From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nicolas Ferre Subject: Re: [PATCH v2 net-next 2/3] net: ethernet: cadence-macb: Add fallback to read DT provided caps Date: Tue, 8 Dec 2015 17:22:52 +0100 Message-ID: <566703DC.1030501@atmel.com> References: <1449485914-12883-1-git-send-email-narmstrong@baylibre.com> <1449582726-6148-1-git-send-email-narmstrong@baylibre.com> <1449582726-6148-3-git-send-email-narmstrong@baylibre.com> <2217940.QPNsks7uyL@wuerfel> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: , , , , , , , To: Arnd Bergmann , Neil Armstrong Return-path: In-Reply-To: <2217940.QPNsks7uyL@wuerfel> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Le 08/12/2015 16:00, Arnd Bergmann a =E9crit : > On Tuesday 08 December 2015 14:52:05 Neil Armstrong wrote: >> Add 1:1 mapping of software defines caps parsing from DT in case the >> generic macb compatible form is used. >> These properties will provide support for futures implementations >> only defined from DT without need to update the driver code to suppo= rt >> new variants. >> >> Signed-off-by: Neil Armstrong >> >=20 > Translating the Linux implementation specific configuration into > DT properties directly is usually not the best way. >=20 > Could we instead have a lookup table by compatible string to set the > flags? It seems that there are lots of different flags but only a > couple of different users of this IP block. Also, the fact that > you are now adding yet another quirk tells me that the set you > define today is unlikely to cover all the future requirements. This is basically what I told Neil in my previous email. I understand you point Neil, but I don't find it makes sense and Arnd described it better that I did. So please find a proper compatibility string and simply use it. What about: "cdns,the_name_of_the_product_that_first_implemented_this_no_usrio_spec= ial_case-gem"? Bye, --=20 Nicolas Ferre