Message ID | 20240215160458.1727237-4-ast@fiberby.net (mailing list archive) |
---|---|
State | Changes Requested |
Delegated to: | Netdev Maintainers |
Headers | show |
Series | make skip_sw actually skip software | expand |
On Thu, Feb 15, 2024 at 11:06 AM Asbjørn Sloth Tønnesen <ast@fiberby.net> wrote: > > TC filters come in 3 variants: > - no flag (no opinion, process wherever possible) > - skip_hw (do not process filter by hardware) > - skip_sw (do not process filter by software) > > However skip_sw is implemented so that the skip_sw > flag can first be checked, after it has been matched. > > IMHO it's common when using skip_sw, to use it on all rules. > > So if all filters in a block is skip_sw filters, then > we can bail early, we can thus avoid having to match > the filters, just to check for the skip_sw flag. > > +----------------------------+--------+--------+--------+ > | Test description | Pre | Post | Rel. | > | | kpps | kpps | chg. | > +----------------------------+--------+--------+--------+ > | basic forwarding + notrack | 1264.9 | 1277.7 | 1.01x | > | switch to eswitch mode | 1067.1 | 1071.0 | 1.00x | > | add ingress qdisc | 1056.0 | 1059.1 | 1.00x | > +----------------------------+--------+--------+--------+ > | 1 non-matching rule | 927.9 | 1057.1 | 1.14x | > | 10 non-matching rules | 495.8 | 1055.6 | 2.13x | > | 25 non-matching rules | 280.6 | 1053.5 | 3.75x | > | 50 non-matching rules | 162.0 | 1055.7 | 6.52x | > | 100 non-matching rules | 87.7 | 1019.0 | 11.62x | > +----------------------------+--------+--------+--------+ > > perf top (100 n-m skip_sw rules - pre patch): > 25.57% [kernel] [k] __skb_flow_dissect > 20.77% [kernel] [k] rhashtable_jhash2 > 14.26% [kernel] [k] fl_classify > 13.28% [kernel] [k] fl_mask_lookup > 6.38% [kernel] [k] memset_orig > 3.22% [kernel] [k] tcf_classify > > perf top (100 n-m skip_sw rules - post patch): > 4.28% [kernel] [k] __dev_queue_xmit > 3.80% [kernel] [k] check_preemption_disabled > 3.68% [kernel] [k] nft_do_chain > 3.08% [kernel] [k] __netif_receive_skb_core.constprop.0 > 2.59% [kernel] [k] mlx5e_xmit > 2.48% [kernel] [k] mlx5e_skb_from_cqe_mpwrq_nonlinear > The concept makes sense - but i am wondering when you have a mix of skip_sw and skip_hw if it makes more sense to just avoid looking up skip_sw at all in the s/w datapath? Potentially by separating the hashes for skip_sw/hw. I know it's a deeper surgery - but would be more general purpose....unless i am missing something > Test setup: > DUT: Intel Xeon D-1518 (2.20GHz) w/ Nvidia/Mellanox ConnectX-6 Dx 2x100G > Data rate measured on switch (Extreme X690), and DUT connected as > a router on a stick, with pktgen and pktsink as VLANs. > Pktgen was in range 12.79 - 12.95 Mpps across all tests. > Hrm. Those are "tiny" numbers (25G @64B is about 3x that). What are the packet sizes? Perhaps the traffic generator is a limitation here? Also feels like you are doing exact matches? A sample flower rule would have helped. cheers, jamal > Signed-off-by: Asbjørn Sloth Tønnesen <ast@fiberby.net> > --- > include/net/pkt_cls.h | 5 +++++ > net/core/dev.c | 3 +++ > 2 files changed, 8 insertions(+) > > diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h > index a4ee43f493bb..a065da4df7ff 100644 > --- a/include/net/pkt_cls.h > +++ b/include/net/pkt_cls.h > @@ -74,6 +74,11 @@ static inline bool tcf_block_non_null_shared(struct tcf_block *block) > return block && block->index; > } > > +static inline bool tcf_block_has_skip_sw_only(struct tcf_block *block) > +{ > + return block && atomic_read(&block->filtercnt) == atomic_read(&block->skipswcnt); > +} > + > static inline struct Qdisc *tcf_block_q(struct tcf_block *block) > { > WARN_ON(tcf_block_shared(block)); > diff --git a/net/core/dev.c b/net/core/dev.c > index d8dd293a7a27..7cd014e5066e 100644 > --- a/net/core/dev.c > +++ b/net/core/dev.c > @@ -3910,6 +3910,9 @@ static int tc_run(struct tcx_entry *entry, struct sk_buff *skb, > if (!miniq) > return ret; > > + if (tcf_block_has_skip_sw_only(miniq->block)) > + return ret; > + > tc_skb_cb(skb)->mru = 0; > tc_skb_cb(skb)->post_ct = false; > tcf_set_drop_reason(skb, *drop_reason); > -- > 2.43.0 >
On Thu 15 Feb 2024 at 16:04, Asbjørn Sloth Tønnesen <ast@fiberby.net> wrote: > TC filters come in 3 variants: > - no flag (no opinion, process wherever possible) > - skip_hw (do not process filter by hardware) > - skip_sw (do not process filter by software) > > However skip_sw is implemented so that the skip_sw > flag can first be checked, after it has been matched. > > IMHO it's common when using skip_sw, to use it on all rules. > > So if all filters in a block is skip_sw filters, then > we can bail early, we can thus avoid having to match > the filters, just to check for the skip_sw flag. > > +----------------------------+--------+--------+--------+ > | Test description | Pre | Post | Rel. | > | | kpps | kpps | chg. | > +----------------------------+--------+--------+--------+ > | basic forwarding + notrack | 1264.9 | 1277.7 | 1.01x | > | switch to eswitch mode | 1067.1 | 1071.0 | 1.00x | > | add ingress qdisc | 1056.0 | 1059.1 | 1.00x | > +----------------------------+--------+--------+--------+ > | 1 non-matching rule | 927.9 | 1057.1 | 1.14x | > | 10 non-matching rules | 495.8 | 1055.6 | 2.13x | > | 25 non-matching rules | 280.6 | 1053.5 | 3.75x | > | 50 non-matching rules | 162.0 | 1055.7 | 6.52x | > | 100 non-matching rules | 87.7 | 1019.0 | 11.62x | > +----------------------------+--------+--------+--------+ > > perf top (100 n-m skip_sw rules - pre patch): > 25.57% [kernel] [k] __skb_flow_dissect > 20.77% [kernel] [k] rhashtable_jhash2 > 14.26% [kernel] [k] fl_classify > 13.28% [kernel] [k] fl_mask_lookup > 6.38% [kernel] [k] memset_orig > 3.22% [kernel] [k] tcf_classify > > perf top (100 n-m skip_sw rules - post patch): > 4.28% [kernel] [k] __dev_queue_xmit > 3.80% [kernel] [k] check_preemption_disabled > 3.68% [kernel] [k] nft_do_chain > 3.08% [kernel] [k] __netif_receive_skb_core.constprop.0 > 2.59% [kernel] [k] mlx5e_xmit > 2.48% [kernel] [k] mlx5e_skb_from_cqe_mpwrq_nonlinear > > Test setup: > DUT: Intel Xeon D-1518 (2.20GHz) w/ Nvidia/Mellanox ConnectX-6 Dx 2x100G > Data rate measured on switch (Extreme X690), and DUT connected as > a router on a stick, with pktgen and pktsink as VLANs. > Pktgen was in range 12.79 - 12.95 Mpps across all tests. > > Signed-off-by: Asbjørn Sloth Tønnesen <ast@fiberby.net> > --- > include/net/pkt_cls.h | 5 +++++ > net/core/dev.c | 3 +++ > 2 files changed, 8 insertions(+) > > diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h > index a4ee43f493bb..a065da4df7ff 100644 > --- a/include/net/pkt_cls.h > +++ b/include/net/pkt_cls.h > @@ -74,6 +74,11 @@ static inline bool tcf_block_non_null_shared(struct tcf_block *block) > return block && block->index; > } > > +static inline bool tcf_block_has_skip_sw_only(struct tcf_block *block) > +{ > + return block && atomic_read(&block->filtercnt) == atomic_read(&block->skipswcnt); > +} Note that this introduces a read from heavily contended cache-line on data path for all classifiers, including the ones that don't support offloads. Wonder if this a concern for users running purely software tc. > + > static inline struct Qdisc *tcf_block_q(struct tcf_block *block) > { > WARN_ON(tcf_block_shared(block)); > diff --git a/net/core/dev.c b/net/core/dev.c > index d8dd293a7a27..7cd014e5066e 100644 > --- a/net/core/dev.c > +++ b/net/core/dev.c > @@ -3910,6 +3910,9 @@ static int tc_run(struct tcx_entry *entry, struct sk_buff *skb, > if (!miniq) > return ret; > > + if (tcf_block_has_skip_sw_only(miniq->block)) > + return ret; > + > tc_skb_cb(skb)->mru = 0; > tc_skb_cb(skb)->post_ct = false; > tcf_set_drop_reason(skb, *drop_reason);
Thu, Feb 15, 2024 at 06:49:05PM CET, jhs@mojatatu.com wrote: >On Thu, Feb 15, 2024 at 11:06 AM Asbjørn Sloth Tønnesen <ast@fiberby.net> wrote: >> >> TC filters come in 3 variants: >> - no flag (no opinion, process wherever possible) >> - skip_hw (do not process filter by hardware) >> - skip_sw (do not process filter by software) >> >> However skip_sw is implemented so that the skip_sw >> flag can first be checked, after it has been matched. >> >> IMHO it's common when using skip_sw, to use it on all rules. >> >> So if all filters in a block is skip_sw filters, then >> we can bail early, we can thus avoid having to match >> the filters, just to check for the skip_sw flag. >> >> +----------------------------+--------+--------+--------+ >> | Test description | Pre | Post | Rel. | >> | | kpps | kpps | chg. | >> +----------------------------+--------+--------+--------+ >> | basic forwarding + notrack | 1264.9 | 1277.7 | 1.01x | >> | switch to eswitch mode | 1067.1 | 1071.0 | 1.00x | >> | add ingress qdisc | 1056.0 | 1059.1 | 1.00x | >> +----------------------------+--------+--------+--------+ >> | 1 non-matching rule | 927.9 | 1057.1 | 1.14x | >> | 10 non-matching rules | 495.8 | 1055.6 | 2.13x | >> | 25 non-matching rules | 280.6 | 1053.5 | 3.75x | >> | 50 non-matching rules | 162.0 | 1055.7 | 6.52x | >> | 100 non-matching rules | 87.7 | 1019.0 | 11.62x | >> +----------------------------+--------+--------+--------+ >> >> perf top (100 n-m skip_sw rules - pre patch): >> 25.57% [kernel] [k] __skb_flow_dissect >> 20.77% [kernel] [k] rhashtable_jhash2 >> 14.26% [kernel] [k] fl_classify >> 13.28% [kernel] [k] fl_mask_lookup >> 6.38% [kernel] [k] memset_orig >> 3.22% [kernel] [k] tcf_classify >> >> perf top (100 n-m skip_sw rules - post patch): >> 4.28% [kernel] [k] __dev_queue_xmit >> 3.80% [kernel] [k] check_preemption_disabled >> 3.68% [kernel] [k] nft_do_chain >> 3.08% [kernel] [k] __netif_receive_skb_core.constprop.0 >> 2.59% [kernel] [k] mlx5e_xmit >> 2.48% [kernel] [k] mlx5e_skb_from_cqe_mpwrq_nonlinear >> > >The concept makes sense - but i am wondering when you have a mix of >skip_sw and skip_hw if it makes more sense to just avoid looking up >skip_sw at all in the s/w datapath? Potentially by separating the >hashes for skip_sw/hw. I know it's a deeper surgery - but would be Yeah, there could be 2 hashes: skip_sw/rest rest is the only one that needs to be looked-up in kernel datapath. skip_sw is just for control path. But is it worth the efford? I mean, since now, nobody seemed to care. If this patchset solves the problem for this usecase, I think it is enough. In that case, I'm fine with this patch: Reviewed-by: Jiri Pirko <jiri@nvidia.com> >more general purpose....unless i am missing something > >> Test setup: >> DUT: Intel Xeon D-1518 (2.20GHz) w/ Nvidia/Mellanox ConnectX-6 Dx 2x100G >> Data rate measured on switch (Extreme X690), and DUT connected as >> a router on a stick, with pktgen and pktsink as VLANs. >> Pktgen was in range 12.79 - 12.95 Mpps across all tests. >> > >Hrm. Those are "tiny" numbers (25G @64B is about 3x that). What are >the packet sizes? >Perhaps the traffic generator is a limitation here? >Also feels like you are doing exact matches? A sample flower rule >would have helped. > >cheers, >jamal >> Signed-off-by: Asbjørn Sloth Tønnesen <ast@fiberby.net> >> --- >> include/net/pkt_cls.h | 5 +++++ >> net/core/dev.c | 3 +++ >> 2 files changed, 8 insertions(+) >> >> diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h >> index a4ee43f493bb..a065da4df7ff 100644 >> --- a/include/net/pkt_cls.h >> +++ b/include/net/pkt_cls.h >> @@ -74,6 +74,11 @@ static inline bool tcf_block_non_null_shared(struct tcf_block *block) >> return block && block->index; >> } >> >> +static inline bool tcf_block_has_skip_sw_only(struct tcf_block *block) >> +{ >> + return block && atomic_read(&block->filtercnt) == atomic_read(&block->skipswcnt); >> +} >> + >> static inline struct Qdisc *tcf_block_q(struct tcf_block *block) >> { >> WARN_ON(tcf_block_shared(block)); >> diff --git a/net/core/dev.c b/net/core/dev.c >> index d8dd293a7a27..7cd014e5066e 100644 >> --- a/net/core/dev.c >> +++ b/net/core/dev.c >> @@ -3910,6 +3910,9 @@ static int tc_run(struct tcx_entry *entry, struct sk_buff *skb, >> if (!miniq) >> return ret; >> >> + if (tcf_block_has_skip_sw_only(miniq->block)) >> + return ret; >> + >> tc_skb_cb(skb)->mru = 0; >> tc_skb_cb(skb)->post_ct = false; >> tcf_set_drop_reason(skb, *drop_reason); >> -- >> 2.43.0 >>
Hi Jamal, On 2/15/24 17:49, Jamal Hadi Salim wrote: > On Thu, Feb 15, 2024 at 11:06 AM Asbjørn Sloth Tønnesen <ast@fiberby.net> wrote: >> >> TC filters come in 3 variants: >> - no flag (no opinion, process wherever possible) >> - skip_hw (do not process filter by hardware) >> - skip_sw (do not process filter by software) >> >> However skip_sw is implemented so that the skip_sw >> flag can first be checked, after it has been matched. >> >> IMHO it's common when using skip_sw, to use it on all rules. >> >> So if all filters in a block is skip_sw filters, then >> we can bail early, we can thus avoid having to match >> the filters, just to check for the skip_sw flag. >> >> +----------------------------+--------+--------+--------+ >> | Test description | Pre | Post | Rel. | >> | | kpps | kpps | chg. | >> +----------------------------+--------+--------+--------+ >> | basic forwarding + notrack | 1264.9 | 1277.7 | 1.01x | >> | switch to eswitch mode | 1067.1 | 1071.0 | 1.00x | >> | add ingress qdisc | 1056.0 | 1059.1 | 1.00x | >> +----------------------------+--------+--------+--------+ >> | 1 non-matching rule | 927.9 | 1057.1 | 1.14x | >> | 10 non-matching rules | 495.8 | 1055.6 | 2.13x | >> | 25 non-matching rules | 280.6 | 1053.5 | 3.75x | >> | 50 non-matching rules | 162.0 | 1055.7 | 6.52x | >> | 100 non-matching rules | 87.7 | 1019.0 | 11.62x | >> +----------------------------+--------+--------+--------+ >> >> perf top (100 n-m skip_sw rules - pre patch): >> 25.57% [kernel] [k] __skb_flow_dissect >> 20.77% [kernel] [k] rhashtable_jhash2 >> 14.26% [kernel] [k] fl_classify >> 13.28% [kernel] [k] fl_mask_lookup >> 6.38% [kernel] [k] memset_orig >> 3.22% [kernel] [k] tcf_classify >> >> perf top (100 n-m skip_sw rules - post patch): >> 4.28% [kernel] [k] __dev_queue_xmit >> 3.80% [kernel] [k] check_preemption_disabled >> 3.68% [kernel] [k] nft_do_chain >> 3.08% [kernel] [k] __netif_receive_skb_core.constprop.0 >> 2.59% [kernel] [k] mlx5e_xmit >> 2.48% [kernel] [k] mlx5e_skb_from_cqe_mpwrq_nonlinear >> > > The concept makes sense - but i am wondering when you have a mix of > skip_sw and skip_hw if it makes more sense to just avoid looking up > skip_sw at all in the s/w datapath? Potentially by separating the > hashes for skip_sw/hw. I know it's a deeper surgery - but would be > more general purpose....unless i am missing something > >> Test setup: >> DUT: Intel Xeon D-1518 (2.20GHz) w/ Nvidia/Mellanox ConnectX-6 Dx 2x100G >> Data rate measured on switch (Extreme X690), and DUT connected as >> a router on a stick, with pktgen and pktsink as VLANs. >> Pktgen was in range 12.79 - 12.95 Mpps across all tests. >> > > Hrm. Those are "tiny" numbers (25G @64B is about 3x that). What are > the packet sizes? > Perhaps the traffic generator is a limitation here? > Also feels like you are doing exact matches? A sample flower rule > would have helped. Yeah, I would also have liked those number to be higher. Sorry forgot to mention it is 64B packets. Sadly, I used two machine to compensate for my lack of pktgen skills. I know that there are faster packet generators, but I just used the in-kernel one, and haven't invested much time into it, but the normal ethtool params mentioned in the doc didn't change much, and since it was still more than enough packets that DUT would go to 100% CPU, I didn't pursue it further. pktgen A: Xeon E5-1620 v2 @ 3.70GHz w/ ConnectX-6 Dx and 100G link. pktgen B: Xeon(R) D-2123IT @ 2.20GHz w/ Intel X722 and 10G link. Both pktgen boxes are running stock Debian bullseye kernel (v6.1) I found this to work best: ./pktgen_sample05_flow_per_thread.sh -i enp8s0f0np0 -s 64 -d 10.53.22.3 -m 2a:11:22:33:21:11 -p 1024-65000 -t 8 -n 0 Datarates monitored through SNMP using: snmpdelta -Cp 10 -c public -v 2c -Cs $hostname IF-MIB::ifInUcastPkts.1057 IF-MIB::ifOutUcastPkts.1057 VLAN config, MAC and IP addressing: v21: 2a:11:22:33:21:11 - 10.53.21.1/27 - dut v21 (tagged VLAN) 2a:11:22:33:21:12 - 10.53.21.2/27 - pktgen-a (untagged) 2a:11:22:33:21:13 - 10.53.21.3/27 - pktgen-b (untagged) v22: 2a:11:22:33:22:21 - 10.53.22.2/31 - dut v22 (tagged VLAN) 2a:11:22:33:22:22 - 10.53.22.3/31 - packet drop (untagged) Switch MAC address table config: # static entry create fdb 2a:11:22:33:21:11 vlan "v21" port 57 # blackhole pktsink create fdb 2a:11:22:33:22:22 vlan "v22" blackhole flow control are disabled on all links: ethtool -A $dev rx off tx off I have uploaded the script used for the flower rules in the tests here: https://files.fiberby.net/ast/2024/tc_skip_sw/flower_placement_tests.sh > cheers, > jamal >> Signed-off-by: Asbjørn Sloth Tønnesen <ast@fiberby.net> >> --- >> include/net/pkt_cls.h | 5 +++++ >> net/core/dev.c | 3 +++ >> 2 files changed, 8 insertions(+) >> >> diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h >> index a4ee43f493bb..a065da4df7ff 100644 >> --- a/include/net/pkt_cls.h >> +++ b/include/net/pkt_cls.h >> @@ -74,6 +74,11 @@ static inline bool tcf_block_non_null_shared(struct tcf_block *block) >> return block && block->index; >> } >> >> +static inline bool tcf_block_has_skip_sw_only(struct tcf_block *block) >> +{ >> + return block && atomic_read(&block->filtercnt) == atomic_read(&block->skipswcnt); >> +} >> + >> static inline struct Qdisc *tcf_block_q(struct tcf_block *block) >> { >> WARN_ON(tcf_block_shared(block)); >> diff --git a/net/core/dev.c b/net/core/dev.c >> index d8dd293a7a27..7cd014e5066e 100644 >> --- a/net/core/dev.c >> +++ b/net/core/dev.c >> @@ -3910,6 +3910,9 @@ static int tc_run(struct tcx_entry *entry, struct sk_buff *skb, >> if (!miniq) >> return ret; >> >> + if (tcf_block_has_skip_sw_only(miniq->block)) >> + return ret; >> + >> tc_skb_cb(skb)->mru = 0; >> tc_skb_cb(skb)->post_ct = false; >> tcf_set_drop_reason(skb, *drop_reason); >> -- >> 2.43.0 >>
Hi Vlad, On 2/16/24 08:47, Vlad Buslov wrote: > On Thu 15 Feb 2024 at 16:04, Asbjørn Sloth Tønnesen <ast@fiberby.net> wrote: >> TC filters come in 3 variants: >> - no flag (no opinion, process wherever possible) >> - skip_hw (do not process filter by hardware) >> - skip_sw (do not process filter by software) >> >> However skip_sw is implemented so that the skip_sw >> flag can first be checked, after it has been matched. >> >> IMHO it's common when using skip_sw, to use it on all rules. >> >> So if all filters in a block is skip_sw filters, then >> we can bail early, we can thus avoid having to match >> the filters, just to check for the skip_sw flag. >> >> +----------------------------+--------+--------+--------+ >> | Test description | Pre | Post | Rel. | >> | | kpps | kpps | chg. | >> +----------------------------+--------+--------+--------+ >> | basic forwarding + notrack | 1264.9 | 1277.7 | 1.01x | >> | switch to eswitch mode | 1067.1 | 1071.0 | 1.00x | >> | add ingress qdisc | 1056.0 | 1059.1 | 1.00x | >> +----------------------------+--------+--------+--------+ >> | 1 non-matching rule | 927.9 | 1057.1 | 1.14x | >> | 10 non-matching rules | 495.8 | 1055.6 | 2.13x | >> | 25 non-matching rules | 280.6 | 1053.5 | 3.75x | >> | 50 non-matching rules | 162.0 | 1055.7 | 6.52x | >> | 100 non-matching rules | 87.7 | 1019.0 | 11.62x | >> +----------------------------+--------+--------+--------+ >> >> perf top (100 n-m skip_sw rules - pre patch): >> 25.57% [kernel] [k] __skb_flow_dissect >> 20.77% [kernel] [k] rhashtable_jhash2 >> 14.26% [kernel] [k] fl_classify >> 13.28% [kernel] [k] fl_mask_lookup >> 6.38% [kernel] [k] memset_orig >> 3.22% [kernel] [k] tcf_classify >> >> perf top (100 n-m skip_sw rules - post patch): >> 4.28% [kernel] [k] __dev_queue_xmit >> 3.80% [kernel] [k] check_preemption_disabled >> 3.68% [kernel] [k] nft_do_chain >> 3.08% [kernel] [k] __netif_receive_skb_core.constprop.0 >> 2.59% [kernel] [k] mlx5e_xmit >> 2.48% [kernel] [k] mlx5e_skb_from_cqe_mpwrq_nonlinear >> >> Test setup: >> DUT: Intel Xeon D-1518 (2.20GHz) w/ Nvidia/Mellanox ConnectX-6 Dx 2x100G >> Data rate measured on switch (Extreme X690), and DUT connected as >> a router on a stick, with pktgen and pktsink as VLANs. >> Pktgen was in range 12.79 - 12.95 Mpps across all tests. >> >> Signed-off-by: Asbjørn Sloth Tønnesen <ast@fiberby.net> >> --- >> include/net/pkt_cls.h | 5 +++++ >> net/core/dev.c | 3 +++ >> 2 files changed, 8 insertions(+) >> >> diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h >> index a4ee43f493bb..a065da4df7ff 100644 >> --- a/include/net/pkt_cls.h >> +++ b/include/net/pkt_cls.h >> @@ -74,6 +74,11 @@ static inline bool tcf_block_non_null_shared(struct tcf_block *block) >> return block && block->index; >> } >> >> +static inline bool tcf_block_has_skip_sw_only(struct tcf_block *block) >> +{ >> + return block && atomic_read(&block->filtercnt) == atomic_read(&block->skipswcnt); >> +} > > Note that this introduces a read from heavily contended cache-line on > data path for all classifiers, including the ones that don't support > offloads. Wonder if this a concern for users running purely software tc. Unfortunately, I don't have access to any multi-CPU machines, so I haven't been able to test the impact of that. Alternatively I guess I could also maintain a static key in the counter update logic. >> + >> static inline struct Qdisc *tcf_block_q(struct tcf_block *block) >> { >> WARN_ON(tcf_block_shared(block)); >> diff --git a/net/core/dev.c b/net/core/dev.c >> index d8dd293a7a27..7cd014e5066e 100644 >> --- a/net/core/dev.c >> +++ b/net/core/dev.c >> @@ -3910,6 +3910,9 @@ static int tc_run(struct tcx_entry *entry, struct sk_buff *skb, >> if (!miniq) >> return ret; >> >> + if (tcf_block_has_skip_sw_only(miniq->block)) >> + return ret; >> + >> tc_skb_cb(skb)->mru = 0; >> tc_skb_cb(skb)->post_ct = false; >> tcf_set_drop_reason(skb, *drop_reason); >
On Fri, Feb 16, 2024 at 7:57 AM Jiri Pirko <jiri@resnulli.us> wrote: > > Thu, Feb 15, 2024 at 06:49:05PM CET, jhs@mojatatu.com wrote: > >On Thu, Feb 15, 2024 at 11:06 AM Asbjørn Sloth Tønnesen <ast@fiberby.net> wrote: > >> > >> TC filters come in 3 variants: > >> - no flag (no opinion, process wherever possible) > >> - skip_hw (do not process filter by hardware) > >> - skip_sw (do not process filter by software) > >> > >> However skip_sw is implemented so that the skip_sw > >> flag can first be checked, after it has been matched. > >> > >> IMHO it's common when using skip_sw, to use it on all rules. > >> > >> So if all filters in a block is skip_sw filters, then > >> we can bail early, we can thus avoid having to match > >> the filters, just to check for the skip_sw flag. > >> > >> +----------------------------+--------+--------+--------+ > >> | Test description | Pre | Post | Rel. | > >> | | kpps | kpps | chg. | > >> +----------------------------+--------+--------+--------+ > >> | basic forwarding + notrack | 1264.9 | 1277.7 | 1.01x | > >> | switch to eswitch mode | 1067.1 | 1071.0 | 1.00x | > >> | add ingress qdisc | 1056.0 | 1059.1 | 1.00x | > >> +----------------------------+--------+--------+--------+ > >> | 1 non-matching rule | 927.9 | 1057.1 | 1.14x | > >> | 10 non-matching rules | 495.8 | 1055.6 | 2.13x | > >> | 25 non-matching rules | 280.6 | 1053.5 | 3.75x | > >> | 50 non-matching rules | 162.0 | 1055.7 | 6.52x | > >> | 100 non-matching rules | 87.7 | 1019.0 | 11.62x | > >> +----------------------------+--------+--------+--------+ > >> > >> perf top (100 n-m skip_sw rules - pre patch): > >> 25.57% [kernel] [k] __skb_flow_dissect > >> 20.77% [kernel] [k] rhashtable_jhash2 > >> 14.26% [kernel] [k] fl_classify > >> 13.28% [kernel] [k] fl_mask_lookup > >> 6.38% [kernel] [k] memset_orig > >> 3.22% [kernel] [k] tcf_classify > >> > >> perf top (100 n-m skip_sw rules - post patch): > >> 4.28% [kernel] [k] __dev_queue_xmit > >> 3.80% [kernel] [k] check_preemption_disabled > >> 3.68% [kernel] [k] nft_do_chain > >> 3.08% [kernel] [k] __netif_receive_skb_core.constprop.0 > >> 2.59% [kernel] [k] mlx5e_xmit > >> 2.48% [kernel] [k] mlx5e_skb_from_cqe_mpwrq_nonlinear > >> > > > >The concept makes sense - but i am wondering when you have a mix of > >skip_sw and skip_hw if it makes more sense to just avoid looking up > >skip_sw at all in the s/w datapath? Potentially by separating the > >hashes for skip_sw/hw. I know it's a deeper surgery - but would be > > Yeah, there could be 2 hashes: skip_sw/rest > rest is the only one that needs to be looked-up in kernel datapath. > skip_sw is just for control path. > > But is it worth the efford? I mean, since now, nobody seemed to care. If > this patchset solves the problem for this usecase, I think it is enough. > May not be worth the effort - and this is a reasonable use case. The approach is a hack nonetheless and kills at least some insects. To address the issues Vlad brought up, perhaps we should wrap it under some kconfig. cheers, jamal > In that case, I'm fine with this patch: > > Reviewed-by: Jiri Pirko <jiri@nvidia.com> > > > > >more general purpose....unless i am missing something > > > >> Test setup: > >> DUT: Intel Xeon D-1518 (2.20GHz) w/ Nvidia/Mellanox ConnectX-6 Dx 2x100G > >> Data rate measured on switch (Extreme X690), and DUT connected as > >> a router on a stick, with pktgen and pktsink as VLANs. > >> Pktgen was in range 12.79 - 12.95 Mpps across all tests. > >> > > > >Hrm. Those are "tiny" numbers (25G @64B is about 3x that). What are > >the packet sizes? > >Perhaps the traffic generator is a limitation here? > >Also feels like you are doing exact matches? A sample flower rule > >would have helped. > > > >cheers, > >jamal > >> Signed-off-by: Asbjørn Sloth Tønnesen <ast@fiberby.net> > >> --- > >> include/net/pkt_cls.h | 5 +++++ > >> net/core/dev.c | 3 +++ > >> 2 files changed, 8 insertions(+) > >> > >> diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h > >> index a4ee43f493bb..a065da4df7ff 100644 > >> --- a/include/net/pkt_cls.h > >> +++ b/include/net/pkt_cls.h > >> @@ -74,6 +74,11 @@ static inline bool tcf_block_non_null_shared(struct tcf_block *block) > >> return block && block->index; > >> } > >> > >> +static inline bool tcf_block_has_skip_sw_only(struct tcf_block *block) > >> +{ > >> + return block && atomic_read(&block->filtercnt) == atomic_read(&block->skipswcnt); > >> +} > >> + > >> static inline struct Qdisc *tcf_block_q(struct tcf_block *block) > >> { > >> WARN_ON(tcf_block_shared(block)); > >> diff --git a/net/core/dev.c b/net/core/dev.c > >> index d8dd293a7a27..7cd014e5066e 100644 > >> --- a/net/core/dev.c > >> +++ b/net/core/dev.c > >> @@ -3910,6 +3910,9 @@ static int tc_run(struct tcx_entry *entry, struct sk_buff *skb, > >> if (!miniq) > >> return ret; > >> > >> + if (tcf_block_has_skip_sw_only(miniq->block)) > >> + return ret; > >> + > >> tc_skb_cb(skb)->mru = 0; > >> tc_skb_cb(skb)->post_ct = false; > >> tcf_set_drop_reason(skb, *drop_reason); > >> -- > >> 2.43.0 > >>
diff --git a/include/net/pkt_cls.h b/include/net/pkt_cls.h index a4ee43f493bb..a065da4df7ff 100644 --- a/include/net/pkt_cls.h +++ b/include/net/pkt_cls.h @@ -74,6 +74,11 @@ static inline bool tcf_block_non_null_shared(struct tcf_block *block) return block && block->index; } +static inline bool tcf_block_has_skip_sw_only(struct tcf_block *block) +{ + return block && atomic_read(&block->filtercnt) == atomic_read(&block->skipswcnt); +} + static inline struct Qdisc *tcf_block_q(struct tcf_block *block) { WARN_ON(tcf_block_shared(block)); diff --git a/net/core/dev.c b/net/core/dev.c index d8dd293a7a27..7cd014e5066e 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -3910,6 +3910,9 @@ static int tc_run(struct tcx_entry *entry, struct sk_buff *skb, if (!miniq) return ret; + if (tcf_block_has_skip_sw_only(miniq->block)) + return ret; + tc_skb_cb(skb)->mru = 0; tc_skb_cb(skb)->post_ct = false; tcf_set_drop_reason(skb, *drop_reason);
TC filters come in 3 variants: - no flag (no opinion, process wherever possible) - skip_hw (do not process filter by hardware) - skip_sw (do not process filter by software) However skip_sw is implemented so that the skip_sw flag can first be checked, after it has been matched. IMHO it's common when using skip_sw, to use it on all rules. So if all filters in a block is skip_sw filters, then we can bail early, we can thus avoid having to match the filters, just to check for the skip_sw flag. +----------------------------+--------+--------+--------+ | Test description | Pre | Post | Rel. | | | kpps | kpps | chg. | +----------------------------+--------+--------+--------+ | basic forwarding + notrack | 1264.9 | 1277.7 | 1.01x | | switch to eswitch mode | 1067.1 | 1071.0 | 1.00x | | add ingress qdisc | 1056.0 | 1059.1 | 1.00x | +----------------------------+--------+--------+--------+ | 1 non-matching rule | 927.9 | 1057.1 | 1.14x | | 10 non-matching rules | 495.8 | 1055.6 | 2.13x | | 25 non-matching rules | 280.6 | 1053.5 | 3.75x | | 50 non-matching rules | 162.0 | 1055.7 | 6.52x | | 100 non-matching rules | 87.7 | 1019.0 | 11.62x | +----------------------------+--------+--------+--------+ perf top (100 n-m skip_sw rules - pre patch): 25.57% [kernel] [k] __skb_flow_dissect 20.77% [kernel] [k] rhashtable_jhash2 14.26% [kernel] [k] fl_classify 13.28% [kernel] [k] fl_mask_lookup 6.38% [kernel] [k] memset_orig 3.22% [kernel] [k] tcf_classify perf top (100 n-m skip_sw rules - post patch): 4.28% [kernel] [k] __dev_queue_xmit 3.80% [kernel] [k] check_preemption_disabled 3.68% [kernel] [k] nft_do_chain 3.08% [kernel] [k] __netif_receive_skb_core.constprop.0 2.59% [kernel] [k] mlx5e_xmit 2.48% [kernel] [k] mlx5e_skb_from_cqe_mpwrq_nonlinear Test setup: DUT: Intel Xeon D-1518 (2.20GHz) w/ Nvidia/Mellanox ConnectX-6 Dx 2x100G Data rate measured on switch (Extreme X690), and DUT connected as a router on a stick, with pktgen and pktsink as VLANs. Pktgen was in range 12.79 - 12.95 Mpps across all tests. Signed-off-by: Asbjørn Sloth Tønnesen <ast@fiberby.net> --- include/net/pkt_cls.h | 5 +++++ net/core/dev.c | 3 +++ 2 files changed, 8 insertions(+)