[Dexter] Adapt to upcoming lldb stepping behavior

[jasonmolenda]

lldb will change how it reports stop reasons around breakpoints in the near future. I landed an earlier version of this change and noticed debuginfo test failures on the CI bots due to the changes. I'm addressing the issues found by CI at
#105594 and will re-land once I've done all of them.

Currently, when lldb stops at a breakpoint instruction -- but has not yet executed the instruction -- it will overwrite the thread's Stop Reason with "breakpoint-hit". This caused bugs when the original stop reason was important to the user - for instance, a watchpoint on an AArch64 system where we have to instruction-step past the watchpoint to find the new value. Normally we would instruction step, fetch the new value, then report the user that a watchpoint has been hit with the old and new values. But if the instruction after this access is a breakpoint site, we overwrite the "watchpoint hit" stop reason (and related actions) with "breakpoint hit".

dexter sets breakpoints on all source lines, then steps line-to-line, hitting the breakpoints. But with this new behavior, we see two steps per source line: The first step gets us to the start of the next line, with a "step completed" stop reason. Then we step again and we execute the breakpoint instruction, stop with the pc the same, and report "breakpoint hit". Now we can step a second time and move past the breakpoint.

I've changed the step method in LLDB.py to check if we step to a breakpoint site but have a "step completed" stop reason -- in which case we have this new breakpoint behavior, and we need to step a second time to actually hit the breakpoint like the debuginfo tests expect.

[jasonmolenda]

I've never looked at/touched the debuginfo tests / dexter before, I'm not sure who is best to ask to review this change. I'm open to feedback, I know the comment is a bit long but I wanted to be clear what this was doing. I wanted to have Dexter work with old or new lldb's equally but people will slowly forget the old lldb behavior.

[SLTozer]

Looks like a sensible way to solve the problem, thanks for this! Just so I understand correctly, in cases where we set breakpoints but exclusively advance using go(), as with the ConditionalController, the change in LLDB's behaviour will have no effect - is that correct?

[jasonmolenda]

Looks like a sensible way to solve the problem, thanks for this! Just so I understand correctly, in cases where we set breakpoints but exclusively advance using go(), as with the ConditionalController, the change in LLDB's behaviour will have no effect - is that correct?

The lldb go method does SBProcess.Continue(). If we are moving between the breakpoints with Continues, the behavior is the same as it was before. We execute until we hit the breakpoint instruction, and then when we resume the thread (step/continue/etc), the breakpoint is silently stepped past before execution is resumed. The issue I'm fixing in this PR only comes up because the step stops on the breakpoint instruction that hasn't been executed yet. I did run the dexter tests on my desktop with the stepping change and they passed, we'll see what happens when it hits the CI bots but I think this is sufficient. Thanks!