On Wed, 2015-07-29 at 12:57 +0100, Mark Brown wrote:
> > Are there drivers for which the expected firmware update cycle is *more*
> > rapid than the kernel release cycle?  That would be quite a surprise,
> > though not an unpleasant one.
> 
> Some of the audio ones are like this, especially during development -
> firmware is used to get system specific callibration data (to account
> for the plastics and the taste of the system integator).  Those
> firmwares would need system specific lists which would be miserable.
> 
> The code firmwares themselves do also get updated rather rapidly at
> times.

It's also not about "more rapid than the kernel release cycle", but
just about "not 100% strictly nailed to the kernel release cycle".

In some environments, the "kernel release cycle" is basically non
-existent. If you're running RHEL6, you get to remain with your ancient
kernel.

In a world where the driver requires a specific hashed firmware, they
are *not* going to update that driver unless there's a bloody good
reason to do so.

In a world where the firmware is *signed*, you can update it yourself
as long as the new version is also signed by an acceptable key.

-- 
dwmw2