On Tue, Jun 14, 2022 at 11:25:07AM +0200, Andy Shevchenko wrote: > On Mon, Jun 13, 2022 at 9:51 PM Hans de Goede <hdegoede@xxxxxxxxxx> wrote: > > > > Hi All, > > > > I want to start writing a libcamera pipeline handler for the atomisp2, > > but before I can do that I first need to fix mmap support in atomisp2. > > > > My plan for this is to port the atomisp2 handler to videobuf2. To do that > > I first need to understand the existing memory/buffer management so I've > > started with cleaning up the hmm code (with a bit of a detour here and > > there). > > > > This series is the result of that. There are likely more cleanups to > > follow, but I've to focus on some other things for a bit. I hope to be > > able to return to the cleanups + an eventual videobuf2 conversion soon. > > Reviewed-by: Andy Shevchenko <andy.shevchenko@xxxxxxxxx> > > For patches 1-8: The code wise it's good clean up, functionality it > might be that intention was to implement it as some point, but looking > into (internal) history of the driver development I believe it would > require some firmware changes which is impossible for upstreamed > version of the driver and as you noticed never productized that time. > Hence, good that we got less LoCs after all. > > For patches 9-13: I believe that patch 10 and 9 should be swapped in > the series. Logically you drop caller first followed by (unused) > callee. > > For the rest: To be continued... For the rest: Reviewed-by: Andy Shevchenko <andy.shevchenko@xxxxxxxxx> I believe a couple of comments are easy to address, may keep my tag there. -- With Best Regards, Andy Shevchenko