[PATCH net-next] net/sched: get rid of qdisc->padded

[PATCH net-next] net/sched: get rid of qdisc->padded

[Eric Dumazet]

From: Eric Dumazet <edumazet@google.com>

kmalloc() of sufficiently big portion of memory is cache-aligned
in regular conditions. If some debugging options are used,
there is no reason qdisc structures would need 64-byte alignment
if most other kernel structures are not aligned.

This get rid of QDISC_ALIGN and QDISC_ALIGNTO.

Addition of privdata field will help implementing
the reverse of qdisc_priv() and documents where
the private data is.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: Allen Pais <allen.lkml@gmail.com>
---
 include/net/pkt_sched.h   |  5 +----
 include/net/sch_generic.h |  5 ++++-
 net/sched/sch_generic.c   | 23 +++++------------------
 3 files changed, 10 insertions(+), 23 deletions(-)

diff --git a/include/net/pkt_sched.h b/include/net/pkt_sched.h
index ac8c890a2657e35546ec51fe8b8a993a2bd0c91b..4ed32e6b020145afb015c3c07d2ec3a613f1311d 100644
--- a/include/net/pkt_sched.h
+++ b/include/net/pkt_sched.h
@@ -19,12 +19,9 @@ struct qdisc_walker {
 	int	(*fn)(struct Qdisc *, unsigned long cl, struct qdisc_walker *);
 };
 
-#define QDISC_ALIGNTO		64
-#define QDISC_ALIGN(len)	(((len) + QDISC_ALIGNTO-1) & ~(QDISC_ALIGNTO-1))
-
 static inline void *qdisc_priv(struct Qdisc *q)
 {
-	return (char *) q + QDISC_ALIGN(sizeof(struct Qdisc));
+	return &q->privdata;
 }
 
 /* 
diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index 6c762457122fd0091cb0f2bf41bda73babc4ac12..d8fd8676fc724110630904909f64d7789f3a4b47 100644
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -91,7 +91,7 @@ struct Qdisc {
 	struct net_rate_estimator __rcu *rate_est;
 	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
 	struct gnet_stats_queue	__percpu *cpu_qstats;
-	int			padded;
+	int			pad;
 	refcount_t		refcnt;
 
 	/*
@@ -112,6 +112,9 @@ struct Qdisc {
 	/* for NOLOCK qdisc, true if there are no enqueued skbs */
 	bool			empty;
 	struct rcu_head		rcu;
+
+	/* private data */
+	long privdata[] ____cacheline_aligned;
 };
 
 static inline void qdisc_refcount_inc(struct Qdisc *qdisc)
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index 54c417244642a1445c618e6adf0e38b2d4f84565..49eae93d1489dc3513b41c237ca3f572e21ff203 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -802,9 +802,8 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 			  const struct Qdisc_ops *ops,
 			  struct netlink_ext_ack *extack)
 {
-	void *p;
 	struct Qdisc *sch;
-	unsigned int size = QDISC_ALIGN(sizeof(*sch)) + ops->priv_size;
+	unsigned int size = sizeof(*sch) + ops->priv_size;
 	int err = -ENOBUFS;
 	struct net_device *dev;
 
@@ -815,22 +814,10 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 	}
 
 	dev = dev_queue->dev;
-	p = kzalloc_node(size, GFP_KERNEL,
-			 netdev_queue_numa_node_read(dev_queue));
+	sch = kzalloc_node(size, GFP_KERNEL, netdev_queue_numa_node_read(dev_queue));
 
-	if (!p)
+	if (!sch)
 		goto errout;
-	sch = (struct Qdisc *) QDISC_ALIGN((unsigned long) p);
-	/* if we got non aligned memory, ask more and do alignment ourself */
-	if (sch != p) {
-		kfree(p);
-		p = kzalloc_node(size + QDISC_ALIGNTO - 1, GFP_KERNEL,
-				 netdev_queue_numa_node_read(dev_queue));
-		if (!p)
-			goto errout;
-		sch = (struct Qdisc *) QDISC_ALIGN((unsigned long) p);
-		sch->padded = (char *) sch - (char *) p;
-	}
 	__skb_queue_head_init(&sch->gso_skb);
 	__skb_queue_head_init(&sch->skb_bad_txq);
 	qdisc_skb_head_init(&sch->q);
@@ -873,7 +860,7 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 
 	return sch;
 errout1:
-	kfree(p);
+	kfree(sch);
 errout:
 	return ERR_PTR(err);
 }
@@ -941,7 +928,7 @@ void qdisc_free(struct Qdisc *qdisc)
 		free_percpu(qdisc->cpu_qstats);
 	}
 
-	kfree((char *) qdisc - qdisc->padded);
+	kfree(qdisc);
 }
 
 static void qdisc_free_cb(struct rcu_head *head)
-- 
2.28.0.806.g8561365e88-goog

Re: [PATCH net-next] net/sched: get rid of qdisc->padded

[Cong Wang]

On Wed, Oct 7, 2020 at 9:51 AM Eric Dumazet <eric.dumazet@gmail.com> wrote:
>
> From: Eric Dumazet <edumazet@google.com>
>
> kmalloc() of sufficiently big portion of memory is cache-aligned
> in regular conditions. If some debugging options are used,
> there is no reason qdisc structures would need 64-byte alignment
> if most other kernel structures are not aligned.
>
> This get rid of QDISC_ALIGN and QDISC_ALIGNTO.
>
> Addition of privdata field will help implementing
> the reverse of qdisc_priv() and documents where
> the private data is.
>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Cc: Cong Wang <xiyou.wangcong@gmail.com>
> Cc: Allen Pais <allen.lkml@gmail.com>

Acked-by: Cong Wang <xiyou.wangcong@gmail.com>

Thanks.

Re: [PATCH net-next] net/sched: get rid of qdisc->padded

[Jakub Kicinski]

On Wed,  7 Oct 2020 09:51:11 -0700 Eric Dumazet wrote:
> diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
> index 6c762457122fd0091cb0f2bf41bda73babc4ac12..d8fd8676fc724110630904909f64d7789f3a4b47 100644
> --- a/include/net/sch_generic.h
> +++ b/include/net/sch_generic.h
> @@ -91,7 +91,7 @@ struct Qdisc {
>  	struct net_rate_estimator __rcu *rate_est;
>  	struct gnet_stats_basic_cpu __percpu *cpu_bstats;
>  	struct gnet_stats_queue	__percpu *cpu_qstats;
> -	int			padded;
> +	int			pad;
>  	refcount_t		refcnt;
>  
>  	/*

Hi Eric!

Why keep the pad field? the member to lines down is
__cacheline_aligned, so we shouldn't have to manually 
push things out?

        struct gnet_stats_queue __percpu *cpu_qstats;                           
        int                     pad;                                            
        refcount_t              refcnt;                                         
                                                                                
        /*                                                                      
         * For performance sake on SMP, we put highly modified fields at the end                                                                               
         */                                                                     
        struct sk_buff_head     gso_skb ____cacheline_aligned_in_smp;  

Re: [PATCH net-next] net/sched: get rid of qdisc->padded

[Eric Dumazet]

On Fri, Oct 9, 2020 at 2:18 AM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Wed,  7 Oct 2020 09:51:11 -0700 Eric Dumazet wrote:
> > diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
> > index 6c762457122fd0091cb0f2bf41bda73babc4ac12..d8fd8676fc724110630904909f64d7789f3a4b47 100644
> > --- a/include/net/sch_generic.h
> > +++ b/include/net/sch_generic.h
> > @@ -91,7 +91,7 @@ struct Qdisc {
> >       struct net_rate_estimator __rcu *rate_est;
> >       struct gnet_stats_basic_cpu __percpu *cpu_bstats;
> >       struct gnet_stats_queue __percpu *cpu_qstats;
> > -     int                     padded;
> > +     int                     pad;
> >       refcount_t              refcnt;
> >
> >       /*
>
> Hi Eric!
>
> Why keep the pad field? the member to lines down is
> __cacheline_aligned, so we shouldn't have to manually
> push things out?
>
>         struct gnet_stats_queue __percpu *cpu_qstats;
>         int                     pad;

I usually prefer adding explicit fields to tell where the holes are,
for future reuse.
Not many developers are aware of pahole existence :/
I renamed it to pad, I could have used padding or something else.

Re: [PATCH net-next] net/sched: get rid of qdisc->padded

[Jakub Kicinski]

On Fri, 9 Oct 2020 09:35:03 +0200 Eric Dumazet wrote:
> On Fri, Oct 9, 2020 at 2:18 AM Jakub Kicinski <kuba@kernel.org> wrote:
> >
> > On Wed,  7 Oct 2020 09:51:11 -0700 Eric Dumazet wrote:  
> > > diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
> > > index 6c762457122fd0091cb0f2bf41bda73babc4ac12..d8fd8676fc724110630904909f64d7789f3a4b47 100644
> > > --- a/include/net/sch_generic.h
> > > +++ b/include/net/sch_generic.h
> > > @@ -91,7 +91,7 @@ struct Qdisc {
> > >       struct net_rate_estimator __rcu *rate_est;
> > >       struct gnet_stats_basic_cpu __percpu *cpu_bstats;
> > >       struct gnet_stats_queue __percpu *cpu_qstats;
> > > -     int                     padded;
> > > +     int                     pad;
> > >       refcount_t              refcnt;
> > >
> > >       /*  
> >
> > Hi Eric!
> >
> > Why keep the pad field? the member to lines down is
> > __cacheline_aligned, so we shouldn't have to manually
> > push things out?
> >
> >         struct gnet_stats_queue __percpu *cpu_qstats;
> >         int                     pad;  
> 
> I usually prefer adding explicit fields to tell where the holes are,
> for future reuse.
> Not many developers are aware of pahole existence :/
> I renamed it to pad, I could have used padding or something else.

Alright then, applied, thank you!