On 24/11/15 00:12, Simon Arlott wrote: > On 23/11/15 18:22, Florian Fainelli wrote: >> On 22/11/15 14:17, Simon Arlott wrote: >>> The BCM63268 has a NAND interrupt register with combined status and enable >>> registers. It also has a clock for the NAND controller that needs to be >>> enabled. >>> >>> Set up the device by enabling the clock, disabling and acking all >>> interrupts, then handle the CTRL_READY interrupt. >>> >>> Add a "device_remove" function to struct brcmnand_soc so that the clock >>> can be disabled when the device is removed. >>> >>> Signed-off-by: Simon Arlott <simon@xxxxxxxxxxx> >>> --- >>> On 22/11/15 21:59, Rob Herring wrote: >>>>>>> + * "brcm,nand-bcm63268" >>>>>>> + - compatible: should contain "brcm,nand-bcm<soc>", "brcm,nand-bcm63268" >>>>> >>>>> vendor,<soc>-device is preferred. >>> >>> The existing two bindings use brcm,nand-<soc>, but I've changed this one. >> >> Could we stick with the existing binding naming convention of using: >> >> brcm,nand-<soc> just so automated tools or other things can match this >> one too, and +1 for consistency? > > I could submit another patch renaming the existing bindings to > brcm,<soc>-nand, and add that to the drivers? Then they'd be consistent. No, let's not create unnecessary churn because of a minor mistake. So, yes we *should* have used brcm,<soc>-nand in the first place, but now that there are DTSes out there using "brcm,nand-<soc>" there is not really any point in doing this, so please update your patches so they match the existing convention. > >> Other than, that, same comment as Jonas, why do we we need the >> device_remove callback to be called from the main driver down to this one? > > I'll add a "struct brcmnand_soc *brcmnand_get_socdata(struct device *)" > instead so that I can access the soc data before calling brcmnand_remove. > -- Florian -- 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