Test compile: mpn_lshift_com optimization 2

[Emmanuel Thomé]

On Wed, Jan 04, 2017 at 02:31:29PM +0100, Emmanuel Thomé wrote:
> I read that code, I'm puzzled too. Given that ceil(528/23)==23, and that
> 23*89 is 2047, I don't see how one could expect r[2048] and such to
> contain anything different from what is returned by malloc(). Which
> should be assumed to be garbage.

Before that commit, it was an entirely different thing.

https://gmplib.org/repo/gmp/diff/2349e1d2f62e/acinclude.m4

E.