Re: [PATCH v4] powerpc/mpc85xx: Update the clock device tree nodes

On Thu, 2013-09-12 at 21:50 -0500, Tang Yuantian-B29983 wrote:

quoted

-----Original Message-----
From: Wood Scott-B07421
Sent: 2013年9月12日 星期四 22:44
To: Tang Yuantian-B29983
Cc: Wood Scott-B07421; galak-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org; linuxppc-
dev-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org; devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org; Li Yang-Leo-R58472
Subject: Re: [PATCH v4] powerpc/mpc85xx: Update the clock device tree
nodes

On Wed, 2013-09-11 at 20:31 -0500, Tang Yuantian-B29983 wrote:

quoted

quoted

-----Original Message-----
From: Wood Scott-B07421
Sent: 2013年9月12日 星期四 9:10
To: Tang Yuantian-B29983
Cc: galak-XVmvHMARGAS8U2dJNN8I7kB+6BGkLq7r@public.gmane.org; linuxppc-dev-uLR06cmDAlY/bJ5BZ2RsiQ@public.gmane.org;
devicetree-u79uwXL29TY76Z2rM5mHXA@public.gmane.org; Li Yang-Leo-R58472
Subject: Re: [PATCH v4] powerpc/mpc85xx: Update the clock device
tree nodes

This description of "reg" is overly specific (assumes how the parent
node's ranges are set up), incomplete (there's a size as well as the
offset), and does not apply to the clockgen node itself (you
probably shouldn't lump them together like this).

Do you mean I should explain the REG of clockgen and its child node

respectively?

quoted

quoted

quoted

+- clocks : shall be the input parent clock phandle for the clock.

Not required on the clockgen node

Required by child node of clockgen.

My point is that you're lumping several different types of nodes together
with one binding, when some parts of the binding are not applicable to
the clockgen node.

Not several, just two types of nodes.
One is clockgen node, the other is PLL and mux nodes.

clockgen + PLL + mux = 3 = several :-)

The reason they lumped together is that the clockgen node is not only IP block
Node but also a clock provider node

I don't understand why that merits lumping them together.

Just describe them separately.

At first, I want to add a extra fixed-clock node and move the clock-frequency of clockgen 
Node to it, but it is against the backward compatibility

Right.

which I think it is not a big deal, Because nobody hasn't used it yet.

The point is it will require updating U-Boot to use it, versus existing
U-Boots which already patch up the clock-frequency in the clockgen node.
And there's nothing semantically wrong with the way it currently is.

If I add a extra node with the clock-frequency property and don't move the
clock-frequency property of clockgen, that would be redundant because both clockgen node
and the extra node have the same clock-frequency node.
So, I choose what I did now.

I'm not complaining about how you structured the nodes, just how you
documented them.

-Scott



--
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