[Lldb-commits] [lldb] New ThreadPlanSingleThreadTimeout to resolve potential deadlock in single thread stepping (PR #90930)

[via lldb-commits]

================
@@ -1312,11 +1312,13 @@ class Process : public std::enable_shared_from_this<Process>,
 
   size_t GetThreadStatus(Stream &ostrm, bool only_threads_with_stop_reason,
                          uint32_t start_frame, uint32_t num_frames,
-                         uint32_t num_frames_with_source,
-                         bool stop_format);
+                         uint32_t num_frames_with_source, bool stop_format);
 
   void SendAsyncInterrupt();
 
+  // Send an async interrupt and receive stop from a specific /p thread.
+  void SendAsyncInterrupt(Thread *thread);
----------------
jimingham wrote:

Do we need two API's here?  Why not have thread default to nullptr and document that that means we let the system choose which thread gets the interrupt?  It's also worth noting here that the "receive stop from specific thread" is artificial, so you might say something like "and attribute the stop to a specific thread" instead to make it clear we don't require the ability to actually have that thread receive the signal.

https://github.com/llvm/llvm-project/pull/90930