On Wed, Dec 28, 2016 at 11:08 AM, Rafał Miłecki <zaj...@gmail.com> wrote: > On 3 October 2016 at 15:29, Kalle Valo <kv...@codeaurora.org> wrote: >> Arnd Bergmann <a...@arndb.de> writes: >> >>> On Friday 30 September 2016, Felix Fietkau wrote: >>>> >> >> >> + device_type = "pci"; >>>> >> >> >> + mediatek,mtd-eeprom = <&factory 0x8000>; >>>> >> >> >> + mediatek,2ghz = <0>; >>>> >> > >>>> >> > It's not clear what the possible values for the 2ghz property are, >>>> >> > can you be more verbose in the description? How is <0> different >>>> >> > from no property? >>>> >> 0 means disabled, no property means unchanged (compared to EEPROM). >>>> > >>>> > Maybe have a boolean property instead then to say >>>> > "mediatek,2ghz-disabled" ? >>>> > >>>> > If zero is the only possible value, there is no need to put a number in >>>> > there. >>>> 1 is also possible, which will force-enable the capability. >>> >>> Ok, then both those values should be documented in the binding. >> >> Related to this, Martin sent patches which add generic bindings for >> enabling 2 Ghz and 5 Ghz bands. >> >> [RFC,1/3] Documentation: dt-bindings: add IEEE 802.11 binding documentation >> https://patchwork.kernel.org/patch/9359833/ >> >> [RFC,2/3] of: add IEEE 802.11 device configuration support code >> https://patchwork.kernel.org/patch/9359837/ > > I would prefer something more generic. Many tri-band routers split 5 > GHz band into 2 sets of channels and they have separated radios for > them. > > E.g. Netgear R8000 has phy0 which should be used for higher part of 5 > GHz band (channels 149+) and phy2 which should be used for lower part > of 5 GHz band (channels from 36 to 48 or 64). > > What do you think about some more flexible properties like: > ieee80211-min-center-freq > ieee80211-max-center-freq what would happen if only one of these properties was given or would we forbid that (because the .dts should always describe the hardware, and if we describe a lower bound then we should also describe the upper bound)? the benefits of your solution are: - this would allow *enabling* bands as well (my proposal allows this as well, but the .dts is indeed a bit hard to read - unlike your solution which looks nice to me) - we could handle this within generic cfg80211/mac80211 code instead of "duplicating" it per driver
should we describe the center freq in Hz or MHz (cfg80211's ieee80211_channel uses the latter)? @Arnd: what do you think from devicetree perspective? Regards, Martin [0] http://lxr.free-electrons.com/source/include/net/cfg80211.h#L130
