Ohad Ben-Cohen <ohad@xxxxxxxxxx> writes: > On Tue, Sep 20, 2011 at 11:37 AM, Kalle Valo <kvalo@xxxxxxxxxx> wrote: >> One extreme is that wl12xx would set IEEE80211_HW_SUPPORTS_DYNAMIC_PS >> and reimplement the dynps timer in the driver. That way the driver >> would have full control how it works and not complicate the mac80211 >> implementation. > > That sounds like an awful code duplication just to prevent a small API > extension. Heh, that's always relative and actually for me it's the opposite. What you consider small API update is for me a major change as it allows the driver control dynps timer parameters, which we haven't allowed earlier. And for me adding that dynps timer to a driver isn't that much more code, basically it's just a simple timer. And I doubt that you even need to stop the queues when changing the ps state like mac80211 does. I would not describe it as "an awful lot of code". Having your own time would allow you to change to the BT coex implementation of the week your firmware provides and not interferece with mac80211 and other drivers. > Yes, adding API complicates the code, but isn't the whole point of > mac80211 is to prevent egregious code duplication ? Sure, but there's always a limit for everything. mac80211 is growing all the time and becoming more complex. I think the "All in" approach won't work for too long and we need to start considering how to keep mac80211 still maintainable. Like I have said earlier, the power save implementation is a good example of this. -- Kalle Valo -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
