Message-ID: <CAA93jw76hQYZ7CT7hEwvdFvt2kC12XHwOxeEGiDoCei62U7h+A@mail.gmail.com>
Date:   Mon, 8 Apr 2019 07:52:51 +0200
From:   Dave Taht <dave.taht@...il.com>
To:     Gautam Ramakrishnan <gautamramk@...il.com>
Cc:     Toke Høiland-Jørgensen <toke@...hat.com>,
        Jamal Hadi Salim <jhs@...atatu.com>,
        "David S. Miller" <davem@...emloft.net>,
        Linux Kernel Network Developers <netdev@...r.kernel.org>,
        "Mohit P . Tahiliani" <tahiliani@...k.edu.in>,
        "Sachin D . Patil" <sdp.sachin@...il.com>,
        Mohit Bhasi <mohitbhasi1998@...il.com>,
        "V . Saicharan" <vsaicharan1998@...il.com>,
        Leslie Monis <lesliemonis@...il.com>
Subject: Re: [RFC net-next 2/2] net: sched: fq_pie: Flow Queue PIE AQM

On Mon, Apr 8, 2019 at 7:37 AM Dave Taht <dave.taht@...il.com> wrote:
>
> On Mon, Apr 8, 2019 at 7:31 AM Gautam Ramakrishnan <gautamramk@...il.com> wrote:
> >
> > I was trying to refactor the code and I ran into some issues.
> > 1. I moved some of the parameters such as flows_cnt into a new struct
> > called fq_pie_params, instead of keeping them in fq_pie_sched_data.
> > Should I move those parameters back into fq_pie_sched_data?
> > 2. fq_codel maintains the backlog variable as a list in the
> > fq_codel_sched_data, whereas I maintain a backlog in the struct
> > fq_pie_flow. What approach should I follow?
>
> Hmm. I had made some efforts to speed up fq_codel here:
> https://github.com/dtaht/fq_codel_fast
>
> based on ideas here:
> https://lists.bufferbloat.net/pipermail/codel/2018-August/002367.html
>
> Notably I wanted to get rid of the non O(1) bulk dropper search and
> keep the flow backlog closer to the flow stats.
> Needed oprofiling to see if it helped.

I'm sorry to conflate your fq_pie upstreaming attempt with my own long
out of tree improvements for fq_codel. I'd been meaning to upstream
multiple fixes from there for ages, but needed to profile them on mips
first.

Are you in a hurry? Because: hacking in SCE into pie and fq_pie, etc,
is WAY more interesting than mainlining stuff right now. :)

Our updated "some congestion experienced"  internet draft with
suggestions for pie and red is here:

https://github.com/dtaht/bufferbloat-rfcs/blob/master/sce/draft-morton-taht-tsvwg-sce.txt#L272

and running code for sce is in that fq_codel_fast repo and in the out
of tree sch_cake also:

https://github.com/dtaht/sch_cake/commit/755ba4cda56bec45d20f7928f1442fdf83c12573

> > 3. Would maintaining a per flow pie_stats be useful in the future? I
> > do not use the per flow stats anywhere in the code.
>
> In general I have not found per flow stats useful or accessible and it
> was one of the things I eliminated in fq_codel_fast
> > On Tue, Apr 2, 2019 at 10:55 PM Toke Høiland-Jørgensen <toke@...hat.com> wrote:
> > >
> > > Gautam Ramakrishnan <gautamramk@...il.com> writes:
> > >
> > > > Hello, thanks for the feedback
> > > >
> > > > On Tue, Apr 2, 2019 at 4:19 PM Toke Høiland-Jørgensen <toke@...hat.com> wrote:
> > > >>
> > > >> Some suggestions below to make fq_pie and fq_codel more similar (ref. my
> > > >> previous email).
> > > >>
> > > >> Also, a few unrelated nits.
> > > >>
> > > >> > From: Mohit P. Tahiliani <tahiliani@...k.edu.in>
> > > >> >
> > > >> > FQ-PIE incorporates fair/flow queuing in which every queue
> > > >> > is managed by an instance of PIE queueing discipline.
> > > >> > The algorithm provides good control over the queueing delay
> > > >> > while at the same time, ensures fairness across various
> > > >> > flows sharing the same link.
> > > >> >
> > > >> > Principles:
> > > >> >   - Packets are classified stochastically on flows.
> > > >> >   - Each flow has a PIE managed queue.
> > > >> >   - Flows are linked onto two (Round Robin) lists,
> > > >> >     so that new flows have priority on old ones.
> > > >> >   - For a given flow, packets are dropped on arrival according
> > > >> >     to a drop probability.
> > > >> >   - Tail drops only.
> > > >>
> > > >> Why tail drops only?
> > > >
> > > > I had mentioned this because packets are dropped only at enqueue.
> > >
> > > Yup, realise that; was just wondering why you went with this design
> > > instead of doing the head drop that fq_codel does?
> > >
> > > >>
> > > >> > Usage:
> > > >> > tc qdisc ... fq_pie [ limit PACKETS ] [ flows NUMBER ]
> > > >> >                     [ alpha NUMBER ] [ beta NUMBER ]
> > > >> >                     [ target TIME us ] [ tupdate TIME us ]
> > > >> >                   [ bytemode ] [ quantum BYTES ]
> > > >> >                   [ ecn | ecn_prob PERCENTAGE ]
> > > >> >
> > > >> > defaults: 1024 flows, 10240 packets limit, quantum : device MTU
> > > >> >            target: 15ms
> > > >> >            tupdate: 15ms
> > > >> >            alpha: 2 (on a scale of 0 to 16)
> > > >> >            beta: 20 (on a scale of 0 to 16)
> > > >> >            ecn: false
> > > >> >            ecn_prob: 10%
> > > >> >
> > > >> > Signed-off-by: Mohit P. Tahiliani <tahiliani@...k.edu.in>
> > > >> > Signed-off-by: Sachin D. Patil <sdp.sachin@...il.com>
> > > >> > Signed-off-by: Mohit Bhasi <mohitbhasi1998@...il.com>
> > > >> > Signed-off-by: V. Saicharan <vsaicharan1998@...il.com>
> > > >> > Signed-off-by: Leslie Monis <lesliemonis@...il.com>
> > > >> > Signed-off-by: Gautam Ramakrishnan <gautamramk@...il.com>
> > > >> > Cc: Dave Taht <dave.taht@...il.com>
> > > >> > ---
> > > >> >  include/uapi/linux/pkt_sched.h |  28 ++
> > > >> >  net/sched/Kconfig              |  14 +-
> > > >> >  net/sched/Makefile             |   1 +
> > > >> >  net/sched/sch_fq_pie.c         | 485 +++++++++++++++++++++++++++++++++
> > > >> >  4 files changed, 527 insertions(+), 1 deletion(-)
> > > >> >  create mode 100644 net/sched/sch_fq_pie.c
> > > >> >
> > > >> > diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
> > > >> > index 7ee74c3474bf..005413bd09ee 100644
> > > >> > --- a/include/uapi/linux/pkt_sched.h
> > > >> > +++ b/include/uapi/linux/pkt_sched.h
> > > >> > @@ -964,6 +964,34 @@ struct tc_pie_xstats {
> > > >> >       __u32 ecn_mark;         /* packets marked with ecn*/
> > > >> >  };
> > > >> >
> > > >> > +/* FQ PIE */
> > > >> > +enum {
> > > >> > +     TCA_FQ_PIE_UNSPEC,
> > > >> > +     TCA_FQ_PIE_TARGET,
> > > >> > +     TCA_FQ_PIE_LIMIT,
> > > >> > +     TCA_FQ_PIE_TUPDATE,
> > > >> > +     TCA_FQ_PIE_ALPHA,
> > > >> > +     TCA_FQ_PIE_BETA,
> > > >> > +     TCA_FQ_PIE_ECN,
> > > >> > +     TCA_FQ_PIE_QUANTUM,
> > > >> > +     TCA_FQ_PIE_BYTEMODE,
> > > >> > +     TCA_FQ_PIE_FLOWS,
> > > >> > +     TCA_FQ_PIE_ECN_PROB,
> > > >> > +     __TCA_FQ_PIE_MAX
> > > >> > +};
> > > >> > +#define TCA_FQ_PIE_MAX   (__TCA_FQ_PIE_MAX - 1)
> > > >> > +
> > > >> > +struct tc_fq_pie_xstats {
> > > >> > +     __u32 packets_in;       /* total number of packets enqueued */
> > > >> > +     __u32 dropped;          /* packets dropped due to fq_pie_action */
> > > >> > +     __u32 overlimit;        /* dropped due to lack of space in queue */
> > > >> > +     __u32 ecn_mark;         /* packets marked with ecn*/
> > > >> > +     __u32 new_flow_count;   /* number of time packets
> > > >> > +                                created a 'new flow' */
> > > >> > +     __u32 new_flows_len;    /* count of flows in new list */
> > > >> > +     __u32 old_flows_len;    /* count of flows in old list */
> > > >> > +};
> > > >> > +
> > > >> >  /* CBS */
> > > >> >  struct tc_cbs_qopt {
> > > >> >       __u8 offload;
> > > >> > diff --git a/net/sched/Kconfig b/net/sched/Kconfig
> > > >> > index 5c02ad97ef23..49f4dd9894a0 100644
> > > >> > --- a/net/sched/Kconfig
> > > >> > +++ b/net/sched/Kconfig
> > > >> > @@ -358,13 +358,25 @@ config NET_SCH_PIE
> > > >> >       help
> > > >> >         Say Y here if you want to use the Proportional Integral controller
> > > >> >         Enhanced scheduler packet scheduling algorithm.
> > > >> > -       For more information, please see https://tools.ietf.org/html/rfc8033
> > > >> > +       For more information, please see
> > > >> > +       http://tools.ietf.org/html/draft-pan-tsvwg-pie-00
> > > >> >
> > > >> >         To compile this driver as a module, choose M here: the module
> > > >> >         will be called sch_pie.
> > > >> >
> > > >> >         If unsure, say N.
> > > >> >
> > > >> > +config NET_SCH_FQ_PIE
> > > >> > +     tristate "Flow Queue Proportional Integral controller Enhanced (FQ-PIE) scheduler"
> > > >> > +     help
> > > >> > +       Say Y here if you want to use the Flow Queue Proportional Integral controller
> > > >> > +       Enhanced scheduler packet scheduling algorithm.
> > > >> > +
> > > >> > +       To compile this driver as a module, choose M here: the module
> > > >> > +       will be called sch_fq_pie.
> > > >> > +
> > > >> > +       If unsure, say N.
> > > >> > +
> > > >> >  config NET_SCH_INGRESS
> > > >> >       tristate "Ingress/classifier-action Qdisc"
> > > >> >       depends on NET_CLS_ACT
> > > >> > diff --git a/net/sched/Makefile b/net/sched/Makefile
> > > >> > index 8a40431d7b5c..fdcd3f7b2fb2 100644
> > > >> > --- a/net/sched/Makefile
> > > >> > +++ b/net/sched/Makefile
> > > >> > @@ -55,6 +55,7 @@ obj-$(CONFIG_NET_SCH_CAKE)  += sch_cake.o
> > > >> >  obj-$(CONFIG_NET_SCH_FQ)     += sch_fq.o
> > > >> >  obj-$(CONFIG_NET_SCH_HHF)    += sch_hhf.o
> > > >> >  obj-$(CONFIG_NET_SCH_PIE)    += sch_pie.o
> > > >> > +obj-$(CONFIG_NET_SCH_FQ_PIE)    += sch_fq_pie.o
> > > >> >  obj-$(CONFIG_NET_SCH_CBS)    += sch_cbs.o
> > > >> >  obj-$(CONFIG_NET_SCH_ETF)    += sch_etf.o
> > > >> >  obj-$(CONFIG_NET_SCH_TAPRIO) += sch_taprio.o
> > > >> > diff --git a/net/sched/sch_fq_pie.c b/net/sched/sch_fq_pie.c
> > > >> > new file mode 100644
> > > >> > index 000000000000..4ccefa0bc7f0
> > > >> > --- /dev/null
> > > >> > +++ b/net/sched/sch_fq_pie.c
> > > >> > @@ -0,0 +1,485 @@
> > > >> > +/*
> > > >> > + * net/sched/sch_fq_pie.c
> > > >> > + *
> > > >> > + * This program is free software; you can redistribute it and/or
> > > >> > + * modify it under the terms of the GNU General Public License
> > > >> > + * as published by the Free Software Foundation; either version 2
> > > >> > + * of the License.
> > > >> > + *
> > > >> > + * This program is distributed in the hope that it will be useful,
> > > >> > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> > > >> > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> > > >> > + * GNU General Public License for more details.
> > > >>
> > > >> Lose the license boilerplate and replace it with a SPDX header line.
> > > >
> > > > I shall do that.
> > > >
> > > >>
> > > >> > + * Author: Mohit P. Tahiliani <tahiliani@...k.edu.in>
> > > >> > + * Author: Sachin D. Patil <sdp.sachin@...il.com>
> > > >> > + * Author: Mohit Bhasi <mohitbhasi1998@...il.com>
> > > >> > + * Author: V Saicharan <vsaicharan1998@...il.com>
> > > >> > + * Author: Leslie Monis <lesliemonis@...il.com>
> > > >> > + * Author: Gautam Ramakrishnan <gautamramk@...il.com>
> > > >> > + *
> > > >> > + * References:
> > > >> > + * RFC 8033: https://tools.ietf.org/html/rfc8033
> > > >> > + */
> > > >> > +
> > > >> > +#include <linux/jhash.h>
> > > >> > +#include <linux/vmalloc.h>
> > > >> > +#include <net/pie.h>
> > > >> > +
> > > >> > +struct fq_pie_params {
> > > >> > +     struct pie_params p_params;
> > > >> > +     u32 ecn_prob;
> > > >> > +     u32 flows_cnt;
> > > >> > +};
> > > >> > +
> > > >> > +struct fq_pie_stats {
> > > >> > +     u32 packets_in;         /* total number of packets enqueued */
> > > >> > +     u32 dropped;            /* packets dropped due to fq_pie action */
> > > >> > +     u32 overlimit;          /* dropped due to lack of space in queue */
> > > >> > +     u32 ecn_mark;           /* packets marked with ECN */
> > > >> > +     u32 new_flow_count;     /* number of time packets created a new flow */
> > > >> > +};
> > > >> > +
> > > >> > +struct fq_pie_flow {
> > > >> > +     s32 deficit;            /* number of credits remaining for the flow */
> > > >> > +     u32 backlog;            /* size of data in the flow */
> > > >> > +     u32 qlen;               /* number of packets in the flow */
> > > >> > +     struct sk_buff *head;
> > > >> > +     struct sk_buff *tail;
> > > >> > +     struct list_head flowchain;
> > > >> > +     struct pie_vars vars;   /* pie vars for the flow */
> > > >> > +     struct pie_stats stats; /* pie stats for the flow */
> > > >> > +};
> > > >> > +
> > > >> > +struct fq_pie_sched_data {
> > > >> > +     u32 quantum;            /* number of credits in deficit round robin */
> > > >> > +     struct fq_pie_flow *flows;
> > > >> > +     struct Qdisc *sch;
> > > >> > +     struct fq_pie_params params;
> > > >> > +     struct fq_pie_stats stats;
> > > >> > +     struct list_head old_flows;
> > > >> > +     struct list_head new_flows;
> > > >> > +     struct timer_list adapt_timer;
> > > >> > +};
> > > >>
> > > >> The flow and sched_data structs have quite a bit in common with those in
> > > >> fq_codel; but the members are in a different order.
> > > >>
> > > >> > +static void fq_pie_params_init(struct fq_pie_params *params)
> > > >> > +{
> > > >> > +     pie_params_init(&params->p_params);
> > > >> > +     params->ecn_prob = 10;
> > > >> > +     params->flows_cnt = 1024;
> > > >> > +}
> > > >> > +
> > > >> > +static inline void flow_queue_add(struct fq_pie_flow *flow,
> > > >> > +                               struct sk_buff *skb)
> > > >> > +{
> > > >> > +     if (!flow->head)
> > > >> > +             flow->head = skb;
> > > >> > +     else
> > > >> > +             flow->tail->next = skb;
> > > >> > +     flow->tail = skb;
> > > >> > +     skb->next = NULL;
> > > >> > +}
> > > >> > +
> > > >> > +static int fq_pie_qdisc_enqueue(struct sk_buff *skb, struct Qdisc *sch,
> > > >> > +                             struct sk_buff **to_free)
> > > >> > +{
> > > >> > +     struct fq_pie_sched_data *q = qdisc_priv(sch);
> > > >> > +     struct fq_pie_flow *sel_flow;
> > > >> > +     u32 pkt_len;
> > > >> > +     u32 idx;
> > > >> > +     u8 enqueue = false;
> > > >> > +
> > > >> > +     /* Classifies packet into corresponding flow */
> > > >> > +     idx = reciprocal_scale(skb_get_hash(skb), q->params.flows_cnt);
> > > >> > +     sel_flow = &q->flows[idx];
> > > >>
> > > >> This is missing the ability to override the classification from tc
> > > >> filters. See fq_codel_classify().
> > > >
> > > > I wanted to keep it simple initially. Shall add that.
> > > >
> > > >>
> > > >> > +
> > > >> > +     /* Checks if the qdisc is full */
> > > >> > +     if (unlikely(qdisc_qlen(sch) >= sch->limit)) {
> > > >> > +             q->stats.overlimit++;
> > > >> > +             sel_flow->stats.overlimit++;
> > > >> > +             goto out;
> > > >> > +     }
> > > >>
> > > >> The memory_limit checks in fq_codel have turned out to be quite useful
> > > >> on constrained systems. I'd suggest adding them here as well.
> > > >
> > > > I shall add that too.
> > > >
> > > >>
> > > >> > +
> > > >> > +     if (!drop_early(sch, sel_flow->backlog, skb->len, &sel_flow->vars,
> > > >> > +                     &q->params.p_params)) {
> > > >> > +             enqueue = true;
> > > >> > +     } else if (q->params.p_params.ecn &&
> > > >> > +                sel_flow->vars.prob <=
> > > >> > +                (MAX_PROB / 100) * q->params.ecn_prob &&
> > > >> > +                INET_ECN_set_ce(skb)) {
> > > >> > +             /* If packet is ecn capable, mark it if drop probability
> > > >> > +              * is lower than the parameter ecn_prob, else drop it.
> > > >> > +              */
> > > >> > +             q->stats.ecn_mark++;
> > > >> > +             sel_flow->stats.ecn_mark++;
> > > >> > +             enqueue = true;
> > > >> > +     }
> > > >> > +     if (enqueue) {
> > > >> > +             pkt_len = qdisc_pkt_len(skb);
> > > >> > +             q->stats.packets_in++;
> > > >> > +             sch->qstats.backlog += pkt_len;
> > > >> > +             sch->q.qlen++;
> > > >> > +             flow_queue_add(sel_flow, skb);
> > > >> > +             if (list_empty(&sel_flow->flowchain)) {
> > > >> > +                     list_add_tail(&sel_flow->flowchain, &q->new_flows);
> > > >> > +                     q->stats.new_flow_count++;
> > > >> > +                     sel_flow->deficit = q->quantum;
> > > >> > +                     sel_flow->stats.dropped = 0;
> > > >> > +                     sel_flow->qlen = 0;
> > > >> > +                     sel_flow->backlog = 0;
> > > >> > +             }
> > > >> > +             sel_flow->qlen++;
> > > >> > +             sel_flow->stats.packets_in++;
> > > >> > +             sel_flow->backlog += pkt_len;
> > > >> > +             return NET_XMIT_SUCCESS;
> > > >> > +     }
> > > >> > +out:
> > > >> > +     q->stats.dropped++;
> > > >> > +     sel_flow->stats.dropped++;
> > > >> > +     return qdisc_drop(skb, sch, to_free);
> > > >>
> > > >> You probably want to return NET_XMIT_CN instead of NET_XMIT_DROP here.
> > > >>
> > > >
> > > > Should I replace qdisc_drop with __qdisc_drop and return NET_XMIT_CN?
> > >
> > > Yeah, I think that would be more appropriate here, as that would
> > > directly throttle sockets on the same host.
> > >
> > > >> > +}
> > > >> > +
> > > >> > +static const struct nla_policy fq_pie_policy[TCA_FQ_PIE_MAX + 1] = {
> > > >> > +     [TCA_FQ_PIE_TARGET] = {.type = NLA_U32},
> > > >> > +     [TCA_FQ_PIE_LIMIT] = {.type = NLA_U32},
> > > >> > +     [TCA_FQ_PIE_TUPDATE] = {.type = NLA_U32},
> > > >> > +     [TCA_FQ_PIE_ALPHA] = {.type = NLA_U32},
> > > >> > +     [TCA_FQ_PIE_BETA] = {.type = NLA_U32},
> > > >> > +     [TCA_FQ_PIE_ECN] = {.type = NLA_U32},
> > > >> > +     [TCA_FQ_PIE_QUANTUM] = {.type = NLA_U32},
> > > >> > +     [TCA_FQ_PIE_BYTEMODE] = {.type = NLA_U32},
> > > >> > +     [TCA_FQ_PIE_FLOWS] = {.type = NLA_U32},
> > > >> > +     [TCA_FQ_PIE_ECN_PROB] = {.type = NLA_U32}
> > > >> > +};
> > > >> > +
> > > >> > +static inline struct sk_buff *dequeue_head(struct fq_pie_flow *flow)
> > > >> > +{
> > > >> > +     struct sk_buff *skb = flow->head;
> > > >> > +
> > > >> > +     flow->head = skb->next;
> > > >> > +     skb->next = NULL;
> > > >> > +     return skb;
> > > >> > +}
> > > >> > +
> > > >> > +static struct sk_buff *fq_pie_qdisc_dequeue(struct Qdisc *sch)
> > > >> > +{
> > > >> > +     struct fq_pie_sched_data *q = qdisc_priv(sch);
> > > >> > +     struct sk_buff *skb = NULL;
> > > >> > +     struct fq_pie_flow *flow;
> > > >> > +     struct list_head *head;
> > > >> > +     u32 pkt_len;
> > > >> > +
> > > >> > +begin:
> > > >> > +     head = &q->new_flows;
> > > >> > +     if (list_empty(head)) {
> > > >> > +             head = &q->old_flows;
> > > >> > +             if (list_empty(head))
> > > >> > +                     return NULL;
> > > >> > +     }
> > > >> > +
> > > >> > +     flow = list_first_entry(head, struct fq_pie_flow, flowchain);
> > > >> > +     /* Flow has exhausted all its credits */
> > > >> > +     if (flow->deficit <= 0) {
> > > >> > +             flow->deficit += q->quantum;
> > > >> > +             list_move_tail(&flow->flowchain, &q->old_flows);
> > > >> > +             goto begin;
> > > >> > +     }
> > > >> > +
> > > >> > +     if (flow->head) {
> > > >> > +             skb = dequeue_head(flow);
> > > >> > +             pkt_len = qdisc_pkt_len(skb);
> > > >> > +             sch->qstats.backlog -= pkt_len;
> > > >> > +             sch->q.qlen--;
> > > >> > +             qdisc_bstats_update(sch, skb);
> > > >> > +     }
> > > >>
> > > >> If you factor this out into a dequeue_func(), this dequeue() function is
> > > >> very close to identical to the one in fq_codel().
> > > >
> > > > Do you suggest I put the if(flow->head) block in a different function?
> > >
> > > Yeah, that was what I meant. However, without doing the full refactoring
> > > it's only borderline useful, so feel free to just keep it the way it
> > > is...
> > >
> > > -Toke
> >
> >
> >
> > --
> > -------------
> > Gautam |
>
>
>
> --
>
> Dave Täht
> CTO, TekLibre, LLC
> http://www.teklibre.com
> Tel: 1-831-205-9740



-- 

Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-831-205-9740

Powered by blists - more mailing lists