Re: Linux stable inclusion question

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi,

On Wed, Feb 27, 2013 at 07:17:09PM -0800, Greg KH wrote:
> On Thu, Feb 28, 2013 at 10:49:21AM +0800, zhangwei(Jovi) wrote:
> > On 2013/2/28 10:28, Greg KH wrote:
> > > On Thu, Feb 28, 2013 at 10:10:51AM +0800, zhangwei(Jovi) wrote:
> > >> Hi,
> > >>
> > >> I'm writing this mail to ask one question about Linux stable inclusion,
> > >> I would appreciate if you can give me some hints.
> > >>
> > >> As our known:
> > >> Grey KH is maintainer of 3.4 stable tree,
> > > 
> > > And 3.0, and 3.8 :)
> > > 
> > >> Paul is maintainer of 2.6.34 stable tree,
> > >> Willy is maintainer of 2.6.27 stable tree.
> > >>
> > >> Is there have any possibility some stable patches merged into 3.4 stable tree,
> > >> but _missed_ in 2.6.34 and 2.6.27 stable tree? since each stable tree is maintained
> > >> by different person.
> > >> (the stable patches here I mean could applied to each stable tree correctly)
> > > 
> > > Sure, we might miss patches, that's just the nature of the business.  If
> > > you notice any that are missed, please let us know.
> > Hmm, I think all stable maintainers use same channel, stable@xxxxxxxxxxxxxxx,
> 
> Yes.
> 
> > if all maintainers watch all patches in this mailing list, then check it if the patch
> > could apply to own stable tree cleanly, then this will avoid miss patches.
> > (what I mean is we can share all stable patches among stable kernel tree)
> 
> Yes, we try to do this, but of course, we are human, and miss things at
> times.

I would go further, we deliberately skip some patches that we think are
irrelevant or are likely to cause more harm than fix issues. The further
we go back in versions, the less relevant patches are. I would say that
about 2/3 of 3.0 patches don't apply to 2.6.32 at all and half of the
rest has to be applied by hand with careful inspection. So the risk of
adding mistakes is huge. From time to time, we have to ask subsystem
maintainers to perform the backport because they're more likely to do
it correctly. So for minor fixes, we prefer to simply drop them or wait
for someone to request them.  

And despite this extreme care, I can say that every time I release a new
kernel I have to issue a second one the week after because I broke
something ! This has been the case with 2.4, 2.6.27 and 2.6.32.

> > this is not the workflow of stable maintainers currently? or if this
> > workflow is too hard to make in reality?
> 
> No, you are correct, it is just that none of us guarantee anything :)
> 
> What we can do, is use the help of people like you to ensure that we
> don't miss patches.  Can you help us out with this?

I totally agree with you Greg. My best source of inspiration for patches
to backport are users. I have a specific queue for user requests, none of
the patches they request is skipped. And for the rest I pick from 3.0 and
2.6.34 what I consider relevant for 2.6.32, and I feel free to drop what
I can't merge and does not seem important.

Regards,
Willy

--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html


[Index of Archives]     [Linux Kernel]     [Kernel Development Newbies]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite Hiking]     [Linux Kernel]     [Linux SCSI]