Hi Guennadi, On Tuesday 30 December 2014 13:12:27 Guennadi Liakhovetski wrote: > On Tue, 30 Dec 2014, Josh Wu wrote: > > [snip] > > > > And until omap1 is in the mainline we cannot drop v4l2_clk. > > s/until/as lonh as/ > > > So I think the better way right now for ov2640 driver is still request > > both the v4l2_clock: mclk, and the clock: xvclk in probe(). > > In that way, we can take our time to introduce other patches to merged > > these two clocks. Does it make sense? > > How about this sequence, that we cat still try to get in on time for the > next merge window: > > 1. stop using the clock name in v4l2_clk > 2. add support for CCF to v4l2_clk, falling back to current behaviour if > no clock is found > 3. switch ov2640 to using "xvclk" It looks good at first sight. > Otherwise I would propose to add asynchronous probing support to ov2640 > _without_ changing the clock name. Whether or not we changee clock's name > isn't related directly to async support, since the driver already is using > v4l2_clk with the standarf "wrong" name. So, I would consider these two > changes separately - one is a new feature, another is a fix. The only > question is in which order to apply them. In general fix-first is > preferred, but since in this case the fix requires framework changes, we > could go with feature-first. This also makes sense since features need to > catch a merge window, whereas a fix can go in later too. So, if we don't > manage the above 3-step plan, I would priposw the most primitive patch ro > ov2640 just adding async support in line wuth other drivers and without > changing the clock name at first. Async support can go in without the clock rename, but DT support can't. -- Regards, Laurent Pinchart -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
