From mboxrd@z Thu Jan 1 00:00:00 1970 From: sebastian.hesselbarth@gmail.com (Sebastian Hesselbarth) Date: Tue, 15 Apr 2014 14:50:14 +0200 Subject: [PATCH 2/2] ARM: dts: berlin: add GPIO nodes for the BG2Q In-Reply-To: <20140415114848.GA2813@piout.net> References: <1397549245-3338-1-git-send-email-antoine.tenart@free-electrons.com> <1397549245-3338-3-git-send-email-antoine.tenart@free-electrons.com> <534CFA77.9030108@gmail.com> <20140415180058.2fefca78@xhacker> <534D0943.2050100@gmail.com> <20140415114848.GA2813@piout.net> Message-ID: <534D2B06.6000408@gmail.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 04/15/2014 01:48 PM, Alexandre Belloni wrote: > On 15/04/2014 at 12:26:11 +0200, Sebastian Hesselbarth wrote : >> On 04/15/2014 12:00 PM, Jisheng Zhang wrote: >>> On Tue, 15 Apr 2014 02:23:03 -0700 >>> Sebastian Hesselbarth wrote: >>>> I am fine with using nr-gpios property now, but I guess BG2Q also >>>> has that CONFIG[1,2] registers to actually read out the features >>>> synthesized in? If I find some time, I'll prepare a patch for >>>> dw-apb-gpio to exploit that (optional) information instead of >>>> using nr-gpios. >>> >>> The problem is CONFIG1/2 registers don't exist on some versions. >>> For example, the version used in BG2/BG2CD. So nr-gpio is necessary >>> if we want to support these versions. >> >> Hmm, are you sure about BG2? I remember reading it and it contains >> sane values. Anyway, a proper patch for dw-apb-gpio would include >> checking for sane (e.g. non-zero) values. And nr-gpios will always >> stay as fall-back just because e.g. sunxi does not have the CONFIG >> registers. > > Shouldn't that be use nr-gpios and if not available, read the CONFIG > registers? Else, what about bogus registers ? Probably, it is a better idea to only look for CONFIG registers if nr-gpios is not set. Anyway, I'll really have to find more time first ;) Sebastian