Re: [RFC PATCH 0/3] mailbox: hisilicon: add mailbox driver

[Leo Yan <leo.yan-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>]

Hi Jassi,

On Mon, Aug 10, 2015 at 12:32:05PM +0530, Jassi Brar wrote:
> On Thu, Aug 6, 2015 at 4:47 AM, Leo Yan <leo.yan-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
> > Hi Jassi,
> >
> > Thanks for review.
> >
> > On Wed, Aug 05, 2015 at 04:22:01PM +0530, Jassi Brar wrote:
> >> On Mon, Aug 3, 2015 at 6:43 AM, Leo Yan <leo.yan-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
> >
> > [...]
> >
> >> > For easily extending for Hisilicon series SoCs (SoCs may have difference
> >> > for register's definition with each other), so firstly implement common
> >> > mailbox driver; this common mailbox driver provides three mainly
> >> > functionality:
> >> >
> >> >  - help register channels into framework;
> >> >  - hook low level callback functions for register's operations;
> >> >  - Enhance rx channel's message queue, which is based on the code in
> >> >    drivers/mailbox/omap-mailbox.c.
> >> >
> >> Not cool.
> >>  Please don't reinvent the wheel by having platform specific
> >> implementation of the mailbox api. Which vendor doesn't plan to roll
> >> out new SoCs, and hence variations of mailbox controllers?  The OMAP
> >> stack predates the common api, and was actually supposed to be
> >> converted over eventually. Please implement just the
> >> drivers/mailbox/hi6220-mailbox.c (preferably by the name of the
> >> mailbox controller, if any)
> >
> > Understood. Here i have one question, the rx channel's message queue is
> > looks like a common mechanism and can be added into framework file
> > mailbox.c, then Soc driver file can _ONLY_ focus on register level's
> > operations. If so, the common driver in this patch also is unnecessary.
> >
> Yes, that's what I say, no 'common' driver for a platform.
> 
> > Do you suggest to use upper method to rework patches? Or just think
> > it's okay to implement rx channel's message queue in hi6220-mailbox.c?
> >
> The code in drivers/mailbox/ should only manage the controller
> (registers and interrupts). Everything else (queues, shmem etc) should
> be in platform specific client driver(s).

Thanks for suggestion, will send new version patches for reviewing.

Thanks,
Leo Yan
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html