Re: [PATCHv3 0/3] ftm-pwm: Convert to direct regmap API usage

[Thierry Reding <thierry.reding@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2433 bytes --]

On Mon, Aug 18, 2014 at 08:43:56AM +0000, Li.Xiubo@freescale.com wrote:
> > -----Original Message-----
> > From: Thierry Reding [mailto:thierry.reding@gmail.com]
> > Sent: Monday, August 18, 2014 4:32 PM
> > To: Xiubo Li-B47053
> > Cc: linux-pwm@vger.kernel.org; robh+dt@kernel.org; pawel.moll@arm.com;
> > rdunlap@infradead.org; devicetree@vger.kernel.org; linux-doc@vger.kernel.org
> > Subject: Re: [PATCHv3 0/3] ftm-pwm: Convert to direct regmap API usage
> > 
> > On Mon, Aug 18, 2014 at 11:32:32AM +0800, Xiubo Li wrote:
> > >
> > > This is depended on the following patches, which has been applied by
> > > Mark into his Regmap-Tree:
> > > https://lkml.org/lkml/2014/7/15/6
> > > https://lkml.org/lkml/2014/7/15/5
> > > https://lkml.org/lkml/2014/7/15/7
> > 
> > This doesn't seem to have any build dependencies on the above patches.
> 
> Yes, you are right.
> 
> 
> > So with only this applied to the PWM tree at worst the registers will be
> > accessed in the wrong endianess and the PWM won't work as expected. Does
> > that matter? 
> 
> No, I think it's okay, we'll only test our drivers in the Linus's stable branch.
> 
> 
> 
> > Or should I request a stable branch from Mark to pull into
> > the PWM tree so that this doesn't happen?
> > 
> > In linux-next things should be fine since both branches will be merged.
> > I guess there could still be an issue depending on the order in which
> > Linus merges the trees for 3.18, in which case there could be a brief
> > period where Linus' tree runtime-breaks on platforms requiring this. But
> > I'm not sure if that's really an issue.
> > 
> 
> The depended patches has been merged into the Linux-next tree just now already
> today.
> 
> And then could we guarantee that these ftm-pwm patches to be in order after the
> depended ones ?

I could merge Mark's regmap tree into the PWM tree to guarantee that.
I'm just wondering if it's worth the trouble since there's no compile-
time dependency so the issue will only show up when running a kernel
built directly from the PWM tree (or some tree that has the PWM tree
merged but not regmap). And even then it will only happen if there's a
device tree for a device that requires special endianess handling.

But it's really your call. If you think it's safer to merge them in the
correct order then I'll ask Mark for a stable regmap branch that I can
pull.

Thierry

[-- Attachment #2: Type: application/pgp-signature, Size: 819 bytes --]