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

[Rafael Espíndola]

> I'm going to remove the comment and supply the test case.

Good. I have reverted the patch while that happens. I understand you
might have a crash that blocks you, but that is not part of how our
code review process works.

Just having a test with a fixme saying something on the lines of: this
should probably be handled differently is OK.

If you don't even have time to reduce a testcase, please keep changes
like this in a local branch.

So, about the testcase. It is the one in llvm.org/pr15538 that needs
reducing, right? If that is so I am more than happy to help you reduce
it and create a testcase we can add to clang.

> In the case where I am passed a Function Declaration and  Global Value which
> is not convertible to a Function, there is nothing for me to do here. There
> are no mips attributes which could be processed in this code that do not
> conform to that.
>
>
> Reed
>

Cheers,
Rafael