On Fri, Sep 1, 2017 at 2:10 PM, Arend van Spriel <arend.vanspr...@broadcom.com> wrote: > On 01-09-17 18:49, Rob Herring wrote: >> >> On Wed, Aug 30, 2017 at 02:02:18PM +0200, Antony Antony wrote: >>> >>> hi, >>> >>> On Wed, Aug 30, 2017 at 10:28:20AM +0800, Chen-Yu Tsai wrote: >>>> >>>> On Wed, Aug 30, 2017 at 5:43 AM, Antony Antony <ant...@phenome.org> >>>> wrote: >>> >>> >>>>> a/Documentation/devicetree/bindings/net/wireless/brcm,bcm43xx-fmac.txt >>>>> +++ >>>>> b/Documentation/devicetree/bindings/net/wireless/brcm,bcm43xx-fmac.txt >>>>> @@ -6,7 +6,9 @@ connects the device to the system. >>>>> >>>>> Required properties: >>>>> >>>>> - - compatible : Should be "brcm,bcm4329-fmac". >>>>> + - compatible : should be one of the following: >>>>> + * "brcm,bcm4329-fmac" >>>>> + * "brcm,bcm43430-fmac" >>>> >>>> >>>> You updated the bindings, but not the driver. So it's not actually >>>> going to work. More specifically, OOB interrupts won't work. >>>> >>> >>> understood, ignore this patch for now. Thanks Chen-Yu. >>> >>>> IIRC, The compatible string for this particular case, as it was >>>> originally proposed, only serves as a placeholder for the driver >>>> to check against. None of the instances in sunxi device trees >>>> match the actual chip model. Actual model matching is done >>>> through SDIO, as you've already seen. >>> >>> >>> yes it seems SDIO driveer code is smarter, once it initialize >>> brcm,bcm4329-fmac it ignore the DT info and read the chip details to >>> locate >>> firmware file. >>> >>> I also noticed other boards using bcm4329-fmac in similar situations. >>> https://patchwork.kernel.org/patch/9739181/ >>> >>> >>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/arm64/boot/dts/amlogic/meson-gxbb-nanopi-k2.dts?h=v4.13-rc7 >>> >>> I will resend "NanoPi NEO Plus2" dts with "brcm,bcm4329-fmac" and see >>> where >>> it goes. >> >> >> Adding the compatible or instead of? The former would be better. You >> should still have the actual chip in case you do have some difference to >> handle. > > > Hi Rob, > > Actually the Broadcom wifi chips themselves are discoverable. So once the > driver has access to the register space of the device it can determine the > actual chip, its revision, and exactly what cores (and their revision) are > present in the chip. Hence there is a single compatible string as there is > no need to convey the same information through device tree data.
So if a chip has different power on/off sequencing you can discover that? I realize that most often you don't need it, but a more specific compatible is there in case you do and so it doesn't require a DTB update to handle some difference. But you can keep using one compatible because I can't really enforce any of that. Rob