On Wed, Jan 27, 2016 at 9:11 AM, Andrew Lunn <and...@lunn.ch> wrote: > On Wed, Jan 27, 2016 at 08:41:56AM -0800, Olof Johansson wrote: >> On Tue, Jan 26, 2016 at 4:11 PM, Andrew Lunn <and...@lunn.ch> wrote: >> > PHY devices may only list clause 22, 45, and their PHY identifier >> > values as compatible values. No other compatible strings are allowed. >> > Make this clear in the documentation, and remove examples where >> > make/model compatible strings are listed. >> > >> > Signed-off-by: Andrew Lunn <and...@lunn.ch> >> >> I'm not sure I agree with the disallowing here. It's common practice >> to use a specific compatible to describe the actual hardware used, in >> case it's needed in the future for some driver to distinguish >> behavior, etc. >> >> So while it should be required to include the clause compats, having a >> more specific one in there should be acceptable. > > Hi Olof > > Matching PHY devices to drivers has never used to compatible string, > other than the "ethernet-phy-XXXX.YYYY" string. The PHY has two > registers containing a manufacture id, device id and revision, > registers 2 and 3. These are the XXXX and YYYY. The core code reads > these values, or uses the values from the ethernet-phy-XXXX.YYYY, and > uses them to find a driver which supports these values. > > A make/model string is less specific than ethernet-phy-XXXX.YYYY. > > I will reword the changelog to make it clear that > "ethernet-phy-XXXX.YYYY" is allowed.
Only case I can see the need for a make/model string is if there's a need to add model-specific properties since you'd need a compatible then (or make those properties shared between all phy bindings). Anyway, we've never had a huge issue with this on other probable busses, so we should be fine with the above. With the clarification I'm OK with this change. -Olof
