Re: [Cerowrt-devel] Latest build test - new sqm-scripts seem to work; "cake overhead 40" [still] didn't

Re: [Cerowrt-devel] Latest build test - new sqm-scripts seem to work; "cake overhead 40" [still] didn't

[Alan Jenkins]

On 19/06/2015, Dave Taht <dave.taht@gmail.com> wrote:
> Fresh bits! Get your fresh bits here! (totally untested)
>
> http://snapon.lab.bufferbloat.net/~cero3/lupin/ar71xx/

Good news: it lives!  Yum, fresh bits.

So now I can update and install packages.  Including tc-adv.  It looks
like it was overriding tc already though.

Bad news: I still can't set cake's overhead option (sticks at 0).

# tc -V
tc utility, iproute2-ss4.1.0-git-1-openwrt

# tc qdisc del dev pppoe-wan root
# tc qdisc |grep "dev pppoe-wan"
qdisc fq_codel 0: dev pppoe-wan root refcnt 2 limit 1024p flows 1024
quantum 300 target 5.0ms interval 100.0ms ecn
qdisc ingress ffff: dev pppoe-wan parent ffff:fff1 ----------------

# tc qdisc add dev pppoe-wan root cake bandwidth 970Kbit
besteffort flows atm  overhead 40

# tc qdisc |grep "dev pppoe-wan"
qdisc cake 800d: dev pppoe-wan root refcnt 2 bandwidth 970Kbit
besteffort flows atm overhead 0
qdisc ingress ffff: dev pppoe-wan parent ffff:fff1 ----------------

Alan

Re: [Cerowrt-devel] Latest build test - new sqm-scripts seem to work; "cake overhead 40" [still] didn't

[Jonathan Morton]

[-- Attachment #1: Type: text/plain, Size: 200 bytes --]

I can probably glean more information from "tc -s qdisc".

However, my hypothesis is still that the version of cake you have is old,
while the version of tc that you have is newer.

- Jonathan Morton

[-- Attachment #2: Type: text/html, Size: 265 bytes --]

Re: [Cerowrt-devel] Latest build test - new sqm-scripts seem to work; "cake overhead 40" [still] didn't

[Alan Jenkins]

On 20/06/15 12:00, Jonathan Morton wrote:
>
> I can probably glean more information from "tc -s qdisc".
>
> However, my hypothesis is still that the version of cake you have is 
> old, while the version of tc that you have is newer.
>
> - Jonathan Morton
>

Hi

I missed that, thanks.  (Error checking in qdiscs would be cool, I like 
error messages even that just say -EINVAL).

I can't think of any other way to check, but I had a look in 
ceropackages-3.10.git.  It should be up to date, it fetches the specific 
git commit (by sha1) that's on the head of the cake repo :(.  Strange 
failure on Dave's build if it's not.

I ran sqm-scripts to pass atm/40 to cake.  Here's what that looks like.  
Did you mean me to run some load first, if so what sort?

Regards
Alan


qdisc cake 8011: dev pppoe-wan root refcnt 2 bandwidth 970Kbit 
besteffort flows atm overhead 0
  Sent 850 bytes 9 pkt (dropped 0, overlimits 0 requeues 0)
  backlog 0b 0p requeues 0
            Class 0
   rate       970Kbit
   target      18.7ms
interval     149.8ms
Pk delay         2us
Av delay         0us
Sp delay         0us
   pkts             9
way inds           0
way miss           1
way cols           0
   bytes          850
   drops            0
   marks            0
qdisc ingress ffff: dev pppoe-wan parent ffff:fff1 ----------------
  Sent 1944 bytes 18 pkt (dropped 0, overlimits 0 requeues 0)
  backlog 0b 0p requeues 0
...
qdisc cake 8012: dev ifb4pppoe-wan root refcnt 2 bandwidth 15500Kbit 
besteffort flows atm overhead 0
  Sent 2340 bytes 18 pkt (dropped 0, overlimits 3 requeues 0)
  backlog 0b 0p requeues 0
            Class 0
   rate     15500Kbit
   target       5.0ms
interval     105.0ms
Pk delay       152us
Av delay         2us
Sp delay         2us
   pkts            18
way inds           0
way miss           1
way cols           0
   bytes         2340
   drops            0
   marks            0

Re: [Cerowrt-devel] Latest build test - new sqm-scripts seem to work; "cake overhead 40" [still] didn't

[Jonathan Morton]

[-- Attachment #1: Type: text/plain, Size: 471 bytes --]

It looks like your cake is new enough to support set associative hashing,
but not the new overhead handling. The ATM flag was put in a long time ago.

Looking at the code which grabs those options (cake_change), there doesn't
seem to be a way to detect whether an unsupported option was provided by
userspace, unless nla_parse_nested returns an error if the provided option
struct would be overflowed. Clearly it doesn't, but just truncates it to
fit.

- Jonathan Morton

[-- Attachment #2: Type: text/html, Size: 538 bytes --]

Re: [Cerowrt-devel] Latest build test - new sqm-scripts seem to work; "cake overhead 40" [still] didn't

[Dave Taht]

huh? I pulled from the repo you were committing to, I thought.... is
that not correct?

I can do a build of a newer cake... before I get on the plane...

On Sat, Jun 20, 2015 at 5:46 AM, Jonathan Morton <chromatix99@gmail.com> wrote:
> It looks like your cake is new enough to support set associative hashing,
> but not the new overhead handling. The ATM flag was put in a long time ago.
>
> Looking at the code which grabs those options (cake_change), there doesn't
> seem to be a way to detect whether an unsupported option was provided by
> userspace, unless nla_parse_nested returns an error if the provided option
> struct would be overflowed. Clearly it doesn't, but just truncates it to
> fit.
>
> - Jonathan Morton



-- 
Dave Täht
worldwide bufferbloat report:
http://www.dslreports.com/speedtest/results/bufferbloat
And:
What will it take to vastly improve wifi for everyone?
https://plus.google.com/u/0/explore/makewififast