter
2015-07-29 21:10 ` [net-next PATCH 2/3] net: sched: set tx_queue_len to default when changing noqueue device's qdisc Florian Westphal
2015-07-29 21:10 ` Eric Dumazet
2015-07-29 22:08 ` Phil Sutter
2015-07-29 21:06 ` [net-next PATCH 1/3] net: make default tx_queue_len configurable Florian Westphal
2015-07-29 21:34 ` Phil Sutter
2015-07-29 21:37 ` David Miller [this message]
2015-08-11 15:48 ` Jesper Dangaard Brouer
2015-08-11 16:23 ` Phil Sutter
2015-08-12 1:13 ` Alexei Starovoitov
2015-08-12 14:55 ` Eric Dumazet
2015-08-13 1:13 ` Phil Sutter
2015-08-13 13:10 ` Jesper Dangaard Brouer
2015-08-13 15:06 ` Phil Sutter
2015-07-29 23:10 ` [net-next PATCH 0/3] Backwards-compatible noqueue in virtual interfaces Cong Wang
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=20150729.143731.834219016906219771.davem@davemloft.net \
--to=davem@davemloft.net \
--cc=brouer@redhat.com \
--cc=cwang@twopensource.com \
--cc=eric.dumazet@gmail.com \
--cc=fw@strlen.de \
--cc=netdev@vger.kernel.org \
--cc=phil@nwl.cc \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.