any idea why a ioctl(VIDIOC_S_FMT) never reaches the kernel (at least not in v4l2-core, or the device driver -- exynos-gsc) ? other stuff like QUERY and ENUM work, it's just S_FMT that somehow gets lots (although gstreamer reports a -22 EINVAL errno return)
strace doesn't snow VIDIOC_S_FMT either
memeka: the only thing I can think of is that the ioctl never reachs to the kernel
memeka: you are using gst right? are you using libv4l2 or not?
maybe libv4l2 is the one to blame here?
javier__: i should not be using libv4l2 but lemme check the compile args again
memeka: what does ldd libgstvideo4linux2.so say?
javier__: libv4l2.so.0 :(
javier__: ill try a new build ... i was going crazy with this missing ioctl
debugged gst and the kernel like crazy ... forgot about libv4l2 :(
memeka: I don't use libv4l2 and exynos-gsc works well for me with gst
so libv4l2 may be the one to blame. Also, is the only reasonable explanation for me if gst returns an error but the ioctl never reaches the kernel
javier__: this must've been it, i remember in older gst, enabling libv4l2 would break MFC ... strangely, MFC works now, but not GSC
building now :D
javier__: it works, can't thank you enough!
memeka: great, I'm glad that I could be of any help :)
javier__: do you know if gsc will be used by default by gst programs, or do i need to set an option for it to be default?
memeka: I believe it should, but I'm not really a gst expert. I know that for example decodebin picks s5p-mfc by default, but not sure for converters
memeka: it's a question for ndufresne (although I see he isn't in the channel now)
and I don't have access to my exynos board now to check
memeka: I'm going to sleep now since is quite late here, bye!
javier__: thanks again, gn