[lld] [LLD] Tombstone LocalTU entry in .debug_names (PR #70701)

[Alexander Yermolovich via llvm-commits]

ayermolo wrote:

> > My previous comment still applies: the description should describe the choice of UINT32_MAX and UINT64_MAX, so that a curious reader doesn't have to read several external links to understand the choice.
> 
> Is there a way we could avoid hardcoding this, and just do "max of whatever the relocation size is"? (some earlier versions of this patch had switches over relocation types to try to do this - which seemed a bit verbose/brittle/repetitive - but if there was some way to generalize it tidily, I think that'd be ideal)

That is already what is happening. We always extend to 64bit MAX, and value is set by relocation type. 32 bit for DWARF32, and 64bit for DWARF64.

https://github.com/llvm/llvm-project/pull/70701