On Thu, 2017-09-21 at 17:47 +0200, Georg Chini wrote: > On 21.09.2017 16:47, James Bottomley wrote: > > On Thu, 2017-09-21 at 17:28 +0300, Tanu Kaskinen wrote: > > > I think we should use the native backend for the HFP AG role by > > > default. If the native HFP AG implementation causes a conflict with > > > ofono in its default configuration (which seems likely to be the > > > case), then "headset=auto" should not enable the native HFP AG > > > implementation, which then means that we should use "headset=native" > > > by default. > > > > > > Using "headset=native" by default means that we lose HFP HF support > > > in the default configuration, but I don't think that's a big loss. > > > > Setting the default to native works for me: it's basically what all > > distros get today anyway because they don't install ofono by default. > > Â That probably means we don't need the elaborate switching for HSP_AG > > either, but it would become harmless, so could be removed later. > > > > > If we want to support the "HFP AG support through PA, HFP HF support > > > through ofono" case, then some new configuration option is needed, > > > and I think it would be clearest if that would be a separate patch > > > after this series. > > This is not true, the patch supplies an option to enable/disable > the HFP implementation. Simple usage of that switch would > provide all the required functionality. Yeah, sorry, I wasn't aware of that option, as I hadn't read the patch. You two just seemed to be choosing between enabling or disabling the native HFP AG implementation by default, and I just wanted to say that I want it to be enabled by default, but without breaking "headset=auto". > The default for the option > would just be disabled in "auto" mode and enabled in "native" > mode. The option seems to be named "enable_profile_hfp_hf", which suggests that disabling it will disable the ofono implementation of HFP AG as well, and that's not what we want. Maybe rename the option to "enable_native_hfp_hf"? > There is not a big code change needed, only a check > if the headset mode is "auto" or "native" and changing the > default accordingly. > The above is the essence of what I proposed to solve the > issue with headset=auto and I really don't understand why > this is causing such discussions. Leaving it as is definitely > breaks "headset=auto". Ok, sounds fine, with the added note that we should set "headset=native" by default. -- Tanu https://www.patreon.com/tanuk