Hi Tim, On Tuesday 17 March 2015 17:57:30 Tim Nordell wrote: > On 04/21/14 07:29, Laurent Pinchart wrote: > > Replace the custom buffers queue implementation with a videobuf2 queue. > > > > Signed-off-by: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx> > > I realize this is late (it's in the kernel now), but I'm noticing that > this does not appear to properly support the scatter-gather buffers that > were previously supported as far as I recall (and can tell with what was > removed with this patch), especially when using USERPTR. You can > observe this using "yavta" with the -u parameter. Can you confirm if > this works for you? I get the following output from the kernel when > attempting to stream a 640x480 UYVY framebuffer: > > [ 111.381256] contiguous mapping is too small 589824/614400 The OMAP3 ISP uses an IOMMU, physically non-contiguous buffers should thus be mapped contiguously into the device memory space. I haven't tried USERPTR support recently, but this surprises me. It requires investigation. Could you give it a try ? -- Regards, Laurent Pinchart -- 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
