Re: [RFC][PATCH 0/2] aarch64: Add optimized ASIMD versions of sinf/cosf

[Joseph Myers <joseph at codesourcery dot com>]

On Tue, 13 Jun 2017, Siddhesh Poyarekar wrote:

> > - math code should not be fsf assigned lgpl code, but universally
> > available, post it under non-restricted license first, then assign
> > it to fsf so it can be used everywhere without legal issues.
> 
> This is not a glibc requirement.  I don't know if we can even make that
> a requirement for arm/aarch64 code under the scope of the glibc project
> (i.e., it seems like a technical limitation - how do we reject arm
> patches in libc-alpha and redirect devs to cortex-strings or whatever
> else?), but that is something that Joseph or Carlos may be able to answer.

There should be no such requirement in glibc.  Contributors are of course 
free to contribute their implementations to other projects if they wish, 
and the standard FSF assignments generally allow you to use under other 
licenses code you contributed (the particular assignments may vary on 
whether prior notice to the FSF is needed before doing so).

I fully expect to continue to license new libm functions I contribute, or 
major pieces of new code (new source files) for existing functions, under 
the usual glibc LGPLv2.1+ license.

> As I mentioned earlier, realistic workloads are more or less a myth
> currently for math, so unless someone comes up with some, synthetic is
> all you'll get.

The synthetic benchmarks can at least have inputs biased to what we expect 
the common case is (for the trig functions I'd expect that to be absolute 
values below a few pi, with not so many large inputs).

-- 
Joseph S. Myers
joseph@codesourcery.com