Potential self-hosting failure

[Renato Golin via cfe-commits]

Hi John,

It seems the LEApcrel patches have broken our self-hosting:

http://lab.llvm.org:8011/builders/clang-cmake-thumbv7-a15-full-sh/builds/1550

http://lab.llvm.org:8011/builders/clang-cmake-armv7-a15-selfhost-neon/builds/1349

http://lab.llvm.org:8011/builders/clang-cmake-armv7-a15-selfhost/builds/1845

The range in each is big, but the overlapping range is actually just
303051 ~ 303054.

Since two of those patches are yours and since this is a self-hosting
issue, my money is on your patches, not the Dwarf one. :)

The tests don't help much, unfortunately.

I have had problems like this in Clang, where the code assumed some
ABI that wasn't as generic as initially assumed, and changes in
relocation are normally the ones that expose those wrong assumptions.

Can you have a look on your side, while we're testing on our side, too?

Thanks!
--renato