The original instigation for FSF
arose from the experience with the Unix OS under Bell Labs. Programmers, such as Stallman, wanted access to the source code so that they could make improvements. They wanted to share those improvements with other such programmers. AT&T came in and restricted the redistribution of the source code because they owned the OS and all surrounding utilities.
Now how this translates into whose freedoms are being protected is a different issue. Does this mean that I have to make all modifications available to the community under all circumstances? Well, the GPL does not have that general level of a requirement, as it sought to hone in only on those that attempted to restrict access to source code when it was redistributed. The current question is not whether a programmer is forced to release modifications, but in what circumstances a trigger is met that software is being distributed.
The purpose of the FSF was to build a toolset (and subsequent OS) that was based on the free exchange of source. The goal is to make that toolset as useful as possible, such that one need not use software for which one does not have source. Use of that software in proprietary systems is not antithetical to the FSF, other than in the way it effects the quality of free software. Learning and building are but secondary goals to the overall goal of making a set of free software that can be used by users.
But all this is pure pedantry. The reasons why people contribute to free software is quite varied, as noted by Stallman's constant irritation of Linus' engineering indifference. One of the interesting questions in all this is whether Linux (excuse me, GNU/Linux) will go with a GPLv3 license. Remember that the goals of Linux are not necessarily in alignment with FSF. Sure, some distributions may use it, but there will also be many that don't. And most will simply distribute the various components under whatever license (incompatible or not) that happened to be attached.
MySQL, the other example cited, is much different than either FSF or Linux. It is issued under the GPL by a company that also sells a proprietary version. It is in their interest to prevent other companies from improving and selling MySQL services - since that's how they make some of their money. They'd just as well be the only ones making proprietary improvements - not exactly what you'd call the same goals as FSF. The authors of Jasper Reports is in the process of trying to use this same sort of business model.