[Bug other/70945] Offloading: compatibility of target and offloading toolchains

["joseph at codesourcery dot com" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70945

--- Comment #4 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
On Fri, 13 May 2016, tschwinge at gcc dot gnu.org wrote:

> We could get rid of this heuristic (the property function_glibc_finite_math in
> combination with matching declarations' names) if the target compiler's early
> code transformation stages would accurately "describe" what they're doing, but
> that sounds like having to add some special/new "attributes" to glibc's
> <bits/math-finite.h>, which sounds more complicated.  I think the heuristic is
> safe enough; symbol names prefixed with an underscore are in the implementation
> namespace.

Even aside from offloading, there is a clear use for extensions to give 
better control over assembler names for different variants of functions: 
the libmvec issues where the vector function variants are not in 
one-to-one correspondence with the scalar function variants.