Plagiarism control is a good tool to protect the journal quality, recently it became a nightmare for the authors who cite their own work, even material and methods are sometimes subjected to plagiarism.
Ammar - you can't really have self-plagiarism. Commonly, plagiarism is mainly seen as presenting someone else work as if it is your own and not acknowledging the other authors work. You can't do that with your own work. What you might be more referring to is 'salami-slicing' publications or research findings i.e. submitting multiple publications to different journals (which are essentially reporting the same things) - but 'pretending' that they are different - or hoping that other journals will not notice. That is more research 'fraud' than plagiarism - depending on the extent of it.
It seems there is no such thing as "self plagiarism", as noted by Mr. Whitehead, but the question is if it is ethical to move large chapters from one work to a future self work if both works deal with the same issue.
COPE do not use the term "self plagiarism" they say "duplicate publication" - and I agree it can be hard for authors. However all plagiarism-checking softwares are simply text-matching programs, and are not intelligent enough to recognise your own works. CrossCheck (which uses TurnItIn software) is probably the best. Google also works pretty well.
Self plagiarism is prohibited in the sense that if an author used part of his/her previously published work in another work and did not properly cite such previous work, then there might be an infringement on the publisher's copyright (of course, if the author doesn't fully owned the copyright, which is the case in many publications where the publisher asked the author for transfer of copyright).
Plagiarism is unethical, and journals, especially refereed ones should check carefully each paper that is a candidate for publication in these journals. I myself, found twice identical papers appearing four years apart with names of different authors.
Self-plagiarism is misleading and against integrity, any research should come up with new knowledge that advance our understanding. Without proper citation, self-plagiarism deceive the reader by implying that the paper presenting completely new discoveries or knowledge. In addition, you do not own those words if you use it in previous published research for the copyrights reasons.