On 09/21/17 14:35, Philipp Zabel wrote: > Hi Ian, > > On Thu, 2017-09-21 at 12:06 +0100, Ian Arkver wrote: > [...] >>> My understanding of Hans' comment: >>> "I'd also add a comment that all other flags must be 0 if the device >>> tree is used. This to avoid mixing the two." >>> >>> is that all the above should only happen if (!!state->pdata). >> >> Except that state->pdata is a copy of the pdata, not a pointer, but you >> know what I mean. Some other check for DT needed here. > > Yes, I'll change this to zero all V4L2_MBUS_CSI2_[1-4]_LANE in the DT > case. I suppose the V4L2_MBUS_CSI2_CONTINUOUS_CLOCK and > V4L2_MBUS_CSI2_NONCONTINUOUS_CLOCK bits should be zeroed as well, then? Yes. Just zero all bits except those in the V4L2_MBUS_CSI2_LANE_MASK. And changing that to (0xf << 10) makes sense. Regards, Hans > >>> I don't know if this would break any existing DT-using bridge drivers. > > The only current users of g_mbus_config are the pxa_camera and > sh_mobile_ceu_camera soc_camera drivers. Neither supports MIPI CSI-2, as > far as I can tell. > > regards > Philipp >
