Re: [PATCH 3.7] spi/bcm63xx: fix transfer bits_per_words check

[Florian Fainelli <florian@openwrt.org>]

On Saturday 24 November 2012 19:19:34 Jonas Gorski wrote:
> On 24 November 2012 19:01, Mark Brown
> <broonie@opensource.wolfsonmicro.com> wrote:
> > On Sat, Nov 24, 2012 at 06:53:10PM +0100, Jonas Gorski wrote:
> >> On 24 November 2012 18:28, Mark Brown
> >
> >> > This is now done by the SPI core, there's no need for individual drivers
> >> > to do this.
> >
> >> So that will go to stable kernels, too?
> >
> > No, I hadn't sent it there.  I'd really expect that anyone using a
> > release kernel would've noticed this if there were a problem, it's
> > pretty obvious when it goes wrong, and I tend to be extremely
> > conservative with changes to stable kernels especially framework ones.
> 
> Well, I noticed this particular problem on a 3.6 kernel (3.6.7 to be
> exact), so by that definition at least one noticed, unless I don't
> count as anyone ;-). So, how can I make this driver less broken for
> release kernels? I somewhat doubt the framework change will be picked
> up by release kernel maintainers.

Considering it is a simple one-liner in spi-bcm63xx and that you don't want to
propagate a framework-level change back to -stable (which makes sense), don't
you want to pick that one?

At least two other users told me about this bug in private conversations.
--
Florian