[Lldb-commits] [lldb] [lldb] Change lldb's breakpoint detection behavior [WIP] (PR #105594)

[Jason Molenda via lldb-commits]

jasonmolenda wrote:

I have a fix for the third of four issues needing investigation - the debuginfo dexter test fails.  In this case, it places a breakpoint on every source line, then step-in's through the program.  With old lldb, we would step to a BreakpointLocation at the start of a source line, and the breakpoint-hit stop reason would override the "step-in completed" stop reason.  With this new algorithm, we stop at the BreakpointLocation that has not executed yet, so our stop reason is "step-in completed".  Then when we step again, we hit the breakpoint we're stopped at, and the stop reason is "breakpoint-hit" -- but we have not advanced.  So dexter needs to step twice to get past these breakpoints-on-every-source-line as it step-in's through the program.  I have a PR that fixes it, asking around if I should put it up as a separate PR or include it as a commit on this.

https://github.com/llvm/llvm-project/pull/105594