On 08/04/16(Fri) 17:14, Patrick Wildt wrote:
> On Fri, Apr 08, 2016 at 04:34:25PM +0200, Martin Pieuchot wrote:
> > On 08/04/16(Fri) 14:46, Patrick Wildt wrote:
> > > Hi,
> > >
> > > so that we can easily check if a node is compatible or to find nodes
> > > that are compatible, I would like to add helpers to the fdt routines.
> > >
> > > This way the drivers can check if they "match" to a node by simply
> > > calling:
> > >
> > > if (fdt_node_compatible(ma->ma_node, "samsung,exynos4210-ehci"))
> > > return (1);
> > >
> > > Sometimes it's helpful to find any node that is compatible. It can
> > > be helpful for instance on finding the early uart. Simplified example:
> > >
> > > if ((node = fdt_find_compatible("arm,pl011")) != NULL)
> > > pl011cnattach(...);
> > >
> > > Thoughts? ok?
> >
> > Always hard to comment without seeing which code would use that.
> >
> > Is a custom function really needed? Why not do like sparc64 and macppc,
> > for example your ehci_match() could be:
> >
> > {
> > char compat[32];
> > ...
> >
> > if (OF_getprop(ma->ma_node, "compatible", compat, sizeof(compat)) == -1)
> > return 0;
> >
> > if (strcmp(compat, "samsung,exynos4210-ehci") == 0
> > return 1;
> >
> > return 0;
> > }
> >
>
> The compatible attribute can be a list of strings:
Yes, that's also true for macppc.
> Implementing parsing multiple strings in a buffer and checking it in
> every _match function seemed a bit too much overhead.
Seems or is overhead? How is the reality? From my experience dealing
with Apple device-tree most of the time you just use one of the
compatible entry.
> Imagine 20 drivers
> doing the same code over and over again.
So? What I find really nice about FDT is that you can use the OF_* API.
That means we can write a driver for ARMv7 the same way we do it for
sparc64 and macppc. So it's easier to learn an maintain. If you
introduce fdt-only functions, we lose this advantage.
Now there might be a real benefit to having such function, but in that
case I'd suggest to implement it in the OF_* API.
For the moment I haven't seen any code that in my opinion justify this.
> Especially if you're also are not looking for one compatible, but
> multiple compatibles:
You have some example, I don't find the code hard to follow:
macppc: aoa, wdc@obio, mpcpci
sparc64: com@ebus, comkbd, comms, ofwiic, pcfiic, pyro, schizo
