If $f$ is a strictly increasing function with $f(f(x))=x^2+2$, then $f(3)=?$ Bdmo 2014 regionals(a tweaked version of question):

If $f$ is a strictly increasing function over the reals with $f(f(x))=x^2+2$, then $f(3)=?$

Obviously,$f(3)=f(1)^2+2$ but I can't see where we are going to use the 'strictly increasing' fact.I don't think there is a way to reverse-engineer such a function without heavy machinery.I have plugged in loads of values but they have yielded nothing.Some help will be appreciated.
EDIT: As others have noted, such a function is not possible with our current domain, but:

If the function is defined over the positive reals, does $f(3)$ have a definite value?

 A: 
If $f:\mathbb{R}\rightarrow \mathbb{R}$ be a strictly increasing
  function. Then $f(f(x))$ cannot be  even.

Proof:
$$x>0\Rightarrow f(-x)<f(x)\Rightarrow f(f(-x))<f(f(x))$$
Contradiction.
A: Any function $f_{base}$ satisfying:
(1) Is defined on $[0,r)$ for some $0<r<2$.
(2) Is continuous and strictly increasing.
(3) $f_{base}(0)=r$ and $\lim\limits_{x\rightarrow r^{-}}f_{base}(x)=2$.
can be extended to a function $f$ on nonnegative real satisfying the above functional equation and is strictly increasing.
This can be proved by induction. We extend to $[r,2)$, $[2,r^{2}+2)$, $[r^{2}+2,6)$ and so on repeatedly using the equation $f(x)=(f^{-1}(x))^{2}+2$. We won't prove it here, as it's not the main point. The main point is that we simply need to specify the function in a small range and trust that it can be extended to the whole nonnegative real. Using this we can show that there are no definite value for $f(3)$.
Try:
$f(x)=1+x^{2}$ for $x\in[0,1)$; $x+1$ for $x\in[1,2]$. Then $f(3)=6=2^{2}+2$ since $f(2)=3$.
Now try:
$f(x)=\frac{3}{2}+\frac{2}{9}x^{2}$ for $x\in[0,\frac{3}{2})$; $\frac{9}{2}(x-\frac{3}{2})+2$ for $x\in[\frac{3}{2},2]$. Then $f(3)=(\frac{31}{18})^{2}+2\not=6$ since $f(\frac{31}{18})=3$.
