On Thu, Jun 4, 2015 at 9:19 AM, Kenneth Klette Jonassen
kenne...@ifi.uio.no wrote:
why not just call tcp_init_cwnd_reduction()?
I deferred considering the ECN implications of doing so. The code to
start PRR was based on tcp_enter_cwr()/tcp_init_cwnd_reduction(), save
that both of these
e can get from a few chipsets, but not enough
was known about the iwl last I looked. And one reason why fq_codel -
unassisted - is not quite the right thing on top of this is that
multicast can take a really long time...
Regardless, I'd highly love to see/use this patch myself in a variety
of real w
> A bursted txop can be as big as 5-10ms. If you consider you want to
> queue 5-10ms worth of data for *each* station at any given time you
> obviously introduce a lot of lag. If you have 10 stations you might
> end up with service period at 10*10ms = 100ms. This gets even worse if
> you consider
On Thu, Feb 11, 2016 at 7:05 AM, Grumbach, Emmanuel
wrote:
> fixing linux-wireless address ...
>
> On 02/11/2016 04:30 PM, Eric Dumazet wrote:
>> On Thu, 2016-02-11 at 16:08 +0200, Emmanuel Grumbach wrote:
>>> Signed-off-by: Emmanuel Grumbach
On Thu, Feb 11, 2016 at 7:29 AM, Grumbach, Emmanuel
wrote:
>
>
> On 02/11/2016 05:12 PM, Eric Dumazet wrote:
>> On Thu, 2016-02-11 at 15:05 +, Grumbach, Emmanuel wrote:
>>
>>
>>> Yeah :) codel_should_drop seemed very long indeed... I wanted to use the
>>>
>> regular fq_codel uses 1024 and there has not been much reason to
>> change it. In the case of an AP which has more limited memory, 256 or
>> 1024 would be a good setting, per station. I'd stick to 1024 for now.
>
> Do note that the 4096 is shared _across_ station-tid queues. It is not
>
it is helpful to name the test files coherently in the flent tests, in
addition to using a directory structure and timestamp. It makes doing
comparison plots in data->add-other-open-data-files simpler. "-t
patched-mac-300mbps", for example.
Also netperf from svn (maybe 2.7, don't remember) will
of my sloppy proof-of-concept change in ath10k).
So I should avoid ben greer's firmware for now?
>
>
> Michał
>
> On 16 March 2016 at 20:48, Jasmine Strong <j...@eero.com> wrote:
>> BK usually has 0 txop, so it doesn't do aggregation.
>>
>> On Wed, Mar
Dave Täht
Let's go make home routers and wifi faster! With better software!
https://www.gofundme.com/savewifi
On Tue, Mar 1, 2016 at 11:38 PM, Michal Kazior wrote:
> On 1 March 2016 at 15:02, Johannes Berg wrote:
>> On Fri, 2016-02-26 at 14:09 +0100, Michal Kazior wrote:
>>>
>>> +typedef u64 codel_time_t;
>>
>> Do we really need this? And if yes, does
On Mon, Mar 7, 2016 at 9:14 AM, Avery Pennarun <apenw...@gmail.com> wrote:
> On Mon, Mar 7, 2016 at 11:54 AM, Dave Taht <dave.t...@gmail.com> wrote:
>> If I can just get a coherent patch set that I can build, I will gladly
>> join you on the testing ath9k in particula
If I can just get a coherent patch set that I can build, I will gladly
join you on the testing ath9k in particular... can probably do ath10k,
too - and do a bit of code review... this week. it's very exciting
watching all this activity...
but I confess that I've totally lost track of what set of
RESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
> + * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
> + * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
> + * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
> + * SPECI
On Mon, May 16, 2016 at 1:14 AM, Roman Yeryomin <leroi.li...@gmail.com> wrote:
> On 16 May 2016 at 01:34, Roman Yeryomin <leroi.li...@gmail.com> wrote:
>> On 6 May 2016 at 22:43, Dave Taht <dave.t...@gmail.com> wrote:
>>> On Fri, May 6, 2016 at 11:56 AM, R
I can't even describe how much I hate the concept of the reorder
buffer in general. Ordering is the endpoints problem.
Someday, after we get fq_codeled, short queues again, I'll be able to show why.
On Mon, May 16, 2016 at 4:41 AM, Luca Coelho wrote:
> On Fri, 2016-05-13 at
On Fri, May 6, 2016 at 11:56 AM, Roman Yeryomin wrote:
> On 6 May 2016 at 21:43, Roman Yeryomin wrote:
>> On 6 May 2016 at 15:47, Jesper Dangaard Brouer wrote:
>>>
>>> I've created a OpenWRT ticket[1] on this issue, as it seems
I have duplicated the issue on my own hardware. I would like to
explore also upping the codel count in this scenario at some point,
but:
Acked-by: Dave Taht
The openwrt tree has long contained a set of patches that correct for
unaligned issues throughout the linux network stack.
In the case of wifi I have 3 issues with this line of thought.
multicast in wifi has generally supposed to be unreliable. This makes
it reliable. reliability comes at a cost -
multicast is typically set at a fixed low rate today. unicast is
retried at different rates until it succeeds - for
On Tue, Jan 10, 2017 at 9:23 AM, Felix Fietkau <n...@nbd.name> wrote:
> On 2017-01-10 18:17, Dave Taht wrote:
>> In the case of wifi I have 3 issues with this line of thought.
>>
>> multicast in wifi has generally supposed to be unreliable. This makes
>> it reli
> netem: remove unnecessary 64 bit modulus
>>
>> net/sched/sch_netem.c | 17 +++--
>> 1 file changed, 7 insertions(+), 10 deletions(-)
>>
>
> Acked-by: Randy Dunlap <rdun...@infradead.org>
Acked-by: Dave Taht <dave.t...@gmail.com>
Thx.
Add support and man page for cake qdisc.
Dave Taht (1):
Add support for cake qdisc
man/man8/tc-cake.8 | 678 ++
tc/Makefile| 1 +
tc/q_cake.c| 771 +
3 files changed, 1450
Hook up sch_cake to the build system.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
net/sched/Kconfig | 11 +++
net/sched/Makefile | 1 +
2 files changed, 12 insertions(+)
diff --git a/net/sched/Kconfig b/net/sched/Kconfig
index c03d86a..3ea22e5 100644
--- a/net/sched/K
3
un_flows 0 0 0
max_len 302830283028
Signed-off-by: Dave Taht <dave.t...@gmail.com>
Tested-by: Pete Heist <petehe...@gmail.com>
Tested-by: Georgios Amanakis <gamana...@gmail.com>
---
include/net/cobalt.h |
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
include/uapi/linux/pkt_sched.h | 58 ++
1 file changed, 58 insertions(+)
diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
index af3cc2f..ed7c111 100644
--- a/include/uapi
ferbloat.net mailing list.
Dave Taht (3):
pkt_sched.h: add support for sch_cake API
Add Common Applications Kept Enhanced (cake) qdisc
Add support for building the new cake qdisc
include/net/cobalt.h | 152 +++
include/uapi/linux/pkt_sched.h | 58 +
net/sched/Kconfig
ferbloat.net mailing list.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
man/man8/tc-cake.8 | 678 ++
tc/Makefile| 1 +
tc/q_cake.c| 771 +
3 files changed, 1450 insertions(
I have been thinking we'd try to submit sch_cake to mainline on this
go-around. It's been out of tree for way too long.
I look forward to understanding your patches soon in the tbf case.
(I'm only responding because cake uses deficit, rather than a token
bucket, scheduler, and is not reliant on
On Wed, Nov 1, 2017 at 9:04 AM, Cong Wang <xiyou.wangc...@gmail.com> wrote:
> On Tue, Oct 31, 2017 at 11:42 PM, Dave Taht <dave.t...@gmail.com> wrote:
>> I am using a fairly complex htb + three tiers of fq_codel and a couple
>> tc filters (it's the sqm-scripts &qu
On Thu, Nov 2, 2017 at 11:09 AM, Cong Wang <xiyou.wangc...@gmail.com> wrote:
> On Wed, Nov 1, 2017 at 1:17 PM, Dave Taht <dave.t...@gmail.com> wrote:
>>
>> That is not in net-next, and the "net" version of that one patch does
>> not apply to net-next
transition where reducing a tick to nsec equivalence
would constrain the max delay in prior versions of netem to only 4.3
seconds.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
include/uapi/linux/pkt_sched.h | 2 ++
net/sched/sch_netem.c | 16
2 files change
diameters of several hundred light-years. However, only
conversions to and from ns, us, ms, and seconds are provided.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
tc/tc_core.h | 3 +++
tc/tc_util.c | 56
tc/tc_util.h | 3
rate, and 20ms path
delay:
tc qdisc add dev eth0 root handle 1: netem delay 20ms rate 200mbit \
limit 1
tc qdisc add dev eth0 parent 1:1 handle 10:1 netem delay 200us \
slot 800us 10ms bytes 64k packets 42 limit 512
Signed-off-by: Dave Taht <dave.t...@
This starts to obsolete the old "ticks" api in favor of well
defined nanoseconds for the kernel/userspace netem interface.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
tc/q_netem.c | 68 +++-
1 file changed, 44 i
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
man/man8/tc-netem.8 | 32 +++-
1 file changed, 31 insertions(+), 1 deletion(-)
diff --git a/man/man8/tc-netem.8 b/man/man8/tc-netem.8
index b31384f..8c41d77 100644
--- a/man/man8/tc-netem.8
+++ b/man/man8/tc-n
This patch series adds support for specifying time in nanoseconds
to tc, updates netem to use such (requires pkt_sched.h from the kernel
patch series to be imported), and lastly adds a new slotting
feature intended to emulate better how LTE and WiFi work.
Dave Taht (4):
tc: support conversions
ying rate, and 20ms path
delay:
tc qdisc add dev eth0 root handle 1: netem delay 20ms rate 200mbit \
limit 1
tc qdisc add dev eth0 parent 1:1 handle 10:1 netem delay 200us \
slot 800us 10ms bytes 64k packets 42 limit 512
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
Upgrade the internal netem scheduler to use nanoseconds rather than
ticks throughout.
Convert to and from the std "ticks" userspace api automatically,
while allowing for finer grained scheduling to take place.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
net/sched/
This patch series converts netem away from the old "ticks" interface and
userspace API, and adds support for a new "slot" feature intended to
emulate bursty macs such as WiFi and LTE better.
Dave Taht (3):
netem: convert to qdisc_watchdog_schedule_ns
netem: add u
This patch series converts netem away from the old "ticks" interface and
userspace API, and adds support for a new "slot" feature intended to
emulate bursty macs such as WiFi and LTE better.
Changes since v1:
Always pass new nanosecond APIs to userspace
Dave Taht (
On Tue, Nov 7, 2017 at 4:26 PM, Stephen Hemminger
<step...@networkplumber.org> wrote:
> On Tue, 7 Nov 2017 12:59:33 -0800
> Dave Taht <dave.t...@gmail.com> wrote:
>
>> This patch series converts netem away from the old "ticks" interface and
>> user
Upgrade the internal netem scheduler to use nanoseconds rather than
ticks throughout.
Convert to and from the std "ticks" userspace api automatically,
while allowing for finer grained scheduling to take place.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
net/sched/
On Wed, Nov 8, 2017 at 3:24 PM, Stephen Hemminger
<step...@networkplumber.org> wrote:
> On Wed, 8 Nov 2017 15:12:27 -0800
> Dave Taht <dave.t...@gmail.com> wrote:
>
>> --- a/net/sched/sch_netem.c
>> +++ b/net/sched/sch_netem.c
>> @@ -819,6 +819,8 @@ static c
ying rate, and 20ms path
delay:
tc qdisc add dev eth0 root handle 1: netem delay 20ms rate 200mbit \
limit 1
tc qdisc add dev eth0 parent 1:1 handle 10:1 netem delay 200us \
slot 800us 10ms bytes 64k packets 42 limit 512
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
Upgrade the internal netem scheduler to use nanoseconds rather than
ticks throughout.
Convert to and from the std "ticks" userspace api automatically,
while allowing for finer grained scheduling to take place.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
net/sched/
nd,q->slot.slot_next)
Changes since v1:
Always pass new nanosecond APIs to userspace
Dave Taht (3):
netem: convert to qdisc_watchdog_schedule_ns
netem: add uapi to express delay and jitter in nanoseconds
netem: support delivering packets in delayed time slots
include/uapi/linux/pkt_sched
transition where reducing a tick to nsec
equivalence would constrain the max delay in prior versions of
netem to only 4.3 seconds.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
Suggested-by: Eric Dumazet <eduma...@google.com>
Reviewed-by: Eric Dumazet <eduma...@google.com>
---
in
ying rate, and 20ms path
delay:
tc qdisc add dev eth0 root handle 1: netem delay 20ms rate 200mbit \
limit 1
tc qdisc add dev eth0 parent 1:1 handle 10:1 netem delay 200us \
slot 800us 10ms bytes 64k packets 42 limit 512
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
transition where reducing a tick to nsec
equivalence would constrain the max delay in prior versions of
netem to only 4.3 seconds.
Signed-off-by: Dave Taht <dave.t...@gmail.com>
---
include/uapi/linux/pkt_sched.h | 2 ++
net/sched/sch_netem.c | 14 ++
2 files change
, 2017 at 2:56 AM, Jesper Dangaard Brouer
<bro...@redhat.com> wrote:
>
> On Sun, 03 Dec 2017 20:19:33 -0800 Dave Taht <d...@taht.net> wrote:
>
>> Changing the topic, adding bloat.
>
> Adding netdev, and also adjust the topic to be a rant on that the Linux
> kernel
I am using a fairly complex htb + three tiers of fq_codel and a couple
tc filters (it's the sqm-scripts "simple.qos" model). I rebased on
net-next head an hour ago, and
[8.357963] IPv6: ADDRCONF(NETDEV_CHANGE): enp2s0: link becomes ready
[9.759942] u32 classifier
[9.759944]
very happy to see this. is there a specific chip or devboard this runs on?
On Tue, Jun 19, 2018 at 11:24 AM, Li Yang wrote:
> On Tue, Jun 19, 2018 at 11:30 AM, Joakim Tjernlund
> wrote:
>> Signed-off-by: Joakim Tjernlund
>
> Acked-by: Li Yang
>
>> ---
>>
On Tue, May 1, 2018 at 11:53 AM, Toke Høiland-Jørgensen <t...@toke.dk> wrote:
> Eric Dumazet <eric.duma...@gmail.com> writes:
>
>> On 04/30/2018 02:27 PM, Dave Taht wrote:
>>
>>> I actually have a tc - bpf based ack filter, during the development of
>>&
On Mon, Apr 30, 2018 at 2:21 PM, Cong Wang wrote:
> On Sun, Apr 29, 2018 at 2:34 PM, Toke Høiland-Jørgensen wrote:
>> sch_cake targets the home router use case and is intended to squeeze the
>> most bandwidth and latency out of even the slowest ISP links
, MA
+ * 02110-1301, USA.
+ */
+
+/*
+ * Author: dave.t...@gmail.com (Dave Taht)
+ *
+ * ack_recognizer: An eBPF program that correctly recognizes modern TCP ACKs,
+ * with tcp option fields like SACK and timestamps, and no additional data.
+ *
+ * ack_match call: Recognize "pure acks" wi
On Tue, May 1, 2018 at 9:12 PM, David Ahern <dsah...@gmail.com> wrote:
> On 5/1/18 12:32 PM, Dave Taht wrote:
>> ack_recognize can shift pure tcp acks into another flowid.
>> ---
>> examples/bpf/ack_recognize.c | 98
>> ++
On Fri, Aug 3, 2018 at 10:50 AM Matteo Croce wrote:
>
> When 'ip link show dev DEVICE' is processed in a batch mode, ip exits
> and stop processing further commands.
> This because ipaddr_list_flush_or_save() calls exit() to avoid printing
> the link information twice.
> Replace the exit with a
-by: Dave Taht
---
tc/q_cake.c | 9 +
1 file changed, 9 insertions(+)
diff --git a/tc/q_cake.c b/tc/q_cake.c
index f1e232a..727d673 100644
--- a/tc/q_cake.c
+++ b/tc/q_cake.c
@@ -79,6 +79,7 @@ static void explain(void)
" dual-srchost | dual-dsthost | triple-isolate
On Sun, Aug 12, 2018 at 3:09 PM David Ahern wrote:
>
> On 8/6/18 11:09 AM, Yousuk Seung wrote:
> > diff --git a/tc/tc_core.h b/tc/tc_core.h
> > index 1dfa9a4f773b..a0fe0923d171 100644
> > --- a/tc/tc_core.h
> > +++ b/tc/tc_core.h
> > @@ -7,6 +7,10 @@
> >
> > #define TIME_UNITS_PER_SEC 100
On Wed, Aug 8, 2018 at 10:00 AM Andrew Lunn wrote:
>
> On Wed, Aug 08, 2018 at 05:27:05PM +0200, Marek Behún wrote:
> > For some reason on Armada 3720 boards (EspressoBin and Turris Mox) the
> > networking driver behaves weirdly when using napi_gro_receive.
> >
> > For example downloading a big
On Mon, Aug 27, 2018 at 9:11 AM Stephen Hemminger
wrote:
>
> On Sun, 26 Aug 2018 19:42:28 -0700
> Yousuk Seung wrote:
>
> > +int get_time(unsigned int *time, const char *str)
> > +{
> > + double t;
> > + char *p;
> > +
> > + t = strtod(str, );
> > + if (p == str)
> > +
In my dreamworld, a packet with a timestamp achieved at rx time on
ethX, or via local traffic, would be consistent with the right clock
throughout the system and reliably still be there when it goes to
ethY.
This would save having to timestamp (again) inside the cb block in
fq_codel, cake, etc,
On Tue, Jul 17, 2018 at 2:16 AM Paolo Abeni wrote:
>
> Hi,
>
> On Mon, 2018-07-16 at 16:39 -0700, Cong Wang wrote:
> > On Fri, Jul 13, 2018 at 2:55 AM Paolo Abeni wrote:
> > >
> > > When mirred is invoked from the ingress path, and it wants to redirect
> > > the processed packet, it can now use
On Tue, Mar 13, 2018 at 11:24 AM, Jakob Unterwurzacher
wrote:
> During stress-testing our "ucan" USB/CAN adapter SocketCAN driver on Linux
> v4.16-rc4-383-ged58d66f60b3 we observed that a small fraction of packets are
> delivered out-of-order.
>
> We
On Mon, Aug 27, 2018 at 9:39 AM Dave Taht wrote:
>
> On Mon, Aug 27, 2018 at 9:11 AM Stephen Hemminger
> wrote:
> >
> > On Sun, 26 Aug 2018 19:42:28 -0700
> > Yousuk Seung wrote:
> >
> > > +int get_time(unsigned int *time, const char *str)
> >
65 matches
Mail list logo