[Lldb-commits] [lldb] [lldb] Change lldb's breakpoint handling behavior (PR #96260)

[Pavel Labath via lldb-commits]

https://github.com/labath commented:

I like this change a lot, but I'm wondering if there's a clearer way to store the intermediate state. It seems like this info could go into the StopInfo class, which would have the benefit of automatically being backed up when doing expression evaluation. Is there a reason to not do that?

If the reason is that the StopInfo of the thread has been cleared by the time we call SetupForResume (and there's no good way to change that), maybe we could pass the old stop info to the function as an argument, since that's effectively what the function is doing -- deciding how to resume the thread based on how it was previously stopped.

Regarding the scenarios you mention in the description (moving the PC to a breakpoint site, setting a breakpoint site at the current pc), are these tested by any of the existing tests?

> Caroline Tice mentioned it is also a problem they've had with putting a breakpoint on _dl_debug_state

I believe this was fixed by 60b90b523323f8196a9e4a68b1f33358624c09eb.

https://github.com/llvm/llvm-project/pull/96260