Get Correct Camera Depth

Code (CSharp):

1. struct v2f {
2. ...
3. float2 uv : TEXCOORD0;
4. }
5.  

In vert:

Code (CSharp):

1. o.uv=v.texcoords;

in frag:

Code (CSharp):

1. o.col.r = tex2D(_CameraDepthTexture, i.uv.xy);

So red cube(in the image below) outputs depth as color red on his texture.

But it is only 0 to 1, while it would seem like _CameraDepthTexture should have larger value, since I get depth<1(red<1) only when object is very close to the camera.

And second question:
I don't want to render depth texture as texture, I want shader to read from it before writing pixel to it. I want to read the depth value that will be replaced by this pixel. How do I do it?

http://s9.postimg.org/mhlw2i0jj/F***eng_shiet.png