[lkml]   [2012]   [Jul]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [squeeze] atl1c: AR8152: "transmit queue 0 timed out" and network is unusable until reset
On Wed, Jul 04, 2012 at 01:16:14PM -0500, Jonathan Nieder wrote:
> Jonathan Nieder wrote:
> > Luis R. Rodriguez wrote:
> >> FWIW we already provide daily backports of code through compat-wireless.
> >> compat-wireless will eventually be changed to "compat-drivers" to reflect
> >> that it has drivers backported other than 802.11. We also have stable releases
> >> of the Linux kernel backported for use on older releases.
> >
> > I looked at the relevant repositories and am afraid I am too dim to
> > see how to use them.
> Ok, it's becoming a little clearer now. Is the appropriate procedure
> something like this?
> git clone git://
> git clone git://

git clone git://
git clone git://

> cd compat-wireless
> git checkout linux-3.0.y
> GIT_COMPAT_TREE=$(pwd)/compat/

this ones always assumes $HOME/compat so if you have it there
already there is no need to specify this variable.

> NEXT_TREE=/path/to/src/linux
> GIT_TREE=/path/to/linux/repo

export GIT_TREE=/home/foo/linux-stable.git


I use Greg's linux-stable.git so I can use all the
extra version stable kernels, but on top of that I have
a remote set up to also pull in Linus' junk:

[remote "linus"]
fetch = +refs/heads/*:refs/remotes/origin/*
url = git://

So I just do this on linux-stable:

git fetch linus
git reset --hard origin

Then I get reset --hard v3.5-rc5 given that Linus will typically
have a delta on top of the latest RC.

> scripts/ <some appropriate arguments>

The arguments allows you to specify which delta you want to
suck in, if at all.

> And then this will not generate a list of patches but just a patched
> source code tree with appropriate #ifdefs to make the code build
> against old kernels.

#ifdef'ing around code to provide kernel backporting is a strategy
of the stone age. We have taken a slightly different approach, we
have stuffed as much into a module / headers: compat.git so that the
code can remain as pristine as possible. This also means that you can
backport *more* subsystems with less effort and I've proven this through
a graph which shows the overhead cost of backporting a new subsystem
once you have a lot of code within a shared compat module:

The only #ifdef crap needed really would then just be things that *cannot* be
backported through a module / headers. In turn we've been discovering that some
of this #ifdef'able stuff actually can be represented also using SmPL and that
this SmPL actually represents collateral evolutions of the Linux kernel. So
the patches/ directory represents just this: collateral evolutions of the Linux
kernel. I've started to formally document this slowly. The first one is the
patch with 4 digits:


Jesper Andersen however has written a tool called spdiff which I intend
on using to extract SmPL *from* a patch file ! What this means is if
one collateral evolution *which could not be backported through compat.git*
is backported for *two drivers* it means that we can backport that collateral
evolution for *all drivers*. Furthermore if we design collateral evolutions
on the Linux kernel *with* SmPL it means we also can backport that respective
collateral evolution with the *inverse* of SmPL !

> That sounds useful.

How about with all the other junk I just mentioned ? :)

I should mention that stable releases are already made:

But surely you can also just make your own as well. The stable releases
also are test compiled accross 21 kernels, and soon, thanks to Ozan's
GSoC project I hope we'll have video backported as well.


 \ /
  Last update: 2012-07-05 23:01    [W:0.024 / U:21.004 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site