[Cerowrt-devel] did a chaos calmer openwrt build, untested

[Cerowrt-devel] did a chaos calmer openwrt build, untested

[Dave Taht]

[-- Attachment #1: Type: text/plain, Size: 1380 bytes --]

I folded in the following:

  * minstrel-blues
  * minstrel patches for minimum variance (hurts udp, helps tcp)
  * babel-1.6.0 release candidate
  * cake and the usual slew of alternate bufferbloat-fighting qdiscs

I intended to use this build to test my ubnt gear and the archer c7v2. It
is NOT cerowrt...

http://snapon.lab.bufferbloat.net/~cero3/ubnt/ar71xx/

1) But before I even got so far as trying that build, ran across this
ticket.

https://dev.openwrt.org/ticket/19194

2) Is the consensus here we should resume using the "device naming by
security model" mode?

3) Things I would like to be developing and then testing
  * latest dnsmasq code (with fixed date getting stuff to hopefully finally
make it up to openwrt)
  * latest FIB patches backported to 3.18 (or, possibly, moving to 4.0)
  * babel with atomic routing updates
  * cake, etc.
  * bobbie (policer)
  * ???

4) Generic problem:

We need some form of autodetection of whether or not IPV6_SUBTREES is
enabled in the kernel (whether in userspace or with additional kernel
support for making the query - perhaps with kernel support) -
presently that is a compile time only option in babel and I'd like it to be
the
default.

-- 
Dave Täht
Let's make wifi fast, less jittery and reliable again!

https://plus.google.com/u/0/107942175615993706558/posts/TVX3o84jjmb

[-- Attachment #2: Type: text/html, Size: 1956 bytes --]

Re: [Cerowrt-devel] did a chaos calmer openwrt build, untested

[Dave Taht]

[-- Attachment #1: Type: text/plain, Size: 1430 bytes --]

On Mon, Mar 16, 2015 at 10:23 AM, Steven Barth <steven@midlink.org> wrote:

>
>
>> 3) Things I would like to be developing and then testing
>>   * latest dnsmasq code (with fixed date getting stuff to hopefully
>> finally make it up to openwrt)
>>
> Interesting are there any updates on this? I haven't followed this issue
> lately.


time getting still borked in dnsmasq-full so far as I know in mainline
openwrt.



>
>> We need some form of autodetection of whether or not IPV6_SUBTREES is
>> enabled in the kernel (whether in userspace or with additional kernel
>> support for making the query - perhaps with kernel support) -
>> presently that is a compile time only option in babel and I'd like it to
>> be the
>> default.
>>
> OpenWrt relies on subtrees to be enabled since it creates source-dest
> routes by default.
>

I would like the IPV6_SUBTREES facility to be enabled by default in
babel-1.6 generally, for distros that might not have got that feature
enabled. While I am pretty sure ubuntu, debian, and fedora have got this
option right, others may not, as yet.

And I have been trying to set aside the time for 2 years now to get atomic
updates working, with no such luck. Only thing I see that does it "right"
is ip route change.

-- 
Dave Täht
Let's make wifi fast, less jittery and reliable again!

https://plus.google.com/u/0/107942175615993706558/posts/TVX3o84jjmb

[-- Attachment #2: Type: text/html, Size: 2412 bytes --]

Re: [Cerowrt-devel] did a chaos calmer openwrt build, untested

[Felix Fietkau]

On 2015-03-16 17:58, Dave Taht wrote:
> I folded in the following:
> 
>   * minstrel-blues
As far as I know, this is not really "minstrel-blues", this is a minor
set of minstrel tweaks.

>   * minstrel patches for minimum variance (hurts udp, helps tcp)
Is this the same patches that you sent a while back?

>   * babel-1.6.0 release candidate
>   * cake and the usual slew of alternate bufferbloat-fighting qdiscs
> 
> I intended to use this build to test my ubnt gear and the archer c7v2.
> It is NOT cerowrt...
> 
> http://snapon.lab.bufferbloat.net/~cero3/ubnt/ar71xx/
Where's the source? I can't find it on your github...

> 1) But before I even got so far as trying that build, ran across this
> ticket.
> 
> https://dev.openwrt.org/ticket/19194
Haven't seen any similar reports from other people yet, might be a
device specific quirk or broken hw or something, I don't know yet.

> 2) Is the consensus here we should resume using the "device naming by
> security model" mode?
> 
> 3) Things I would like to be developing and then testing
>   * latest dnsmasq code (with fixed date getting stuff to hopefully
> finally make it up to openwrt)
>   * latest FIB patches backported to 3.18 (or, possibly, moving to 4.0)
The FIB patches that went into 4.0 are already in OpenWrt trunk ;)

- Felix

Re: [Cerowrt-devel] did a chaos calmer openwrt build, untested

[Dave Taht]

[-- Attachment #1: Type: text/plain, Size: 1957 bytes --]

On Mon, Mar 16, 2015 at 11:53 AM, Felix Fietkau <nbd@openwrt.org> wrote:

> On 2015-03-16 17:58, Dave Taht wrote:
> > I folded in the following:
> >
> >   * minstrel-blues
> As far as I know, this is not really "minstrel-blues", this is a minor
> set of minstrel tweaks.
>
> >   * minstrel patches for minimum variance (hurts udp, helps tcp)
> Is this the same patches that you sent a while back?
>
> Yes. They did leverage a function call from the above.


> >   * babel-1.6.0 release candidate
> >   * cake and the usual slew of alternate bufferbloat-fighting qdiscs
> >
> > I intended to use this build to test my ubnt gear and the archer c7v2.
> > It is NOT cerowrt...
> >
> > http://snapon.lab.bufferbloat.net/~cero3/ubnt/ar71xx/
> Where's the source? I can't find it on your github...
>
>
One of my goals is to work more closely within openwrt's structures this
time around,
I only did that build because half my campus network is breaking on a
regular basis and I have *gotta*  fix that.

> 1) But before I even got so far as trying that build, ran across this
> > ticket.
> >
> > https://dev.openwrt.org/ticket/19194
> Haven't seen any similar reports from other people yet, might be a
> device specific quirk or broken hw or something, I don't know yet.
>
> > 2) Is the consensus here we should resume using the "device naming by
> > security model" mode?
> >
> > 3) Things I would like to be developing and then testing
> >   * latest dnsmasq code (with fixed date getting stuff to hopefully
> > finally make it up to openwrt)
> >   * latest FIB patches backported to 3.18 (or, possibly, moving to 4.0)
> The FIB patches that went into 4.0 are already in OpenWrt trunk ;)
>

Would like to update the ar71xx code also, if possible?


> - Felix
>



-- 
Dave Täht
Let's make wifi fast, less jittery and reliable again!

https://plus.google.com/u/0/107942175615993706558/posts/TVX3o84jjmb

[-- Attachment #2: Type: text/html, Size: 3446 bytes --]