On Mon, Feb 13, 2023 at 11:49:26AM +0200, Tomi Valkeinen wrote: > Hi, > > On 10/02/2023 13:55, Tomi Valkeinen wrote: > > Add support to get and set subdev routes and to get and set > > configurations per stream. > > > > Based on work from Jacopo Mondi <jacopo@xxxxxxxxxx> and > > Niklas Söderlund <niklas.soderlund+renesas@xxxxxxxxxxxx>. > > > > Signed-off-by: Tomi Valkeinen <tomi.valkeinen@xxxxxxxxxxxxxxxx> > > --- > > utils/v4l2-ctl/v4l2-ctl-subdev.cpp | 288 +++++++++++++++++++++++++---- > > utils/v4l2-ctl/v4l2-ctl.cpp | 2 + > > utils/v4l2-ctl/v4l2-ctl.h | 2 + > > 3 files changed, 259 insertions(+), 33 deletions(-) > > > > diff --git a/utils/v4l2-ctl/v4l2-ctl-subdev.cpp b/utils/v4l2-ctl/v4l2-ctl-subdev.cpp > > index 33cc1342..81236451 100644 > > --- a/utils/v4l2-ctl/v4l2-ctl-subdev.cpp > > +++ b/utils/v4l2-ctl/v4l2-ctl-subdev.cpp > > @@ -1,5 +1,13 @@ > > #include "v4l2-ctl.h" > > > > +#define ARRAY_SIZE(array) (sizeof(array) / sizeof((array)[0])) > > + > > +/* > > + * The max value comes from a check in the kernel source code > > + * drivers/media/v4l2-core/v4l2-ioctl.c check_array_args() > > + */ > > +#define NUM_ROUTES_MAX 256 > > + > > struct mbus_name { > > const char *name; > > __u32 code; > > @@ -19,45 +27,57 @@ static const struct mbus_name mbus_names[] = { > > #define SelectionFlags (1L<<4) > > > > static __u32 list_mbus_codes_pad; > > +static __u32 list_mbus_codes_stream = 0; > > static __u32 get_fmt_pad; > > +static __u32 get_fmt_stream = 0; > > static __u32 get_sel_pad; > > +static __u32 get_sel_stream = 0; > > static __u32 get_fps_pad; > > +static __u32 get_fps_stream = 0; > > static int get_sel_target = -1; > > static unsigned int set_selection; > > static struct v4l2_subdev_selection vsel; > > static unsigned int set_fmt; > > static __u32 set_fmt_pad; > > +static __u32 set_fmt_stream = 0; > > static struct v4l2_mbus_framefmt ffmt; > > static struct v4l2_subdev_frame_size_enum frmsize; > > static struct v4l2_subdev_frame_interval_enum frmival; > > static __u32 set_fps_pad; > > +static __u32 set_fps_stream = 0; > > static double set_fps; > > +static struct v4l2_subdev_routing routing; > > +static struct v4l2_subdev_route routes[NUM_ROUTES_MAX]; > > > > void subdev_usage() > > { > > printf("\nSub-Device options:\n" > > - " --list-subdev-mbus-codes <pad>\n" > > + " --list-subdev-mbus-codes pad=<pad>,stream=<stream>\n" > > " display supported mediabus codes for this pad (0 is default)\n" > > " [VIDIOC_SUBDEV_ENUM_MBUS_CODE]\n" > > - " --list-subdev-framesizes pad=<pad>,code=<code>\n" > > + " --list-subdev-framesizes pad=<pad>,stream=<stream>,code=<code>\n" > > " list supported framesizes for this pad and code\n" > > " [VIDIOC_SUBDEV_ENUM_FRAME_SIZE]\n" > > " <code> is the value of the mediabus code\n" > > - " --list-subdev-frameintervals pad=<pad>,width=<w>,height=<h>,code=<code>\n" > > + " --list-subdev-frameintervals pad=<pad>,stream=<stream>,width=<w>,height=<h>,code=<code>\n" > > " list supported frame intervals for this pad and code and\n" > > " the given width and height [VIDIOC_SUBDEV_ENUM_FRAME_INTERVAL]\n" > > " <code> is the value of the mediabus code\n" > > - " --get-subdev-fmt [<pad>]\n" > > - " query the frame format for the given pad [VIDIOC_SUBDEV_G_FMT]\n" > > - " --get-subdev-selection pad=<pad>,target=<target>\n" > > + " --get-subdev-fmt pad=<pad>,stream=<stream>\n" > > + " query the frame format for the given pad and optional stream [VIDIOC_SUBDEV_G_FMT]\n" > > + " <pad> the pad to get the format from\n" > > + " <stream> the stream to get the format from (0 if not specified)\n" > > + " --get-subdev-selection pad=<pad>,stream=<stream>,target=<target>\n" > > " query the frame selection rectangle [VIDIOC_SUBDEV_G_SELECTION]\n" > > " See --set-subdev-selection command for the valid <target> values.\n" > > - " --get-subdev-fps [<pad>]\n" > > + " --get-subdev-fps pad=<pad>,stream=<stream>\n" > > " query the frame rate [VIDIOC_SUBDEV_G_FRAME_INTERVAL]\n" > > Laurent noticed that the above option, and a few others, break backward > compatibility. > > If no one has other suggestions, I'll additionally add back the old > option format. In other words, if the parameter is a number, it's a pad. > Otherwise we look for pad=<pad>,stream=<stream> style option. > > Well, another alternative would be to use an "extended" pad format > instead. I.e. for each <pad> in the old style, we'd accept something > like <pad[/stream]>. E.g. "--list-subdev-framesizes > pad=<pad[/stream]>,code=<code>" > > Opinions? I'd go for pad=<pad>,stream=<stream>. For options that just take a numerical value, <pad>/<stream> is fine, but pad=<pad>/<stream> is confusing. -- Regards, Laurent Pinchart