All Binary Combinations to Decimal

Python, 60 bytes

lambda x,y:[n for n in range(1<<x+y)if bin(n).count('1')==y]

Test it on Ideone.

How it works

All positive numbers that can be represented in binary with x zeroes and y ones is clearly smaller than 2^{x + y}, since the canonical binary representation of the latter has x + y + 1 digits.

The lambda simply iterates over the integers in [0, 2^{x + y}) and keeps all integers n in that range that have y ones. Since n < 2^{x + y} is can be represented with x (or less) zeroes.

Jelly, 8 bytes

0,1xŒ!ḄQ

Try it online!

How it works

0,1xŒ!ḄQ Main link. Argument: [x, y]

0,1x     Repeat 0 x times and 1 y times.
    Œ!   Compute all permutations of the result.
      Ḅ   Unbinary; convert each permutation from base 2 to integer.
       Q  Unique; deduplicate the results.

Mathematica, 59 57 bytes

A usual outcome with Mathematica: high-level functions = good, long function names = bad.

#+##&~Fold~#&/@Permutations@Join[0&~Array~#,1&~Array~#2]&

Join[0&~Array~#,1&~Array~#2] creates a list with the correct number of 0s and 1s. Permutations generates all permutations of that list, without repetitions (as I learned) and in sorted order. #+##&~Fold~# (a golfuscated version of #~FromDigits~2) converts a list of base-2 digits into the integer they represent.

Previous version, before Martin Ender's comment:

#~FromDigits~2&/@Permutations@Join[0&~Array~#,1&~Array~#2]&