Select closest matching display resolution if requested by user is not available

[seriyps]

I think in case none of the resolutions supported by connected monitor matches exactly the value of --screen-mode, I think it might be better to anyway select one of the resolutions that is the closest one to the requested one and use it instead of crashing?

Eg, if user requests --screen-mode 1920x1080@60 but the supported resolutions are

9 : 1920x1080@30
12 : 1920x1080@24
13 : 1600x1200@60
14 : 1680x1050@60
15 : 1280x1024@75
16 : 1280x1024@60
17 : 1152x864@75

pixelpilot should select the closest one, eg, 1920x1080@24.

The way heuristics may work is by finding the minimum of the abs( (mode_width - hdisplay) + (mode_height - hdisplay) + (mode_vrefresh - vrefresh) ) here

PixelPilot_rk/src/drm.c

Lines 441 to 450 in 4579eda

	printf( "Available modes:\n");
	for (int i = 0; i < conn->count_modes; i++ ) {
	printf( "%d : %dx%d@%d\n",i, conn->modes[i].hdisplay, conn->modes[i].vdisplay , conn->modes[i].vrefresh );
	if (conn->modes[i].hdisplay == mode_width &&
	conn->modes[i].vdisplay == mode_height &&
	conn->modes[i].vrefresh == mode_vrefresh
	) {
	fc = i;
	}
	}

[seriyps]

Or it might be not the best idea actually - to give user not what they explicitly asked. Given that we already have "flexible" solution: if --screen-mode is not provided, pixelpilot selects the 1st resolution that it detects...

[zhouruixi]

I think the --screen-mode setting should be determined by the PixelPilot_rk startup script. Imagine that when I set a resolution, the program starts normally, but the resolution is not what I set. This is very strange. Generally, when passing unsupported parameters to a program, the program should give a prompt that it does not support the parameter, rather than trying to correct it.

[zhouruixi]

Should drop to preferred screen mode when specifying an invalid screen mode.