If you get an insight (or are referencing an idea/discovery), you need to cite where it comes from. Good/bad journal. Even trade journals, websites, personal communications, etc.
In addition, if you are surveying the literature (or semi-surveying at the intro to a paper), than you should list all the papers that would be useful for readers. I would use some judgment (this is a favor you are doing for the reader). List the most useful papers (good ones, mini-reviews, etc.). You should not feel the need to show everything (impossible in active fields, e.g. electronic thin films). But give the reader the best ones--after that he is on his own to get "cites from the cites" or to do his own ad hoc lit search.
In the event that citations are limited, I would probably lean more towards inclusion (of junkier stuff). Conversely, if there's a lot out there...show the best stuff and skip the "this is a paper about an individual film, not a set of them" type dross.
At the end of the day, this is not a Euclidean proof, but a thoughtful help to the reader. And some element of selection/exclusion is what you are delivering to them (and they will appreciate it...after all they can do their own search in addition). It can't be done perfectly since it is always up for debate. But it can be done thoughtfully and helpfully. Just like most practical problems in business, government, etc.
I definitely wouldn't be tentative about junky journals if you know the papers are good or if that's all there is. If anyone contests it, just say (truthfully hopefully) "I have spent years reading and working in this area and am one of the world's experts in it" (as an older grad student, this is actually a reasonable position...you are the expert in your thesis.)
All that said, I have found that junky journals are usually worse. But you will develop your own intuitions here, within your field. I know that in physics and chemistry, the APS and ACS specialty journals still are quite good. Even some of the Elsevier and the like competition. And Nature/Science (while very prestigious and many good papers) do tend to have more of a problem with headline-seekers and fraud. Since it is a big deal to get papers there and since they see themselves as more than just science journals, but somewhat news-y. Non-US national societies (especially excepting France, Germany, UK) tend to be worse. TEND. But I would look with a wary eye at Japanese Applied Films or the like. And I definitely wouldn't bother ever skimming the table of contents of such a magazine--only coming to their papers as parts of lit searches or "cite from cite" tracing.
But of course, look at the specific papers. You should develop enough perspective (and be enough of a critical thinker) to recognize some of the attention-seeker crappiness in Nature and/or find a diamond in the rough in Japanese Applied Films. And then obviously put the better stuff forward. Don't be a mouse.