Hi Hans, Thank you for the review. On Wednesday 28 January 2015 11:27:02 Hans Verkuil wrote: > On 01/28/15 10:17, Laurent Pinchart wrote: > > Expose the module as two controls, one for the 3x3 multiplier matrix and > > one for the 3x1 offset vector. > > > > Signed-off-by: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx> > > --- > > > > drivers/staging/media/omap4iss/iss_ipipe.c | 129 +++++++++++++++++++++++- > > drivers/staging/media/omap4iss/iss_ipipe.h | 17 ++++ > > 2 files changed, 144 insertions(+), 2 deletions(-) > > > > diff --git a/drivers/staging/media/omap4iss/iss_ipipe.c > > b/drivers/staging/media/omap4iss/iss_ipipe.c index 73b165e..624c5d2 > > 100644 > > --- a/drivers/staging/media/omap4iss/iss_ipipe.c > > +++ b/drivers/staging/media/omap4iss/iss_ipipe.c > > @@ -119,6 +119,105 @@ static void ipipe_configure(struct iss_ipipe_device > > *ipipe)> > > } > > > > /* ---------------------------------------------------------------------- > > + * V4L2 controls > > + */ > > + > > +#define OMAP4ISS_IPIPE_CID_BASE (V4L2_CID_USER_BASE | 0xf000) > > Private control ranges should be reserved in uapi/linux/v4l2-controls.h. > > See e.g. V4L2_CID_USER_SAA7134_BASE. My bad, I'll fix that. > > +#define OMAP4ISS_IPIPE_CID_RGB2RGB_MULT (OMAP4ISS_IPIPE_CID_BASE + 0) > > +#define OMAP4ISS_IPIPE_CID_RGB2RGB_OFFSET (OMAP4ISS_IPIPE_CID_BASE + 1) > > Can you give some information how the values are interpreted? That should > be documented anyway, but I would like to see how this compares to the > adv drivers. This is something that we might want to make available as > standard controls. I will have to think about that a bit more. Sure. http://www.ti.com/lit/pdf/swpu235, section 8.3.3.4.6, page 1863. / \ / \ / \ / \ | R_out | | gain_RR gain_GR gain_BR | | R_in | | offset_R | | G_out | = | gain_RG gain_GG gain_BG | x | G_in | + | offset_G | | B_out | | gain_RB gain_GB gain_BB | | B_in | | offset_B | \ / \ / \ / \ / The two controls correspond to the multiplication matrix and offset vector. Coefficients are stored in 16 bits each and expressed as S3.8 (-4 to +3.996) for the gains and S11 (-1024 to 1023) for the offsets. Note that the ISS IPIPE has two RGB to RGB blending matrices as shown on figure 8-132, page 1859. This patch implements support for the first one only. We should probably consider how to expose the second one as well. -- 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
