Hi Geert, On 10 April 2018 15:29 Geert Uytterhoeven wrote: > On Tue, Apr 10, 2018 at 4:23 PM, Phil Edworthy wrote: > > On 10 April 2018 07:24 Phil Edworthy wrote: > >> On 09 April 2018 20:20 Rob Herring wrote: > >> > On Wed, Mar 28, 2018 at 03:22:30PM +0100, Phil Edworthy wrote: > > [...] > >> > > +- interrupt-mask : a 32-bit bit mask that specifies which > >> > > +interrupts in the list > >> > > + of interrupts is valid, bit is 1 for a valid irq. > >> > > >> > This is not a standard property and would need a vendor prefix. > >> > However, > >> I'd > >> > prefer you just skip any not connected interrupts with an invalid > >> > interrupt number. Then the GPIO number is the index into "interrupts". > >> Makes sense, I'll rework it to do this. > > Err, what would an invalid interrupt number be? > > If I use -1, I get a DT parsing error and 0 is certainly valid. If the > > number is larger than the valid interrupt range I get errors during probe. > > Perhaps using interrupts-extended instead of interrupts? > > E.g. > > interrupts-extended = <&intc1 5 1>, <0>, <&intc2 1 0>; Thanks for the pointer, I'll have a look. Phil
