[llvm] [DebugInfo] Merge partially matching chains of textual inclusions (PR #125780)

[Stephen Tozer via llvm-commits]

SLTozer wrote:

> If I understand your initial comment correctly, your idea means that we get a line-0 location if L1 and L2 from a `getMergedLocation(L1, L2)` call belong to different files even though there is an "include site" they have in common. Am I right?

That is correct - the main point is that if the line is 0, file typically doesn't matter.

> With the approach in this commit, we're trying to find a "common include location" and use it for locations from different files. Such resulting location should be visible for a debugger.

Do you have a test case for what you're describing? The test case in this commit right now covers the basic case, i.e. merging two instructions with the same line and the same file. A test case where we have two instructions with different files but a "common include location" might help clarify the intent here (and also should be part of the patch coverage).

With that said, as far as I understand what you're describing, a simpler approach is still possible. Even if we use the common include location as the merged location, that is still only a single location - the inlined-chain logic you've adapted here is necessary to produce a list of locations for each inline frame. In this case, the only thing you'd need AFAICT is to modify the `GetNearestCommonScope` code in `getMergedLocation` to track inclusions in addition to scopes, and use the nearest common inclusion if one exists.

https://github.com/llvm/llvm-project/pull/125780