[PATCH] Consider the frame base isn't moving in calculateDbgValueHistory.

[David Blaikie]

This seems like it sort of devolves to something like what I suggested during Alexey's last round of improvements (that we just special case the frame pointer rather than trying to detect all the non-changing registers)

Alexey - is Fred's assessment here correct? That is, your previous changes/improvements are still being stopped by any call instruction?
If that's the case, are we gaining much/anything by having the complexity of detecting non-clobbered registers? Or should we just do a single special case for the frame pointer and throw the rest of the logic out?

http://reviews.llvm.org/D5366