[llvm] r258471 - Do not lower VSETCC if operand is an f16 vector

[Tim Northover via llvm-commits]

> On 28 Jan 2016, at 10:03, Hans Wennborg <hans at chromium.org> wrote:
> It was suggested on PR26348 that this should be merged to 3.8. What do
> you think?

Ah, was just about to reply to that one! Seems like it's obviously the right thing to do: the patch is trivially right, and the entire space was broken in 3.7 so it's virtually impossible to regress anyway.

Cheers.

Tim.