Cite just the more extensive one.
If this is indeed a case of self-plagiarism (in the ill-intended sense, i.e. artificially increasing one's apparent productivity), you will at least not provide the benefit for giving the authors two citations "at the price of one". At the same time, you provide your readers with the pointer to the more extensive resource.
Note that there are quite some arrangements where authors may publish an extended version of a previously published work elsewhere. Normally, this implies
- more than just a short section of additional material (e.g. in CS, the threshold is usually said to be around 30% of new content when publishing an extended journal version of a conference article) and
- a reference to the earlier work in the more extensive work.
However, the former may vary by field and venue and be subject to exceptional circumstances (e.g. a "special issue" may aim at mostly repeating an existing work with the primary goal of reaching a wider audience rather than disseminating new material as such), while failure to achieve the latter may be malicious, but may just as well be an oversight.