Em Mon, 26 Sep 2016 14:39:06 +0530 Ajith Raj <ajith.raj@xxxxxxxxxxxx> escreveu: > Hi Maintainers, > > I was working on a CPI based camera driver based on soc_camera framework > which is compiled as a static module in kernel, due to the non-availability > of modprobe framework. Don't rely on soc_camera. We're in the process of rework such drivers to use directly the V4L2 core or to remove them. > While I was pushing the changes, I got a question from the internel code > review that why the driver is being compiled as static while all other > drivers available in the open source are compiled as dynamic .ko modules. > > I understand that camera as a component need not have to be built static and > can be added later on the need basis. Is there any reason apart from this? > Since the modprobe framework is not available, if I build it as dynamic .ko > module, I need to insert all dependent modules independently and in the > right order. There are many; > > videobuf2-dma-contig.ko, videobuf2-v4l2.ko, videobuf2-memops.ko, > videobuf2-vmalloc.ko, …..etc > v4l2-common.ko, v4l2-mem2mem.ko, v4l2-dv-timings.ko > soc_mediabus.ko,soc_camera.ko ,……etc > media.ko, videodev.ko You don't need to take care of the order, as modprobe inserts them with the right dependencies automatically. You may need to use the async probe inside V4L, though, in order to ensure that the main driver will be initializing the sub-drivers in the right order, as OF doesn't ensure that. > Do you see any issues in submitting camera as a static module? > Or it has to be dynamic? If so, could you please help me understand the > reasons. All media drivers should build with modprobe. The more important reason (at least for my PoV) is that it should be built with allmodconfig/allyesconfig, and on all architectures, if COMPILE_TEST Kconfig define is defined. The rationale is that building the driver this way is required for us to avoid building problems. Also, some static code analyzers we have (Coverity) only runs on x86. Another thing is that drivers that don't accept dynamic module load very likely have hidden bugs (usually, race bugs, PM support issues, dynamic bind/unbind issues). > > Thanks in advance. > > Regards, > Ajith > -- > 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 Thanks, Mauro -- 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
