On Wednesday 09 December 2015 09:57:40 Lucas Stach wrote: > Am Mittwoch, den 09.12.2015, 16:50 +0800 schrieb Peter Chen: > > On Tue, Dec 08, 2015 at 08:36:00AM -0700, Mathieu Poirier wrote: > > > On 7 December 2015 at 18:37, Peter Chen <peter.chen@xxxxxxxxxxxxx> wrote: > > > > + > > > > + if (dev->of_node) { > > > > + struct device_node *node = dev->of_node; > > > > + > > > > + hub_data->clk = devm_clk_get(dev, "external_clk"); > > > > + if (IS_ERR(hub_data->clk)) { > > > > + dev_dbg(dev, "Can't get external clock: %ld\n", > > > > + PTR_ERR(hub_data->clk)); > > > > + } > > > > > > Is the intended behaviour to keep going here event when there is an > > > error? Can the "hub_data" really work without a clock? > > > > Yes, some HUB may work with fixed 24M OSC at the board, but they need to > > reset through external IO, so the clock is not need at this case, but > > reset pin is mandatory. > > > If the hub always requires a clock it must not be optional. If you have > a fixed 24MHz clock on board add this to the DT as a fixed-clock and use > it as an input to the hub. I think it's fine to make the clock optional in the sense that you only need to list non-fixed clocks that have to be enabled and or controlled. Which reminds me, should the driver call clk_set_rate()? It currently doesn't, but other machines might need that. Arnd -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html