[openib-general] Re: was [PATCH] enable the fmr pool user to set the page size - 2 Q on upstream pushes

[Muli Ben-Yehuda]

On Wed, Jan 18, 2006 at 05:36:05PM +0200, Or Gerlitz wrote:

> first, what does it means that the 2.6.16-rc1 window was missed?

Current kernel development follows a model where there's an
approximately two week merge window after each major kernel release
(e.g., 2.6.15). Then follow approximately 6 weeks of stabilizing the
massive ammount of code that went in during the merge window.

>  what 
> changes makes sense for non -rc1 releases? 

Theoretically anything can go in in -rc1 (assuming it is appropriate
to go in in the first place). Other rc's have much stricter criteria
for what can go in. It's the maintainer's decision what should go in.

> is it correct that the parity 
> of the X minor number in 2.6.X relates to which changes can be pushed? 
> that is for even X (eg 2.6.14/16) only bug fixes and for odd X (eg 
> 2.6.15/17) also new features (eg srp & open-iscsi "waited" for
> 2.6.15)?

No. Once upon a time odd numbered kernels (e.g., 2.3.X, 2.5.X)
were experimental and even numbered kernel (2.2.X, 2.4.X) were
stable. Nowdays all 2.6.X kernels are equally experimental or stable,
depending on your point of view.

> Is there an FAQ explaining all this...?

Not really, since it's a constantly changing constantly evolving
process. This stuff is usually discussed on the kernel mailing
list, and summarized or one of the many summaries (kerneltraffic,
lwn.net's weekly kernel page).

Cheers,
Muli
-- 
Muli Ben-Yehuda
http://www.mulix.org | http://mulix.livejournal.com/