Re: [PATCH v8 04/10] reset: eyeq5: add platform driver

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hello,

On Thu Feb 29, 2024 at 12:22 PM CET, Andy Shevchenko wrote:
> On Wed, Feb 28, 2024 at 06:04:47PM +0100, Théo Lebrun wrote:
> > On Tue Feb 27, 2024 at 6:27 PM CET, Andy Shevchenko wrote:
> > > On Tue, Feb 27, 2024 at 03:55:25PM +0100, Théo Lebrun wrote:
>
> ...
>
> > > > +	u32 offset = id & GENMASK(7, 0);
> > > > +	u32 domain = id >> 8;
> > >
> > > Perhaps
> > >
> > > 	u32 offset = (id & GENMASK(7, 0)) >> 0;
> > > 	u32 domain = (id & GENMASK(31, 8)) >> 8;
> > >
> > > for better understanding the split?
> > 
> > Do the additional zero-bit-shift and GENMASK() help understanding? My
> > brain needs time to parse them to then notice they do nothing and
> > simplify the code in my head, back to the original version.
>
> In my opinion yes, as you exactly showing the split.
> But. The better is to use FIELD_GET().

I'll go with the FIELD_GET() option!

[...]

>
> > > > +	priv->rcdev.of_node = np;
> > >
> > > It's better to use device_set_node().
> > 
> > I don't see how device_set_node() can help? It works on struct device
> > pointers. Here priv->rcdev is a reset_controller_dev struct. There are
> > no users of device_set_node() in drivers/reset/.
>
> No users doesn't mean it's good. The API is relatively "new" and takes
> care of two things:
> 1) it uses agnostic interface;
> 2) it doesn't require any firmware node direct dereference.
>
> The 2) is most important here as allows us to refactor (firmware node) code
> in the future.

I think I get the point of device_set_node(). I still do not understand
how it could help me fill the ->of_node field in a reset_controller_dev
structure?

Should I be using device_set_node() to fill the struct device pointer
and the reset subsystem, by some magic, will pick this up and use it
for its own of_node field? I've not seen any magic/code doing that.

[...]

> > > > +		priv->rcdev.nr_resets += __builtin_popcount(eq5r_valid_masks[i]);
> > >
> > > Please, use corresponding hweightXX() API.
> >
> > Noted. I did not find this keyword even though I searched quite a bit
> > for it. "popcount" sounds more logical to me. :-)
>
> Hmm... But it's fundamental, it's called Hamming weight.
> https://en.wikipedia.org/wiki/Hamming_weight

Makes sense now. I've always called it population count following the
name of the matching instruction on x86 (and I believe other ISAs). TIL.

Regards,

--
Théo Lebrun, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com






[Index of Archives]     [Device Tree Compilter]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux PCI Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Yosemite Backpacking]


  Powered by Linux