On 03/17/2014 01:54 PM, Hans Verkuil wrote: > These five patches fix sparse errors and warnings coming from the v4l2 > core. There are more, but those seem to be problems with sparse itself (see > my posts from Saturday on that topic). > > Please take a good look at patch 3/5 in particular: that fixes sparse > errors introduced by my vb2 changes, and required some rework to get it > accepted by sparse without errors or warnings. > > The rework required the introduction of more type-specific call_*op macros, > but on the other hand the fail_op macros could be dropped. Sort of one > step backwards, one step forwards. > > If someone can think of a smarter solution for this, then please let me > know. > > Regards, > > Hans > > Changes since v1: > Forgot to mention: - in patch 1/5 moved v4l2_subdev_notify from v4l2-subdev.h to v4l2-device.h and made it a static inline function as per Laurent's suggestion. > - added patch 2/5: the call_ptr_memop function checks for IS_ERR_OR_NULL > to see if a pointer is valid or not. The __qbuf_dmabuf code only used > IS_ERR. Made this consistent with both call_ptr_memop and the other > pointer checks elsewhere in the vb2 core code. > > - fixed a small typo in a comment that Pawel remarked upon. > > - Rewrote patch 5/5: Laurent wanted to keep the __user annotation with the > user_ptr. The reason I hadn't done that was that I couldn't make it work, > but I had an idea that moving the __user annotation before the '**' might > do the trick, and that helped indeed. > > -- > 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 > -- 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
