On Tue, Jun 28, 2016 at 05:18:22PM +0100, Mark Brown wrote: > On Tue, Jun 28, 2016 at 06:23:38PM +0300, Mika Westerberg wrote: > > On Tue, Jun 28, 2016 at 04:07:52PM +0100, Mark Brown wrote: > > > > If that's the case then it's definitely broken - we either always warn > > > or never warn, neither is good. > > > For DT we warn if the device is used directly (by the node name) as > > previously. For ACPI we never warn because it is not possible to use the > > node name directly. We always require use of a proper compatible string. > > > I don't see how that is broken. > > So you're saying that it's not possible for ACPI to use anything > *except* an explicitly listed compatible string to bind? What we want > to avoid is any ACPI tables that explicitly list spidev since that's a > total abstraction failure. That's exactly what I'm saying. We never match using the node name (which is always 32-bit "string" like "SPI0"). For PRP0001 match to succeeed you need to have "compatible" property. The ASL code looks like: Device (SPID) { Name (_HID, "PRP0001") Name (_CRS, ResourceTempate () { SpiSerialBus (1, PolarityLow, FourWireMode, 0x08, ControllerInitiated, 0x007A1200, ClockPolarityLow, ClockPhaseSecond, "\\_SB.SPI1", 0x00, ResourceConsumer) }) Name (_DSD, Package () { ToUUID("daffd814-6eba-4d8c-8a91-bc9bbf4aa301"), Package () { Package () {"compatible", "rohm,dh2228fv"}, } }) } Without the "compatible" property you get a warning from ACPI core: \_SB.SPI1.SPID requires 'compatible' property and it will not match anything. -- To unsubscribe from this list: send the line "unsubscribe linux-spi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html