On Friday 28 February 2014 11:19:36 Lucas Stach wrote: > Am Donnerstag, den 27.02.2014, 17:44 +0100 schrieb Arnd Bergmann: > > On Thursday 27 February 2014 17:41:44 Lucas Stach wrote: > > > num-lanes = <1>; > > > - interrupts = <0 123 0x04>; > > > + interrupt-names = "inta", "intb", "intc", "intd/msi"; > > > + interrupts = <0 123 0x04>, <0 122 0x04>, <0 121 0x04>, <0 120 0x04>; > > > clocks = <&clks 189>, <&clks 187>, <&clks 206>, <&clks 144>; > > > > > > > The standard PCI interrupts should not be listed here, you need to > > put them into the "interrupt-map" property so the of_irq_parse_and_map_pci() > > function can translate them. > > > > So as INTA is already listed and implemented in the driver this way, > this means the binding is totally bogus (taking into account that it > didn't match the documented designware binding in more places). > > I wonder if we should just break the binding to sort things out, given > that there are not that many users of imx-pcie yet. That may be best, yes. If we have to provide backwards compatibility, the driver can have a fallback for the case where no interrupt-map property is present, but it should not try to handle multiple interrupt lines that way, only the trivial case where you have a single IntA line for all devices. 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