public inbox for netdev@vger.kernel.org 
 help / color / mirror / Atom feed
From: John Fastabend <john.fastabend@gmail•com>
To: Eric Dumazet <eric.dumazet@gmail•com>
Cc: jhs@mojatatu•com, davem@davemloft•net, brouer@redhat•com,
	xiyou.wangcong@gmail•com, alexei.starovoitov@gmail•com,
	john.r.fastabend@intel•com, netdev@vger•kernel.org
Subject: Re: [net-next PATCH 02/15] net: sched: allow qdiscs to handle locking
Date: Tue, 23 Aug 2016 15:32:22 -0700	[thread overview]
Message-ID: <57BCCEF6.3090405@gmail.com> (raw)
In-Reply-To: <1471986499.14381.56.camel@edumazet-glaptop3.roam.corp.google.com>

On 16-08-23 02:08 PM, Eric Dumazet wrote:
> On Tue, 2016-08-23 at 13:23 -0700, John Fastabend wrote:
>> This patch adds a flag for queueing disciplines to indicate the
>> stack does not need to use the qdisc lock to protect operations.
>> This can be used to build lockless scheduling algorithms and
>> improving performance.
>> 

[...]

>> * Heuristic to force contended enqueues to serialize on a *
>> separate lock before trying to get qdisc main lock. @@ -3898,19
>> +3913,22 @@ static void net_tx_action(struct softirq_action *h)
>> 
>> while (head) { struct Qdisc *q = head; -			spinlock_t *root_lock; +
>> spinlock_t *root_lock = NULL;
>> 
>> head = head->next_sched;
>> 
>> -			root_lock = qdisc_lock(q); -			spin_lock(root_lock); +			if
>> (!(q->flags & TCQ_F_NOLOCK)) { +				root_lock = qdisc_lock(q); +
>> spin_lock(root_lock); +			} /* We need to make sure
>> head->next_sched is read * before clearing __QDISC_STATE_SCHED */ 
>> smp_mb__before_atomic(); clear_bit(__QDISC_STATE_SCHED,
>> &q->state); qdisc_run(q); -			spin_unlock(root_lock); +			if
>> (!(q->flags & TCQ_F_NOLOCK))
> 
> This might be faster to use : if (root_lock) (one less memory read
> and mask)
> 

hmm this actually gets factored out in patch 12 but I'll go ahead
and make this change and then I think it reads a bit better through
the series.

>> +				spin_unlock(root_lock); } } } diff --git
>> a/net/sched/sch_generic.c b/net/sched/sch_generic.c index
>> e305a55..af32418 100644 --- a/net/sched/sch_generic.c +++
>> b/net/sched/sch_generic.c @@ -170,7 +170,8 @@ int
>> sch_direct_xmit(struct sk_buff *skb, struct Qdisc *q, int ret =
>> NETDEV_TX_BUSY;
>> 
>> /* And release qdisc */ -	spin_unlock(root_lock); +	if (!(q->flags
>> & TCQ_F_NOLOCK)) +		spin_unlock(root_lock);
> 
> You might use the same trick, if root_lock is NULL for lockless
> qdisc.

So what I just did is pass NULL into sch_direct_xmit() for root_lock
when the qdisc is lockless. This replaces the qdisc flags checks in this
call to checking root_lock.

Seems like a nice cleanup/optimization. I'll wait a bit and then push
it in v2 after giving folks a day or two to review this set.

  reply	other threads:[~2016-08-23 22:33 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-08-23 20:22 [net-next PATCH 00/15] support lockless qdisc John Fastabend
2016-08-23 20:22 ` [net-next PATCH 01/15] net: sched: cleanup qdisc_run and __qdisc_run semantics John Fastabend
2016-08-23 20:38   ` Eric Dumazet
2016-08-23 20:23 ` [net-next PATCH 02/15] net: sched: allow qdiscs to handle locking John Fastabend
2016-08-23 21:08   ` Eric Dumazet
2016-08-23 22:32     ` John Fastabend [this message]
2016-08-23 20:23 ` [net-next PATCH 03/15] net: sched: remove remaining uses for qdisc_qlen in xmit path John Fastabend
2016-08-23 21:10   ` Eric Dumazet
2016-08-23 20:24 ` [net-next PATCH 04/15] net: sched: provide per cpu qstat helpers John Fastabend
2016-08-23 23:25   ` Eric Dumazet
2016-08-23 23:50     ` John Fastabend
2016-08-23 20:24 ` [net-next PATCH 05/15] net: sched: a dflt qdisc may be used with per cpu stats John Fastabend
2016-08-24 16:29   ` Eric Dumazet
2016-08-24 16:41   ` Eric Dumazet
2016-08-24 17:13     ` John Fastabend
2016-08-24 17:26       ` Eric Dumazet
2016-08-24 17:50         ` John Fastabend
2016-08-24 19:08           ` Eric Dumazet
2016-08-23 20:25 ` [net-next PATCH 06/15] net: sched: per cpu gso handlers John Fastabend
2016-08-23 20:25 ` [net-next PATCH 07/15] net: sched: drop qdisc_reset from dev_graft_qdisc John Fastabend
2016-08-23 20:26 ` [net-next PATCH 08/15] net: sched: support qdisc_reset on NOLOCK qdisc John Fastabend
2016-08-23 20:26 ` [net-next PATCH 09/15] net: sched: support skb_bad_tx with lockless qdisc John Fastabend
2016-08-23 20:26 ` [net-next PATCH 10/15] net: sched: qdisc_qlen for per cpu logic John Fastabend
2016-08-23 20:27 ` [net-next PATCH 11/15] net: sched: helper to sum qlen John Fastabend
2016-08-23 20:27 ` [net-next PATCH 12/15] net: sched: lockless support for netif_schedule John Fastabend
2016-09-07 14:50   ` John Fastabend
2016-08-23 20:28 ` [net-next PATCH 13/15] net: sched: add support for TCQ_F_NOLOCK subqueues to sch_mq John Fastabend
2016-08-23 20:28 ` [net-next PATCH 14/15] net: sched: add support for TCQ_F_NOLOCK subqueues to sch_mqprio John Fastabend
2016-08-23 20:28 ` [net-next PATCH 15/15] net: sched: pfifo_fast use skb_array John Fastabend
2016-09-01  8:26   ` [lkp] [net] c4c75f963d: inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage kernel test robot

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=57BCCEF6.3090405@gmail.com \
    --to=john.fastabend@gmail$(echo .)com \
    --cc=alexei.starovoitov@gmail$(echo .)com \
    --cc=brouer@redhat$(echo .)com \
    --cc=davem@davemloft$(echo .)net \
    --cc=eric.dumazet@gmail$(echo .)com \
    --cc=jhs@mojatatu$(echo .)com \
    --cc=john.r.fastabend@intel$(echo .)com \
    --cc=netdev@vger$(echo .)kernel.org \
    --cc=xiyou.wangcong@gmail$(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