Extensive typological study doesn't appear to have been done on this. However, this is probably so because it has been difficult to say precisely what the mass/count distinction is about.
What we can understand is that the mass/count distinction tracks (imperfectly) an important human cognitive distinction between objects (or, countables) and substances (or, uncountables, "fake mass nouns" like furniture aside). As with many conceptual distinctions, different languages may mark the difference overtly, others may not. To give a different example, a language like West Greenlandic has a host of "pluractional" morphemes that attach to verbs, one of which indicates whether an event description picks out a series of iterative events, or whether it picks out a single event of continuous extent. English doesn't have such morphemes, but interestingly in this language we see certain event descriptions as underdetermining the two uses (e.g., John ran more than Mary last year, if true, can mean he ran on more occasions (iterative), or for a greater temporal or spatial extent (continuative)). So a conceptual distinction that is overtly marked in West Greenlandic is left implicit in English.
Similarly, English does not mark mass occurrences of nouns, but marks individual (with the indefinite article a(n)) and plural (with the -s suffix) occurrences of nouns. The bare form of a noun is often understood on a material reading (which is why the gruesome interpretation in There was boy all over the road, as if uttered after a horrific accident), and singular/plural occurrences are understood as picking out a singularity or multiplicity of individuals that fall under the concept named by the noun. If a language, for example, doesn't have a singular/plural distinction, one might think that noun occurrences there are understood via a combination of the preferred conceptualizations of things falling under the concept (e.g., we think of boy best as describing individual boys, and water best as describing an unindividuated substance), and extralinguistic context. The grammar doesn't force such speakers to express which conception they intend.
An oft-discussed counterexample to the idea that the distinction is universal, is the claim that all nouns are actually mass in Chinese---"count" nouns don't exist except in the sense that they are "built up" in combination with classifiers. However, as Cheng and Sybesma argue, and Chierchia accepts, even this language has reflexes of the conceptual distinction---i.e., it uses only "classifiers" for nouns that are understood as "count" (i.e., highly individuated), and only "massifiers" for those understood as "mass" (i.e., low individuation).
In sum, the role of grammar seems to be specifying what kind of conceptualization we're after. In English, there are a lot of nouns that are comfortable occurring as either mass or count, the difference ending up in whether we want to express a notion of the material, rock, or that of some quantity of individuals comprised of that material, rocks. Other occurrences, like muds seem less acceptable, but this could just be a matter of the frequency with which we encounter individuated MUD; if we can talk about kinds of muds or jars that contain some mud in a context, so that I bought three muds at the mud store isn't really so bad. Extralinguistic context can help us individuate even under the concept MUD.
Whether a language uses grammar to express the mass/count distinction will be a fact about particular languages, but the conceptual distinction is likely universal. (Hat tip to Dustin Chacón for discussion of Chinese.)