Hi all, This is my first intended-to-be-complete RFC patch to get rid of the enums in V4L2 IOCTL structs. The approach is the one outlined first by Mauro (AFAIR): <URL:http://www.spinics.net/lists/linux-media/msg46504.html> A set of compat structs (and compat IOCTLs) are created and there are a few functions to convert between the in-kernel representation and the old representation with enums the user space may well be using. On many archs the two IOCTLs are actually the same. This patchset depends on my earlier patch to remove v4l2_buffer.input: <URL:http://www.spinics.net/lists/linux-media/msg47144.html> All three patches are also available here: <URL:http://git.linuxtv.org/sailus/media_tree.git/shortlog/refs/heads/enum-fix> Open questions: - Orring the return values of {get,put}_user etc. is time-consuming on modern CPUs with deep pipelines. Would if be better to use | (logical or) instead? The regular case where the access is successful would be optimised on the expense of the error case. The end result in error cases may be different, too, but does that matter? - Testing this patch completely is difficult. I've only got access to capture hardware on 32-bit systems which generally do not easily exhibit problems with enums in IOCTLs (since they're 32-bit ints) in first place. I've tested this by changing fields from __u32 to __u64 where the old code had enums; that works, so at least something is working correctly. Help in testing would be very much appreciated. Questions and comments are welcome. Kind regards, -- Sakari Ailus e-mail: sakari.ailus@xxxxxx jabber/XMPP/Gmail: sailus@xxxxxxxxxxxxxx -- 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
