[llvm] [JITLink] Some cleanups to EHFrameSupport (PR #66707)

[via llvm-commits]

lhames wrote:

There were multiple simplifications in this piece of code. Assuming one record per block seems reasonable (we have the splitter as you noted), but rejecting objects with existing FDE-to-CIE edges is a change in behavior -- I'm asking for the reasoning behind that.

https://github.com/llvm/llvm-project/pull/66707