Conversation
I grepped through the FreeBSD source tree and for me, it seems to be totally unclear why these two specific functions are weak references. Such a construct is commonly used by FreeBSD's threading library (libthr) to override certain functions, but I can't find any traces of that. Just use the function name directly. This fixes a compiler warning as well (-Wmissing-prototypes).
…s(). - Add missing prototypes to openlibm.h for sincos() and sincosf(). - Mark the internal kernel functions static.
The global signgam variable is only part of the X/Open System Interfaces. It is not part of the POSIX base definitions nor the C standard. I'd rather have it disabled for my specific use-case, so introduce a new compilation flag that we can use to disable it.
OpenLibm uses the __weak_reference() macro for platforms where double and long double use the same layout. That way functions only need to be provided by the library once. The point is, in this specific case we want to use strong references; not weak references. Strong references can be used to give a symbol a second name. If you look at the resulting object file, you will have two symbols with the same offset and size. Weak references are different, in the sense that they are marked in such a way that they act as fallbacks. They are only used if an explicitly matching symbol is missing.
I guess the idea would be to eventually also install all of the openlibm*.h headers, instead of just openlibm.h. Make openlibm_fenv.h suitable for this purpose by moving all of the $ARCH/fenv.h headers next to it. We actually need this change to make OPENLIBM_USE_HOST_FENV_H work. Right now it's still broken, because the "#include <fenv.h>" performed by openlibm_fenv.h still pulls in $ARCH/fenv.h as $ARCH/ is added to the compiler include path.
|
Oh, wait. It turns out that our __strong_reference() macro only works if the target symbol has the same type as the source. Let me remove that part of the change for now. Edit: Done. |
…eeded." Unlike the __weak_reference() macro, __strong_reference() does type checking. It can only create the reference if the type of the source and the destination function match exactly. Even if double == long double in practice, they remain unequal at the language level.
|
Do you think the header files are all better off living in |
|
Thanks for merging! Yes. I think having the files in include/ would be a bit better. I would even consider moving the internal headers that are stored into include/ into src/ instead. We currently use them the wrong way around in my opinion. The reason why I decided to move the headers into src/ was only because openlibm.h was already stored there. As you mentioned, we should also add the install targets. Would you be interested in addressing these issues, or shall I send you a patch? By the way, I now only have a very small number of local changes against OpenLibm left. I'm only removing a couple of files from SRCS for functions that don't need to be built in my environment. I don't think there is anything useful that can be merged into OpenLibm itself. Feel free to release a new version of OpenLibm. Thanks! Ed |
|
You are right to do it the other way around. I realised it as I was looking at it. It would be great if you can send a PR. Thanks again for all the patches. These tremendously increase the quality of openlibm. |
2 participants
Hi there,
Attached is a small number of random, not so strongly related cleanup changes. A couple of these are actually needed for my use-case (fenv.h, OPENLIBM_ONLY_THREAD_SAFE). The others are just some cleanups for things I noticed along the way.
After these changes get merged, I no longer need to patch any OpenLibm source files. I only need tweak the Makefiles here and there.
Thanks,
Ed