public inbox for netdev@vger.kernel.org 
 help / color / mirror / Atom feed
From: Hannes Frederic Sowa <hannes@stressinduktion•org>
To: "Michael S. Tsirkin" <mst@redhat•com>, Jason Wang <jasowang@redhat•com>
Cc: Paolo Abeni <pabeni@redhat•com>,
	netdev@vger•kernel.org, "David S. Miller" <davem@davemloft•net>,
	"Eric W. Biederman" <ebiederm@xmission•com>,
	Greg Kurz <gkurz@linux•vnet.ibm.com>
Subject: Re: [PATCH RFC 2/2] tun: don't set a default qdisc
Date: Thu, 14 Apr 2016 12:09:57 +0200	[thread overview]
Message-ID: <570F6C75.8010001@stressinduktion.org> (raw)
In-Reply-To: <20160414125822-mutt-send-email-mst@redhat.com>

On 14.04.2016 12:01, Michael S. Tsirkin wrote:
> On Thu, Apr 14, 2016 at 05:21:40PM +0800, Jason Wang wrote:
>>
>>
>> On 04/14/2016 05:10 PM, Michael S. Tsirkin wrote:
>>> On Thu, Apr 14, 2016 at 05:07:50PM +0800, Jason Wang wrote:
>>>>
>>>> On 04/14/2016 05:05 PM, Michael S. Tsirkin wrote:
>>>>> On Thu, Apr 14, 2016 at 02:49:28PM +0800, Jason Wang wrote:
>>>>>>>
>>>>>>> On 04/13/2016 06:26 PM, Michael S. Tsirkin wrote:
>>>>>>>>> On Wed, Apr 13, 2016 at 11:04:47AM +0200, Paolo Abeni wrote:
>>>>>>>>>>> This patch disables the default qdisc by explicitly setting the
>>>>>>>>>>> IFF_NO_QUEUE private flag so that now the tun xmit path do not
>>>>>>>>>>> require any lock by default.
>>>>>>>>>>>
>>>>>>>>>>> The default qdisc was first removed as a side effect of commit
>>>>>>>>>>> f84bb1eac027 ("net: fix IFF_NO_QUEUE for drivers using alloc_netdev")
>>>>>>>>>>> and recently restored with commit 016adb7260f4 ("tuntap: restore default qdisc")
>>>>>>>>>>>
>>>>>>>>>>> Signed-off-by: Paolo Abeni <pabeni@redhat•com>
>>>>>>>>> I wonder about this back and forth.
>>>>>>>>> Jason, do you see a workload where the default qdisc
>>>>>>>>> is preferable?
>>>>>>> I don't know, but we used to behave like this so we'd better keep it.
>>>>>>>
>>>>>>> An interesting thing is I vaguely remember that you have some concern
>>>>>>> when I propose IFF_NO_QUEUE for macvtap[1] :)
>>>>>>>
>>>>>>> [1] https://lkml.org/lkml/2015/8/24/147
>>>>> It's the same concern - that we aren't fully addressing
>>>>> the problem, so if user configures a qdisc, we are back to square 1.
>>>>> It's especially annoying that IIUC in this setup, if one
>>>>> does configured a non default qdisc, there's no way to go back.
>>>>> It doesn't necessarily mean we must not do it as an intermediate step,
>>>>> though.
>>>>>
>>>>>>> I think this could be done by management or more safe by introducing a
>>>>>>> new tun flag (TUN_NO_QUEUE).
>>>>> What exactly does this flag do/mean?
>>>> It means we don't need qdisc for this tuntap, so we can set IFF_NO_QUEUE
>>>> flag.
>>> But what does it mean for the user? When to set it and when not to set
>>> it?
>>
>> It was used for user that don't want qdisc (e.g for the user that only
>> cares about performance).
>
> However
> - for tun, how is a fifo qdisc functionally different?

pfifo wouldn't matter match, but the pfifo_qdisc classifies packets into 
different bands and does alter the way ip packets are transmitted 
outside. With the LLTX change I don't see how in any case a queue should 
build up, so I don't really see this as a problem.

The only reason why this change could do harm is scripting and API 
compatibility, e.g. tools expect after creation a qdisc to be present to 
modify its behavior.

> - it doesn't prevent configuring a qdisc, does it?

Correct, you can still add qdiscs later on and switch back to no_queue, 
also. We are just talking about changing the default.

Bye,
Hannes

  reply	other threads:[~2016-04-14 10:10 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-04-13  9:04 [PATCH RFC 0/2] tun: lockless xmit Paolo Abeni
2016-04-13  9:04 ` [PATCH RFC 1/2] tun: don't require serialization lock on tx Paolo Abeni
2016-04-13  9:41   ` Michael S. Tsirkin
2016-04-13  9:48     ` Hannes Frederic Sowa
2016-04-13 12:57       ` Michael S. Tsirkin
2016-04-13 13:27         ` Eric Dumazet
2016-04-13 13:54           ` Michael S. Tsirkin
2016-04-13 14:39             ` Eric Dumazet
2016-04-13 12:52   ` Eric Dumazet
2016-04-13 14:26   ` Sergei Shtylyov
2016-04-14  6:50   ` Jason Wang
2016-04-14 10:27   ` Hannes Frederic Sowa
2016-04-13  9:04 ` [PATCH RFC 2/2] tun: don't set a default qdisc Paolo Abeni
2016-04-13 10:26   ` Michael S. Tsirkin
2016-04-13 15:22     ` David Miller
2016-04-14  6:49     ` Jason Wang
2016-04-14  9:05       ` Michael S. Tsirkin
2016-04-14  9:07         ` Jason Wang
2016-04-14  9:10           ` Michael S. Tsirkin
2016-04-14  9:21             ` Jason Wang
2016-04-14 10:01               ` Michael S. Tsirkin
2016-04-14 10:09                 ` Hannes Frederic Sowa [this message]
2016-04-13 11:08 ` [PATCH RFC 0/2] tun: lockless xmit Michael S. Tsirkin
2016-04-13 12:50   ` Eric Dumazet
2016-04-13 12:56     ` Michael S. Tsirkin
2016-04-13 13:09       ` Eric Dumazet
2016-04-13 13:17         ` Michael S. Tsirkin
2016-04-13 13:43           ` Eric Dumazet
2016-04-13 16:42             ` Eric Dumazet

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=570F6C75.8010001@stressinduktion.org \
    --to=hannes@stressinduktion$(echo .)org \
    --cc=davem@davemloft$(echo .)net \
    --cc=ebiederm@xmission$(echo .)com \
    --cc=gkurz@linux$(echo .)vnet.ibm.com \
    --cc=jasowang@redhat$(echo .)com \
    --cc=mst@redhat$(echo .)com \
    --cc=netdev@vger$(echo .)kernel.org \
    --cc=pabeni@redhat$(echo .)com \
    /path/to/YOUR_REPLY

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