In fact, thinking some more about it, it strikes me that the proper
approach is to just do:
- the shared library just assumes that the startup state is always the
state that has been introduced by a "fninit" (that's what Linux has
done since version 0.11, modulo bugs)
- if that ever changes, _than_ can we start thinking about it.
Let's face it: the whole _idea_ with shared libraries is that you share
code AND that you're able to react to changes in the OS without having to
recompile all your binaries.
As such, a shared library that imposes changes on the OS is just a stupid
idea. You shouldn't go searching for problems you think might be an issue
some day, you should fix problems you know about. FPUCW is not a problem
today, and as far as I can tell there is no reason why it should ever
become one.
Let's aim for a fast, lean and mean shared library rather than bloating it
up without good reason. I already hate how all my binaries do this
completely stupid search for some random "locale" file even though I don't
have one and never asked for one, and it just slows the system down simply
because of a stupid shared library.
Linus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/