That's certainly a fair point. However this leads me to another problem with Foucault's theories which has been commented on many times before. Namely, that it does not begin from any normative basis. This has two consequences which are extremely important:
(1) His language mirrors that which it criticises. Baudrillard put it extremely succinctly - provided you've wasted enough time figuring out how decipher post-structuralist jargon:
"Foucault's discourse is a mirror of the power relations he describes. Its strength and seduction lie there, and not in its 'truth' index..... No, its strength and its seduction are in the analysis which unwinds the subtle meanderings of its object, describing it with tactile and tactical exactness, where seduction feeds analytical force and WHERE LANGUAGE ITSELF GIVES BIRTH TO THE OPERATION OF NEW POWERS. Such is also the operation of myth, right down to the symbolic effectiveness described by Levi-Strauss. Foucault's is not therefore a discourse of truth but a mythic discourse in the strong sense of the word, and I secretly believe that it has no illusions about the effects of truth it produces." (Baudrillard, Forget Foucault p.30 - My Emphasis)
The point is that since Foucault's theories claim to be ones of praxis if his discourse does in fact directly "mirror the power relations he describes" the solutions he proposes will necessarily be of this nature also. Here's an analogy: imagine all you ever learnt was a very strict variant of neo-classical economic theory and you wished to help build a socialist economy. You'd necessarily be unable to do so because the very language which you use to organise economic functions already contains too many presuppositions. Similarly Foucault's language is one of "power" and yet the problems he highlights are also ones of "power". Thus the term "power" is completely ambivalent - it is both good AND bad. (Towards the end of his life he began to realise this, especially in his last few lecture series, but these are generally ignored in most people's use of Foucault's thought).
(2) Following from this Foucault's theories, as you highlighted above, cannot call something dysfunctional. It presumes a perfect positivity, a perfect functionality all the time. I say the prison system is fucked, most would agree with me from a common sense perspective, you tell me, following Foucault's theories that its completely functional. This is a sort of functionalism. However, its a sort of functionalism that ALWAYS assumes the object is functional - there are no pathologies! Murder rate rises by 500%... no problem, its part of the system.
This is an impossible way to theorise society. I mean, I'll take a perfect example: hyperinflation under Mugabe. A Foucauldian would have to admit that this was not a failure at all. This was perfectly functional in the context of the web of power-relations in contemporary Zimbabwe. This is of course true. However, to maintain this you have to adhere to a relativism which crosses the line into total absurdity. To call Foucault's theories cynical would be an understatement. They prove so detached from reality as to be completely meaningless.