RE: [PATCH V3 1/4] mfd: da9062: DA9062 MFD core driver

[Opensource [Steve Twiss]]

On 11 June 2015 09:57 Lee Jones wrote:

> To: Opensource [Steve Twiss]
> Subject: Re: [PATCH V3 1/4] mfd: da9062: DA9062 MFD core driver
> 
> > > > > diff --git a/include/linux/mfd/da9062/registers.h

[...]

> > > > > +#define DA9062AA_WRITE_MODE_SHIFT	6
> > > > > +#define DA9062AA_WRITE_MODE_MASK	(0x01 << 6)
> > > >
> > > > For 1 << X, you should use BIT(X).
> > > >
> > >
> > > For the two comments above "Registers" and "Bit fields" and the (1<<x)
> > > definitions ...
> > >
> > > The whole of this file is automatically generated by our hardware designers
> > > I would prefer it if the register definitions and bit fields are not altered using
> > > the #define BIT(nr) (1UL<<(nr)) macro and the comments removed because
> > > we have scripts that can be used to check this file automatically.
> > >
> > > Also if the register map is ever updated, then it will be easier for me to diff
> > > the new delivered register and bit field definitions with the old one.
> > >
> > > My preference would be not to change this header file.
> > >
> > > [...]
> >
> > If these last two things are a problem can you please let me know.

HI Lee,

Thanks for replying so quickly.

> I'm still not particularly happy with this.  Can yo speak to your H/W
> guys and get them to change their scripts to output sensible header
> files?

Ah. Ok. 

For our side, the generated headers might not just be used for Linux. I've
just discussed this with my colleagues and they will need it to remain.
So I guess internally we will keep the headers like this, but as it enters
my submission process I can change it for the Linux community.
 
> To be honest, it's probably not a blocker for acceptance, but if someone
> writes a patch next week to change all of the (0x01 << X) lines to
> start using the BIT() macro, I will accept it.  Better to influenced
> your guys so you are not overly inconvenienced.

Yep: I will change this BIT() macro for the submissions in future.
Depending on the next step, I will send a patch to this or update the submission
if there are further comments on this patch set.

> FWIW, when upstreaming code, the excuse "someone else wrote it", has
> never been a good one to use on the lists.   Believe me, I've
> tried. ;)

heh okay :)

Regards,
Steve