Hi Andy, On Mon, 19 May 2025 15:06:33 +0300 Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx> wrote: > On Mon, May 19, 2025 at 01:58:18PM +0200, Herve Codina wrote: > > On Thu, 8 May 2025 17:27:52 +0300 > > Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx> wrote: > > > On Wed, May 07, 2025 at 09:12:47AM +0200, Herve Codina wrote: > > ... > > > > > if (of_property_match_string(np, "compatible", match->compatible) == 0) > > > > > > Side note, there is an fwnode_is_device_compatible() API for such cases. And IIRC > > > there is also OF variant of it. > > > > fwnode_device_is_compatible() checked for all compatible string. I mean, if > > we have compatible = "foo,custom-bus", "simple-bus"; > > fwnode_device_is_compatible() checking against "simple-bus" returns true. > > > > Here, we want "simple-bus" as the first position in the compatible string. > > In other word, we want to match the more specific compatible string as > > mentioned in the comment. > > I admit I'm not an expert in DT, but why is the compatibility position > dependent? > > ... > > > > > + if (pdev->dev.of_node) > > > > > > Why do you need this check? AFAICS it dups the one the call has already in it. > > > > of_platform_populate() was called only if an OF node is present. > > I want to call of_platform_depopulate() on removal also only if an OF node > > is present. > > > > I don't see the other call that duplicated this check. > > > > Can you clarify? > > The of_...() is already NULL-aware (AFAICS), why do you need the duplicated > check? Oh, I see. I was focused on previous of_device_get_match_data() call. My bad. Indeed, you're right, I can call directly of_platform_depopulate() without checking pdev->dev.of_node before the call. The check is already present in of_platform_depopulate() itself. I will do that in the next iteration. Thanks for pointing out. Best regards, Hervé
