Thank you for taking the time to write to us, but this is not
a bug. Please double-check the documentation available at
http://www.php.net/manual/ and the instructions on how to report
a bug at http://bugs.php.net/how-to-report.php

Currently it's expected, as the 'b' prefix doesn't produces a separated token.

I doubt that this is expected. Only single characters can be represented using a one-char string token, because the token number and token text are identical in this case.
'b"' is not a single character and thus can't be returned as a string token, because the token number is chr('"'), not chr('b"') (which is chr('b')).

This is defined in the scanner this way.
Currently we don't need the b prefix and have it for forward compatibility only. I doubt we won't add an extra token for the 'b' prefix (which most likely creates conflicts)
If anything like this is done it should be done according to the needs of a future use of that prefix.

@johannes: My main issue here is that the output of token_get_all is malformed. 'b"' isn't a valid string token, because it has more than one character.

But really, I have to admit that I can't think of a way how to cleanly solve this issue, without breaking BC for the tokenizer ext. So I'd say I agree with the Wontfix, especially as this is easy to detect and fix in userland code.