This is the Dwarves' name for the Vala Aulë. No translation is ever given for it, but the most common guess I've seen is "maker", due to the Dwarves' claim that Aulë created them. This is a very reasonable idea, especially in light of the Dwarven usage of descriptive names. The issue then is the etymology.
There are a number of way to break this name down. The first is fairly straightforward. The root for "to make, create" would be M-H-L, which would be placed in a singular vowel template CaCaC to form "a thing or person that makes, creates; a maker, creator". This would be parallel to F-L-K "to hew" and felak "hewer". Both would be seen as as active participles: the "making" and the "hewing". The only issue I take with this view is that the vocalization CaCaC is not seen elsewhere for a noun. The closest is zâram. That doesn't negate the possibility, though, especially since at least some Semitic languages (Arabic) treat nouns and adjectives as being the same word class. So, as we see CaCaC in several adjectives, like narag and zahar, it's very possible that the pattern is also used for words English speaker would normally think of as nouns as well. We just don't see one in our limited corpus of Khuzdul words.
Another similar analysis is to theorize a root M-H-' "to make, create" and place that into a same template CaCaC to form a verb maha "make, create". From there, an agentive suffix would be added, such as -al to become Mahal "maker". Some would suggest that we see the same suffix in Azaghâl, with '-Z-Gh being a verbal root "to war". Azaghâl would then mean "warrior". I don't think that's the case. In any event, an explanation would be needed for why the long vowel in Azaghâl but not Mahal.
The third explanation does have some circumstantial evidence, inside Khuzdul and out, to back it up. Mahal could be divided into ma- + hal'. H-L-' would be a root, while ma- would be the passive participle marker seen in Mazarbul. In fact, Mahal would then exactly match mazarb "a record" in structure. As a verbal root, the best concept I can think of for H-L-' would be "to revere, worship". Thus, Mahal would be "Revered". This break down becomes even more interesting when you see that the root H-L-' can be viewed as very similar to the root(s) for Allah and Elohim. That root is '-L-H, or something similar. There are different theories about it, and some of those have those sound rearranged or reinterpreted. Nonetheless, H-L-' is comparable. One suggested meanings of the Semitic root(s) for Allah and Elohim is "to adore, worship". That would line up exactly with the proposal of Mahal "Revered".
The gloss of "Revered" might be even more interesting if one of the other two suggestions are also true. Picture a root H-L-' "to revere", with passive participle prefix ma- added becomes Mahal "Revered". Also, the root M-H-L means "to make, create", and added to the template CaCaC becomes mahal "maker". So, it could actually be seen as being both "Revered" and "Maker".
The Silmarillion, ch 2
The War of the Jewels, pg 10