Inverse Marking is often described in ways that are somewhat confusing. I hope to make matters a bit clearer.
Generally we have some "clear" thing that distinguishes subjects from objects - case marking, word order or some kind of congruence on verbs. But languages can lack all of these as distinguishing strategies, yet be able to distinguish which is the object and which is the subject.
Languages with direct/inverse alignment determine which to parse as subject and which to parse as object by a hierarchy. Normally, the noun higher up in the hierarchy is the subject.
The hierarchy might be something like 1 > 2 > human > animate > inanimate. Various complications may exist - a synonym might be in an unexpected place, zero-morph impersonal pronouns might occupy surprising places in the hierarchy giving odd ways of forming ~passives, which we'll get back to later.
So, if we have a sentence like {Noun1, Noun2, Verb}, it is the hierarchy that determines which is subject and which is object, no matter the order of the nouns in the actual sentence. Noun1>Noun2 implies Noun1 is the subject, Noun2 is the object, and vice versa. However, situations may occur where we want the other way around to obtain. For these, there is a morpheme that goes on the verb that simply reverses the parsing.
This is not a passive - a passive demotes the subject to an oblique position, and promotes the object to subject. Conceptually it might be a bit similar, but they don't really map easily to each other.
Noun1 > Noun2, Direct = Noun1 acts on Noun2Noun1 > Noun2, Reverse ≃ Noun1 is acted on by Noun2
What ruins this similarity is that Noun2 is a proper subject in the second clause, not an oblique thing. Syntactically it behaves like a subject. A better description is, really
Noun1 > Noun2, Reverse = Noun2 acts on Noun1
The differences might not be all that clear if you're not all that knowledgeable about the more special properties of subjects (and objects) - a post for a future occasion.
