Hi Kaneko-san, On Tue, 21 Oct 2014, Yoshihiro Kaneko wrote: > From: Koji Matsuoka <koji.matsuoka.xm@xxxxxxxxxxx> > > Up until now scaling has been forbidden for the NV16 capture format. > This patch adds support for horizontal scaling-up for NV16. Vertical > scaling-up for NV16 is forbidden by the H/W specification. Here and also in the subject - what do you mean by "scaling-up?" Do you really mean increasing sizes, i.e. scaling from smaller sizes to larger ones? Is down-scaling not supported? Maybe someone with a better English knowledge, then myself, could advise - is "add scaling-up support" ok or would "add up-scaling support" or, if we don't really mean increasing, just "add scaling support" be better? > > Signed-off-by: Koji Matsuoka <koji.matsuoka.xm@xxxxxxxxxxx> > Signed-off-by: Yoshihiro Kaneko <ykaneko0929@xxxxxxxxx> > --- > v3 [Yoshihiro Kaneko] > * no changes > > v2 [Yoshihiro Kaneko] > * Updated change log text from Simon Horman > * Code-style fixes as suggested by Sergei Shtylyov > > drivers/media/platform/soc_camera/rcar_vin.c | 17 +++++++++++++---- > 1 file changed, 13 insertions(+), 4 deletions(-) > > diff --git a/drivers/media/platform/soc_camera/rcar_vin.c b/drivers/media/platform/soc_camera/rcar_vin.c > index ecdbd48..fd2207a 100644 > --- a/drivers/media/platform/soc_camera/rcar_vin.c > +++ b/drivers/media/platform/soc_camera/rcar_vin.c > @@ -644,7 +644,7 @@ static int rcar_vin_setup(struct rcar_vin_priv *priv) > /* output format */ > switch (icd->current_fmt->host_fmt->fourcc) { > case V4L2_PIX_FMT_NV16: > - iowrite32(ALIGN(ALIGN(cam->width, 0x20) * cam->height, 0x80), > + iowrite32(ALIGN((cam->out_width * cam->out_height), 0x80), > priv->base + VNUVAOF_REG); > dmr = VNDMR_DTMD_YCSEP; > output_is_yuv = true; > @@ -1614,9 +1614,17 @@ static int rcar_vin_set_fmt(struct soc_camera_device *icd, > * At the time of NV16 capture format, the user has to specify the > * width of the multiple of 32 for H/W specification. > */ > - if ((pixfmt == V4L2_PIX_FMT_NV16) && (pix->width & 0x1F)) { > - dev_err(icd->parent, "Specified width error in NV16 format.\n"); > - return -EINVAL; > + if (pixfmt == V4L2_PIX_FMT_NV16) { > + if (pix->width & 0x1F) { > + dev_err(icd->parent, > + "Specified width error in NV16 format. Please specify the multiple of 32.\n"); > + return -EINVAL; > + } > + if (pix->height != cam->height) { > + dev_err(icd->parent, > + "Vertical scaling-up error in NV16 format. Please specify input height size.\n"); > + return -EINVAL; > + } Similar to the previous patch - shouldn't this new test be added to _try_fmt() and there you would just fix the size instead of erroring out? Thanks Guennadi > } > > switch (pix->field) { > @@ -1661,6 +1669,7 @@ static int rcar_vin_set_fmt(struct soc_camera_device *icd, > case V4L2_PIX_FMT_YUYV: > case V4L2_PIX_FMT_RGB565: > case V4L2_PIX_FMT_RGB555X: > + case V4L2_PIX_FMT_NV16: /* horizontal scaling-up only is supported */ > can_scale = true; > break; > default: > -- > 1.9.1 > -- 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