r177329 - This code works around what appears to be a bug in another part of clang.

[Rafael Espíndola]

> Hi Rafael,
>
> I don't think I should have to revert this patch.
>
> I'm not even sure if there is another bug.
>
> I don't know the code that is calling this. It's just my opinion that there
> is some other issue.

Then you should investigate that. What we should never do is add

    if (!Fn) return; // should not happen

If it should not happen, this should be an assert (as it was before
your patch). It it is a logically valid condition, instead of the
wrong comment we should have a test where Fn is null.

> Someone that knows clang can explain how FD is a function declaration but GV
> is not a Function.
> Why is that my responsibility to sort that out?

Because you changed the code.

> This code is need for the mips test-suite to not regress and for my
> attribute work to continue.

Sorry. You should not push problems to others. Looks like you actually
have a testcase and you just need to reduce it. Please revert this
patch and reduce the testcase. Depending on what you find you can put
this patch back, but without the comment and with a testcase where Fn
is null.

> Reed
>
>

Cheers,
Rafael