saugustine added a comment. I've cleaned up this patch a bit. Now the only files that are in the x86_ARCH group are those that require 80 bits floats and cpu_model.c. Tests for all of these were already disabled on arm and powerpc (because neither has 80-bit floats), so we knew these library functions don't work on arm, but we compiled them anyway. It makes more sense to just not compile them at all if we know they require something the target architecture doesn't support.
I'm cc'ing some relevant people on this. The arm folks can comment on this patch (and I will add them to review) if there is a problem. In https://reviews.llvm.org/D36555#839829, @mgorny wrote: > Also, I think that if you're splitting them up, it'd also logical to move > them into a subdirectory, `x86-common` maybe. I'm reluctant to do this because that would require adding another layer to the file the filtering code around line 526, and that already has some special casing and is a bit hard to follow. Nevertheless, if you insist, I will. https://reviews.llvm.org/D36555 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
