Preserve ThreadPlanStacks for unreported threads

[jingham]

Summary

This is the final patch in the work I've done so far to support Process plugins (or their associated OS Plugins) not reporting all threads on every stop.

This commit moves the thread plan updating and dumping into the ThreadPlanStackMap. It adds a setting (target.process.plugin-reports-all-threads) you can use to indicate that you want us to preserve unreported thread plans, and a "thread plan prune" command you can use to remove any orphaned plans.

If this is approved, I'd like to check in all these pieces. At this stage if you opt into preserving ThreadPlanStacks, the process is pretty minimal. I want to get this more by-hand version working because there are a lot of OS Plugins in the wild (there's on in every mach_kernel dSYM) and we have not supported them well. So I want to get something that supports the extant plugins.

The next step is to add a way for the plugins to sync up with the thread plans, and then call that at public stop time, but I'd like to do that as a separate patch.

[labath]

I still don't know much about thread plans, but I didn't see anything that would stand out. All my comments are fairly superficial.

lldb/source/Target/TargetProperties.td

187

If this is relevant only for os plugins, then it would be good to reflect that in the name as well.

lldb/test/API/functionalities/plugins/python_os_plugin/stepping_plugin_threads/main.cpp

2–8

We don't put licence headers on tests.

21

unistd.h is not portable, and it doesn't look like you are using anything from it.

34

The synchronization here seems to rely on the fact that there will be no spurious wakeups. The simplest way to guard against that is to pass a lambda into the wait function g_cv.wait(func_lock, [&] { return g_value == 2; })

39–42

The clang/gcc/icc anotations are not used, and who knows if they are even accurate at this point.

lldb/test/API/functionalities/thread_plan/TestThreadPlanCommands.py

24

this looks like a model example for using self.filecheck

[jingham]

Addressed most of Pavel's review comments.

[jingham]

Addressed most of the review comments. I don't see that using file check would make the plan output checking any easier. The current function seems clear and easy to use. Trying to dynamically insert the passed in matches into a filecheck test file and then run file check on that doesn't seem like it would be any easier than what is here. I don't need any of the substitutions or any of the other fancy things FileCheck provides, that just seems like overkill. But maybe I'm misunderstanding what you meant.

lldb/source/Target/TargetProperties.td

187

I thought about that. In the future a regular Process plugin might decide it was too expensive to report all threads as well. There's nothing in this patch that wouldn't "just work" with that case as well. Leaving the OS out was meant to indicate this was about how the Process plugin OR any of its helpers (e.g. the OS Plugin) produces threads.

lldb/test/API/functionalities/thread_plan/TestThreadPlanCommands.py

24

I don't see that. Do you mean write a file check matching file with the contents of the three arrays in this function, and then run file check on that? Or did you mean convert the call sites into embedded patterns and then call file check on myself? But then I'd have to also embed the headers in the body such that they came in the right order for all the tests. Neither of these seem like attractive options to me. It doesn't seem like it will be hard to maintain this little checker, and I don't see what I would gain by using file check instead.

[labath]

lldb/source/Target/TargetProperties.td

187

Well, I am hoping that if we ever extend this support to the regular process plugins, we will implement it in a way where the plugin itself can tell us whether it is operating/supporting this mode (which I guess would involve a new gdb-remote packet and some specification of what exactly should happen when it gets sent), instead of relying on the user to set this correctly.

I mean, in an ideal world this is I would want to happen with the python plugins as well, but it seems that we are stuck with some existing plugins which already do that. However, I wouldn't want to plan on the same thing happening again. :)

lldb/test/API/functionalities/thread_plan/TestThreadPlanCommands.py

24

What I meant was doing doing something like:

self.filecheck("thread plan list %d"%(current_id), __file__, "--check-prefix=CHECK1")
# CHECK1:      Active plan stack:
# CHECK1-NEXT: Element 0: Base thread plan
# CHECK1-NEXT: Element 1: Stepping over line main.c
# CHECK1-EMPTY:

instead of self.check_list_output(..., [], ["Stepping over line main.c"])

The main advantage of that I see is in the error messages it produces. If anything goes wrong, the most likely error message we're going to get is "9 != 10, Too many elements in match arrays". The second most likely error message is "False is not True, Didn't find active plan Stepping over line main.c". If the same happens while using filecheck, it would print the string it is trying to match, the line it is trying to match it against, and the line of a "possible intended match", which makes it much easier to figure out what is going on without debugging the test. Now you could embed all of that into this matcher function, but then you're sort of reinventing FileCheck.

The second advantage is that it is easier to see what is the "expected" output supposed to be when it is layout out like this, instead of embedded in the logic of the matcher. That may introduce some amount of repetition for specifying the common parts of the output, but that is not necessarily bad (because it's more readable). If there is a huge amount of repetition, then there are ways to avoid that via using multiple prefixes:

FileCheck --check-prefixes=ALL,FIRST
...
FileCheck --check-prefixes=ALL,SECOND
# ALL: Active plan stack:
# ALL: Element 0: Base thread plan
# FIRST: Element 1: Stepping over line main.c
# SECOND: Element 1: ???

But that again makes it harder to see the expected output, so it should be balanced against that.

[jingham]

lldb/source/Target/TargetProperties.td

187

Right, I put this in mostly to help backwards compatibility. For instance, another OS Plugin I know about handles some older cooperative threading scheme. That one does report all threads on each stop. I didn't want to force them to do anything to keep their plugin working as well as it did before. That's also why I set the default to true here.

Even when we have plugins that actually support not reporting all threads, you could imagine somebody having a Process plugin that supports both modes - particularly early in the development of its support for this feature, and in some corner cases the "doesn't report all threads" mode has some subtle problem. Having this setting will allow people to get the slower but more assuredly correct behavior till it works 100% reliably. So I still think the setting has some value.

But I agree, going forward there should be some kind of handshake between the ThreadPlanStackMap and the Process Plugin, either a "I've reported all threads now" which could trigger a prune, or a "Is TID X still alive" which the generic code could use to balance the cost of keeping outdated stacks alive against when we want to ask about all threads.

lldb/test/API/functionalities/thread_plan/TestThreadPlanCommands.py

24

Yeah, if you don't mind, I'm going to keep it as it is. To do it the way you suggest you either would have to repeat all the header stuff over and over or make some compound section for the filecheck part of the test with FIRST and SECOND etc for the tests. That's sort of okay for a few cases but if you have more that gets hard to follow and edit without making mistakes.

The way it is now, you say explicitly what you want to see in the the categories at the check site, which seems really clear to me. And the processing code is not terribly difficult, so I don't think maintaining it will be a problem.

[labath]

lldb/source/Target/TargetProperties.td

187

All of this sounds fine, and I wouldn't mind having a setting like that, even after the support for partial thread lists is considered "stable". However, that sounds like a different setting to me -- that's like a directive to the process plugin about how it should behave, whereas this setting is more like a statement of fact about what the plugin does.

The setting could be later repurposed to do both, but it's not clear to me whether that is useful. Like, since we already support plugin-specific settings, the plugin which decides to implement both modes of operation could expose that as a custom setting. That way, one wouldn't get the impression that this setting applies to any process plugin...

lldb/test/API/functionalities/plugins/python_os_plugin/stepping_plugin_threads/main.cpp

41

spurious wake up danger here too

lldb/test/API/functionalities/thread_plan/TestThreadPlanCommands.py

24

Well, I wouldn't say I don't mind, but I am not interested in making a protracted debate out of this. But if you are going to stick to this, then I think you should do something to make the error messages coming out of here more reasonable. The least you could do is use self.runCmd instead of handle command (so that the command output is visible in trace mode at least), and replace all assertTrue(needle in haystack) with assertIn(needle, haystack) (to make it easier to see what is being matched against what).

[jingham]

Add a variable backing the std::condition_variable in the source for the stepping test to catch spurious wakeups.

AssertTrue -> AssertIn in the ThreadPlanList test.

[jingham]

lldb/source/Target/TargetProperties.td

187

Okay. I don't want to have to design that right now.

Since this is just a workaround for the fact that the extant OS plugins don't have a way to tell me this fact, I'll go back to using this just for OS plugins. In that case, I think I should move it to experimental. Once we add an API to the various plugins to advertise how they work, it won't be needed. The benefit of experimental settings is that their absence doesn't cause the "settings set" to raise an error, so people can leave this in their lldbinit's and it won't cause problems once we use it.

Now I just have to figure out how experimental settings work in the new tablegen way of doing properties...

lldb/test/API/functionalities/plugins/python_os_plugin/stepping_plugin_threads/main.cpp

41

g_value had a different purpose and using it more widely as the variable backing the condition_variable was awkward, so I added an explicit condition variable.

lldb/test/API/functionalities/thread_plan/TestThreadPlanCommands.py

24

I should have guessed there was an assertIn as well! Converted over to use that.

I didn't use runCmd since I wasn't using any of its facilities besides dumping the command result. But I added a check for self.TraceOn and if true I dump the command output. So that should do what you wanted.

[labath]

lldb/source/Target/TargetProperties.td

187

An experimental setting sounds good to me.

[jingham]

I changed target.process.plugin-reports-all-threads to target.experimental.os-plugin-reports-all-threads.

[labath]

Looks fine to me, though I can't say I'm much of a thread plan expert (but then again, I don't know if anyone except you is).

[jingham]

In D76814#1958988, @labath wrote:

Looks fine to me, though I can't say I'm much of a thread plan expert (but then again, I don't know if anyone except you is).

That's not a great situation, but it is where we currently are...

lldb/include/lldb/Target/Process.h

public:
 
   bool IsRunning() const;
 
   DynamicCheckerFunctions *GetDynamicCheckers() {
     return m_dynamic_checkers_up.get();
   }
 
   void SetDynamicCheckers(DynamicCheckerFunctions *dynamic_checkers);
-  
+/// Prune ThreadPlanStacks for unreported threads.
+///
+/// \param[in] tid
+///     The tid whose Plan Stack we are seeking to prune.
+///
+/// \return
+///     \b true if the TID is found or \b false if not.
+bool PruneThreadPlansForTID(lldb::tid_t tid);
+
+/// Prune ThreadPlanStacks for all unreported threads.
+void PruneThreadPlans();
+
   /// Find the thread plan stack associated with thread with \a tid.
   ///
   /// \param[in] tid
-  ///     The tid whose Plan Stack we are seeking..
+  ///     The tid whose Plan Stack we are seeking.
   ///
   /// \return
   ///     Returns a ThreadPlan if the TID is found or nullptr if not.
   ThreadPlanStack *FindThreadPlans(lldb::tid_t tid);
-  
-  void AddThreadPlansForThread(Thread &thread);
-  
-  void RemoveThreadPlansForTID(lldb::tid_t tid);
+  /// Dump the thread plans associated with thread with \a tid.
+  ///
+  /// \param[in/out] strm
+  ///     The stream to which to dump the output
+  ///
+  /// \param[in] tid
+  ///     The tid whose Plan Stack we are dumping
+  ///
+  /// \param[in] desc_level
+  ///     How much detail to dump
+  ///
+  /// \param[in] internal
+  ///     If \b true dump all plans, if false only user initiated plans
+  ///
+  /// \param[in] condense_trivial
+  ///     If true, only dump a header if the plan stack is just the base plan.
+  ///
+  /// \param[in] skip_unreported_plans
+  ///     If true, only dump a plan if it is currently backed by an
+  ///     lldb_private::Thread *.
+  ///
+  /// \return
+  ///     Returns \b true if TID was found, \b false otherwise
+  bool DumpThreadPlansForTID(Stream &strm, lldb::tid_t tid,
+                             lldb::DescriptionLevel desc_level, bool internal,
+                             bool condense_trivial, bool skip_unreported_plans);
+  /// Dump all the thread plans for this process.
+  ///
+  /// \param[in/out] strm
+  ///     The stream to which to dump the output
+  ///
+  /// \param[in] desc_level
+  ///     How much detail to dump
+  ///
+  /// \param[in] internal
+  ///     If \b true dump all plans, if false only user initiated plans
+  ///
+  /// \param[in] condense_trivial
+  ///     If true, only dump a header if the plan stack is just the base plan.
+  ///
+  /// \param[in] skip_unreported_plans
+  ///     If true, skip printing all thread plan stacks that don't currently
+  ///     have a backing lldb_private::Thread *.
+  void DumpThreadPlans(Stream &strm, lldb::DescriptionLevel desc_level,
+                       bool internal, bool condense_trivial,
+                       bool skip_unreported_plans);
 
   /// Call this to set the lldb in the mode where it breaks on new thread
   /// creations, and then auto-restarts.  This is useful when you are trying
   /// to run only one thread, but either that thread or the kernel is creating
   /// new threads in the process.  If you stop when the thread is created, you
   /// can immediately suspend it, and keep executing only the one thread you
   /// intend.
   ///
public:
 
     virtual ~NextEventAction() = default;
 
     virtual EventActionResult PerformAction(lldb::EventSP &event_sp) = 0;
     virtual void HandleBeingUnshipped() {}
     virtual EventActionResult HandleBeingInterrupted() = 0;
     virtual const char *GetExitString() = 0;
     void RequestResume() { m_process->m_resume_requested = true; }
-    
   protected:
     Process *m_process;
   };
 
   void SetNextEventAction(Process::NextEventAction *next_event_action) {
     if (m_next_event_action_up.get())
       m_next_event_action_up->HandleBeingUnshipped();
 

lldb/include/lldb/Target/Target.h

public:
 
   const ProcessLaunchInfo &GetProcessLaunchInfo();
 
   void SetProcessLaunchInfo(const ProcessLaunchInfo &launch_info);
 
   bool GetInjectLocalVariables(ExecutionContext *exe_ctx) const;
 
   void SetInjectLocalVariables(ExecutionContext *exe_ctx, bool b);
   
+  bool GetOSPluginReportsAllThreads() const;
+
+  void SetOSPluginReportsAllThreads(bool does_report);
+
   void SetRequireHardwareBreakpoints(bool b);
 
   bool GetRequireHardwareBreakpoints() const;
 
   bool GetAutoInstallMainExecutable() const;
 
   void UpdateLaunchInfoFromProperties();
 
+
 private:
   // Callbacks for m_launch_info.
   void Arg0ValueChangedCallback();
   void RunArgsValueChangedCallback();
   void EnvVarsValueChangedCallback();
   void InputPathValueChangedCallback();
   void OutputPathValueChangedCallback();
   void ErrorPathValueChangedCallback();

lldb/include/lldb/Target/Thread.h

public:
   ///   Discard all plans up to and including this user plan given by this
   ///   index.
   ///
   /// \return
   ///    \b true if there was a thread plan with that user index, \b false
   ///    otherwise.
   bool DiscardUserThreadPlansUpToIndex(uint32_t thread_index);
 
-  /// Prints the current plan stack.
-  ///
-  /// \param[in] s
-  ///    The stream to which to dump the plan stack info.
-  ///
-  void DumpThreadPlans(
-      Stream *s,
-      lldb::DescriptionLevel desc_level = lldb::eDescriptionLevelVerbose,
-      bool include_internal = true, bool ignore_boring = false) const;
-
   virtual bool CheckpointThreadState(ThreadStateCheckpoint &saved_state);
 
   virtual bool
   RestoreRegisterStateFromCheckpoint(ThreadStateCheckpoint &saved_state);
 
   virtual bool
   RestoreThreadStateFromCheckpoint(ThreadStateCheckpoint &saved_state);
 
protected:
   friend class StackFrameList;
   friend class StackFrame;
   friend class OperatingSystem;
 
   // This is necessary to make sure thread assets get destroyed while the
   // thread is still in good shape to call virtual thread methods.  This must
   // be called by classes that derive from Thread in their destructor.
   virtual void DestroyThread();
-  
   ThreadPlanStack &GetPlans() const;
 
   void PushPlan(lldb::ThreadPlanSP plan_sp);
 
   void PopPlan();
 
   void DiscardPlan();
 
lldb::StateType m_temporary_resume_state; ///< This state records what the
                                             ///thread was told to do by the
                                             ///thread plan logic for the current
                                             ///resume.
   /// It gets set in Thread::ShouldResume.
   std::unique_ptr<lldb_private::Unwind> m_unwinder_up;
   bool m_destroy_called; // This is used internally to make sure derived Thread
                          // classes call DestroyThread.
   LazyBool m_override_should_notify;
+  mutable std::unique_ptr<ThreadPlanStack> m_null_plan_stack_up;
 
 private:
   bool m_extended_info_fetched; // Have we tried to retrieve the m_extended_info
                                 // for this thread?
   StructuredData::ObjectSP m_extended_info; // The extended info for this thread
 
 private:
-  
   void BroadcastSelectedFrameChange(StackID &new_frame_id);
 
   DISALLOW_COPY_AND_ASSIGN(Thread);
 };
 
 } // namespace lldb_private
 
 #endif // LLDB_TARGET_THREAD_H

lldb/include/lldb/Target/ThreadPlan.h

public:
   ///   A  pointer to the thread plan's owning thread.
   Thread &GetThread();
 
   Target &GetTarget();
 
   const Target &GetTarget() const;
 
   /// Print a description of this thread to the stream \a s.
-  /// \a thread.
+  /// \a thread.  Don't expect that the result of GetThread is valid in
+  /// the description method.  This might get called when the underlying
+  /// Thread has not been reported, so we only know the TID and not the thread.
   ///
   /// \param[in] s
   ///    The stream to which to print the description.
   ///
   /// \param[in] level
   ///    The level of description desired.  Note that eDescriptionLevelBrief
   ///    will be used in the stop message printed when the plan is complete.
   virtual void GetDescription(Stream *s, lldb::DescriptionLevel level) = 0;
protected:
   bool m_takes_iteration_count;
   bool m_could_not_resolve_hw_bp;
   int32_t m_iteration_count = 1;
 
 private:
   // For ThreadPlan only
   static lldb::user_id_t GetNextID();
 
-  Thread *m_thread;
+  Thread *m_thread; // Stores a cached value of the thread, which is set to
+                    // nullptr when the thread resumes.  Don't use this anywhere
+                    // but ThreadPlan::GetThread().
   ThreadPlanKind m_kind;
   std::string m_name;
   std::recursive_mutex m_plan_complete_mutex;
   LazyBool m_cached_plan_explains_stop;
   bool m_plan_complete;
   bool m_plan_private;
   bool m_okay_to_discard;
   bool m_is_master_plan;

lldb/include/lldb/Target/ThreadPlanStack.h

 // them from an owning thread, then reattach them by TID.
 // The ThreadPlanStack holds the ThreadPlans for a given TID.  All its methods
 // are private, and it should only be accessed through the owning thread.  When
 // it is detached from a thread, all you can do is reattach it or delete it.
 class ThreadPlanStack {
   friend class lldb_private::Thread;
 
 public:
-  ThreadPlanStack(Thread &thread) {}
+  ThreadPlanStack(const Thread &thread, bool make_empty = false);
   ~ThreadPlanStack() {}
 
   enum StackKind { ePlans, eCompletedPlans, eDiscardedPlans };
 
   using PlanStack = std::vector<lldb::ThreadPlanSP>;
 
-  void DumpThreadPlans(Stream *s, lldb::DescriptionLevel desc_level,
+  void DumpThreadPlans(Stream &s, lldb::DescriptionLevel desc_level,
                        bool include_internal) const;
 
   size_t CheckpointCompletedPlans();
 
   void RestoreCompletedPlanCheckpoint(size_t checkpoint);
 
   void DiscardCompletedPlanCheckpoint(size_t checkpoint);
 
public:
 
   ThreadPlan *GetInnermostExpression() const;
 
   void WillResume();
 
 private:
   const PlanStack &GetStackOfKind(ThreadPlanStack::StackKind kind) const;
 
+  void PrintOneStack(Stream &s, llvm::StringRef stack_name,
+                     const PlanStack &stack, lldb::DescriptionLevel desc_level,
+                     bool include_internal) const;
+
   PlanStack m_plans;           ///< The stack of plans this thread is executing.
   PlanStack m_completed_plans; ///< Plans that have been completed by this
                                /// stop.  They get deleted when the thread
                                /// resumes.
   PlanStack m_discarded_plans; ///< Plans that have been discarded by this
                                /// stop.  They get deleted when the thread
                                /// resumes.
   size_t m_completed_plan_checkpoint = 0; // Monotonically increasing token for
                                           // completed plan checkpoints.
   std::unordered_map<size_t, PlanStack> m_completed_plan_store;
 };
 
 class ThreadPlanStackMap {
 public:
-  ThreadPlanStackMap() {}
+  ThreadPlanStackMap(Process &process) : m_process(process) {}
   ~ThreadPlanStackMap() {}
 
+  // Prune the map using the current_threads list.
+  void Update(ThreadList &current_threads, bool delete_missing,
+              bool check_for_new = true);
+
   void AddThread(Thread &thread) {
     lldb::tid_t tid = thread.GetID();
     auto result = m_plans_list.emplace(tid, thread);
   }
 
   bool RemoveTID(lldb::tid_t tid) {
     auto result = m_plans_list.find(tid);
     if (result == m_plans_list.end())
public:
   }
 
   void Clear() {
     for (auto plan : m_plans_list)
       plan.second.ThreadDestroyed(nullptr);
     m_plans_list.clear();
   }
 
+  // Implements Process::DumpThreadPlans
+  void DumpPlans(Stream &strm, lldb::DescriptionLevel desc_level, bool internal,
+                 bool ignore_boring, bool skip_unreported);
+
+  // Implements Process::DumpThreadPlansForTID
+  bool DumpPlansForTID(Stream &strm, lldb::tid_t tid,
+                       lldb::DescriptionLevel desc_level, bool internal,
+                       bool ignore_boring, bool skip_unreported);
+                       
+  bool PrunePlansForTID(lldb::tid_t tid);
+
 private:
+  Process &m_process;
   using PlansList = std::unordered_map<lldb::tid_t, ThreadPlanStack>;
   PlansList m_plans_list;
 };
 
 } // namespace lldb_private
 
 #endif // LLDB_TARGET_THREADPLANSTACK_H

lldb/source/Commands/CommandObjectThread.cpp

CommandOptions() : Options() {
       // ()
       OptionParsingStarting(nullptr);
     }
 
     ~CommandOptions() override = default;
 
     Status SetOptionValue(uint32_t option_idx, llvm::StringRef option_arg,
                           ExecutionContext *execution_context) override {
-      Status error;
       const int short_option = m_getopt_table[option_idx].val;
 
       switch (short_option) {
       case 'i':
         m_internal = true;
         break;
+      case 't':
+        lldb::tid_t tid;
+        if (option_arg.getAsInteger(0, tid))
+          return Status("invalid tid: '%s'.", option_arg.str().c_str());
+        m_tids.push_back(tid);
+        break;
+      case 'u':
+        m_unreported = false;
+        break;
       case 'v':
         m_verbose = true;
         break;
       default:
         llvm_unreachable("Unimplemented option");
       }
-      return error;
+      return {};
     }
 
     void OptionParsingStarting(ExecutionContext *execution_context) override {
       m_verbose = false;
       m_internal = false;
+      m_unreported = true; // The variable is "skip unreported" and we want to
+                           // skip unreported by default.
+      m_tids.clear();
     }
 
     llvm::ArrayRef<OptionDefinition> GetDefinitions() override {
       return llvm::makeArrayRef(g_thread_plan_list_options);
     }
 
     // Instance variables to hold the values for command options.
     bool m_verbose;
     bool m_internal;
+    bool m_unreported;
+    std::vector<lldb::tid_t> m_tids;
   };
 
   CommandObjectThreadPlanList(CommandInterpreter &interpreter)
       : CommandObjectIterateOverThreads(
             interpreter, "thread plan list",
             "Show thread plans for one or more threads.  If no threads are "
             "specified, show the "
             "current thread.  Use the thread-index \"all\" to see all threads.",
             nullptr,
             eCommandRequiresProcess | eCommandRequiresThread |
                 eCommandTryTargetAPILock | eCommandProcessMustBeLaunched |
                 eCommandProcessMustBePaused),
         m_options() {}
 
   ~CommandObjectThreadPlanList() override = default;
 
   Options *GetOptions() override { return &m_options; }
 
-protected:
-  bool HandleOneThread(lldb::tid_t tid, CommandReturnObject &result) override {
-    ThreadSP thread_sp =
-        m_exe_ctx.GetProcessPtr()->GetThreadList().FindThreadByID(tid);
-    if (!thread_sp) {
-      result.AppendErrorWithFormat("thread no longer exists: 0x%" PRIx64 "\n",
-                                   tid);
+  bool DoExecute(Args &command, CommandReturnObject &result) override {
+    // If we are reporting all threads, dispatch to the Process to do that:
+    if (command.GetArgumentCount() == 0 && m_options.m_tids.empty()) {
+      Stream &strm = result.GetOutputStream();
+      DescriptionLevel desc_level = m_options.m_verbose
+                                        ? eDescriptionLevelVerbose
+                                        : eDescriptionLevelFull;
+      m_exe_ctx.GetProcessPtr()->DumpThreadPlans(
+          strm, desc_level, m_options.m_internal, true, m_options.m_unreported);
+      result.SetStatus(eReturnStatusSuccessFinishResult);
+      return true;
+    } else {
+      // Do any TID's that the user may have specified as TID, then do any
+      // Thread Indexes...
+      if (!m_options.m_tids.empty()) {
+        Process *process = m_exe_ctx.GetProcessPtr();
+        StreamString tmp_strm;
+        for (lldb::tid_t tid : m_options.m_tids) {
+          bool success = process->DumpThreadPlansForTID(
+              tmp_strm, tid, eDescriptionLevelFull, m_options.m_internal,
+              true /* condense_trivial */, m_options.m_unreported);
+          // If we didn't find a TID, stop here and return an error.
+          if (!success) {
+            result.SetError("Error dumping plans:");
+            result.AppendError(tmp_strm.GetString());
             result.SetStatus(eReturnStatusFailed);
             return false;
           }
+          // Otherwise, add our data to the output:
+          result.GetOutputStream() << tmp_strm.GetString();
+        }
+      }
+      return CommandObjectIterateOverThreads::DoExecute(command, result);
+    }
+  }
 
-    Thread *thread = thread_sp.get();
+protected:
+  bool HandleOneThread(lldb::tid_t tid, CommandReturnObject &result) override {
+    // If we have already handled this from a -t option, skip it here.
+    if (std::find(m_options.m_tids.begin(), m_options.m_tids.end(), tid) !=
+        m_options.m_tids.end())
+      return true;
+
+    Process *process = m_exe_ctx.GetProcessPtr();
 
     Stream &strm = result.GetOutputStream();
     DescriptionLevel desc_level = eDescriptionLevelFull;
     if (m_options.m_verbose)
       desc_level = eDescriptionLevelVerbose;
 
-    thread->DumpThreadPlans(&strm, desc_level, m_options.m_internal, true);
+    process->DumpThreadPlansForTID(strm, tid, desc_level, m_options.m_internal,
+                                   true /* condense_trivial */,
+                                   m_options.m_unreported);
     return true;
   }
 
   CommandOptions m_options;
 };
 
 class CommandObjectThreadPlanDiscard : public CommandObjectParsed {
 public:
if (thread->DiscardUserThreadPlansUpToIndex(thread_plan_idx)) {
           "Could not find User thread plan with index %s.",
           args.GetArgumentAtIndex(0));
       result.SetStatus(eReturnStatusFailed);
       return false;
     }
   }
 };
 
+class CommandObjectThreadPlanPrune : public CommandObjectParsed {
+public:
+  CommandObjectThreadPlanPrune(CommandInterpreter &interpreter)
+      : CommandObjectParsed(interpreter, "thread plan prune",
+                            "Removes any thread plans associated with "
+                            "currently unreported threads.  "
+                            "Specify one or more TID's to remove, or if no "
+                            "TID's are provides, remove threads for all "
+                            "unreported threads",
+                            nullptr,
+                            eCommandRequiresProcess |
+                                eCommandTryTargetAPILock |
+                                eCommandProcessMustBeLaunched |
+                                eCommandProcessMustBePaused) {
+    CommandArgumentEntry arg;
+    CommandArgumentData tid_arg;
+
+    // Define the first (and only) variant of this arg.
+    tid_arg.arg_type = eArgTypeThreadID;
+    tid_arg.arg_repetition = eArgRepeatStar;
+
+    // There is only one variant this argument could be; put it into the
+    // argument entry.
+    arg.push_back(tid_arg);
+
+    // Push the data for the first argument into the m_arguments vector.
+    m_arguments.push_back(arg);
+  }
+
+  ~CommandObjectThreadPlanPrune() override = default;
+
+  bool DoExecute(Args &args, CommandReturnObject &result) override {
+    Process *process = m_exe_ctx.GetProcessPtr();
+    
+    if (args.GetArgumentCount() == 0) {
+      process->PruneThreadPlans();
+      result.SetStatus(eReturnStatusSuccessFinishNoResult);
+      return true;  
+    }
+
+    bool success;
+    const size_t num_args = args.GetArgumentCount();
+
+    std::lock_guard<std::recursive_mutex> guard(
+        process->GetThreadList().GetMutex());
+
+    for (size_t i = 0; i < num_args; i++) {
+      bool success;
+
+      lldb::tid_t tid = StringConvert::ToUInt64(
+          args.GetArgumentAtIndex(i), 0, 0, &success);
+      if (!success) {
+        result.AppendErrorWithFormat("invalid thread specification: \"%s\"\n",
+                                     args.GetArgumentAtIndex(i));
+        result.SetStatus(eReturnStatusFailed);
+        return false;
+      }
+      if (!process->PruneThreadPlansForTID(tid)) {
+        result.AppendErrorWithFormat("Could not find unreported tid: \"%s\"\n",
+                                     args.GetArgumentAtIndex(i));
+        result.SetStatus(eReturnStatusFailed);
+        return false;
+      }
+    }
+    result.SetStatus(eReturnStatusSuccessFinishNoResult);
+    return true;
+  }
+};
+
 // CommandObjectMultiwordThreadPlan
 
 class CommandObjectMultiwordThreadPlan : public CommandObjectMultiword {
 public:
   CommandObjectMultiwordThreadPlan(CommandInterpreter &interpreter)
       : CommandObjectMultiword(
             interpreter, "plan",
             "Commands for managing thread plans that control execution.",
             "thread plan <subcommand> [<subcommand objects]") {
     LoadSubCommand(
         "list", CommandObjectSP(new CommandObjectThreadPlanList(interpreter)));
     LoadSubCommand(
         "discard",
         CommandObjectSP(new CommandObjectThreadPlanDiscard(interpreter)));
+    LoadSubCommand(
+        "prune",
+        CommandObjectSP(new CommandObjectThreadPlanPrune(interpreter)));
   }
 
   ~CommandObjectMultiwordThreadPlan() override = default;
 };
 
 // CommandObjectMultiwordThread
 
 CommandObjectMultiwordThread::CommandObjectMultiwordThread(

lldb/source/Commands/Options.td

def thread_jump_force : Option<"force", "r">, Groups<[1,2,3]>,
     Desc<"Allows the PC to leave the current function.">;
 }
 
 let Command = "thread plan list" in {
   def thread_plan_list_verbose : Option<"verbose", "v">, Group<1>,
     Desc<"Display more information about the thread plans">;
   def thread_plan_list_internal : Option<"internal", "i">, Group<1>,
     Desc<"Display internal as well as user thread plans">;
+  def thread_plan_list_thread_id : Option<"thread-id", "t">, Group<1>,
+    Arg<"ThreadID">, Desc<"List the thread plans for this TID, can be "
+    "specified more than once.">;
+  def thread_plan_list_unreported : Option<"unreported", "u">, Group<1>,
+    Desc<"Display thread plans for unreported threads">;
 }
 
 let Command = "type summary add" in {
   def type_summary_add_category : Option<"category", "w">, Arg<"Name">,
     Desc<"Add this to the given category instead of the default one.">;
   def type_summary_add_cascade : Option<"cascade", "C">, Arg<"Boolean">,
     Desc<"If true, cascade through typedef chains.">;
   def type_summary_add_no_value : Option<"no-value", "v">,

lldb/source/Target/Process.cpp

: ProcessProperties(this), UserID(LLDB_INVALID_PROCESS_ID),
                                   "lldb.process.internal_state_broadcaster"),
       m_private_state_control_broadcaster(
           nullptr, "lldb.process.internal_state_control_broadcaster"),
       m_private_state_listener_sp(
           Listener::MakeListener("lldb.process.internal_state_listener")),
       m_mod_id(), m_process_unique_id(0), m_thread_index_id(0),
       m_thread_id_to_index_id_map(), m_exit_status(-1), m_exit_string(),
       m_exit_status_mutex(), m_thread_mutex(), m_thread_list_real(this),
-      m_thread_list(this), m_extended_thread_list(this),
+      m_thread_list(this), m_thread_plans(*this), m_extended_thread_list(this),
       m_extended_thread_stop_id(0), m_queue_list(this), m_queue_list_stop_id(0),
       m_notifications(), m_image_tokens(), m_listener_sp(listener_sp),
       m_breakpoint_site_list(), m_dynamic_checkers_up(),
       m_unix_signals_sp(unix_signals_sp), m_abi_sp(), m_process_input_reader(),
       m_stdio_communication("process.stdio"), m_stdio_communication_mutex(),
       m_stdin_forward(false), m_stdout_data(), m_stderr_data(),
       m_profile_data_comm_mutex(), m_profile_data(), m_iohandler_sync(0),
       m_memory_cache(*this), m_allocated_memory_cache(*this),
bool Process::SetProcessExitStatus(
   }
   return false;
 }
 
 void Process::UpdateThreadListIfNeeded() {
   const uint32_t stop_id = GetStopID();
   if (m_thread_list.GetSize(false) == 0 ||
       stop_id != m_thread_list.GetStopID()) {
+    bool clear_unused_threads = true;
     const StateType state = GetPrivateState();
     if (StateIsStoppedState(state, true)) {
       std::lock_guard<std::recursive_mutex> guard(m_thread_list.GetMutex());
+      m_thread_list.SetStopID(stop_id);
+
       // m_thread_list does have its own mutex, but we need to hold onto the
       // mutex between the call to UpdateThreadList(...) and the
       // os->UpdateThreadList(...) so it doesn't change on us
       ThreadList &old_thread_list = m_thread_list;
       ThreadList real_thread_list(this);
       ThreadList new_thread_list(this);
       // Always update the thread list with the protocol specific thread list,
       // but only update if "true" is returned
       if (UpdateThreadList(m_thread_list_real, real_thread_list)) {
         // Don't call into the OperatingSystem to update the thread list if we
         // are shutting down, since that may call back into the SBAPI's,
         // requiring the API lock which is already held by whoever is shutting
         // us down, causing a deadlock.
         OperatingSystem *os = GetOperatingSystem();
         if (os && !m_destroy_in_process) {
           // Clear any old backing threads where memory threads might have been
           // backed by actual threads from the lldb_private::Process subclass
           size_t num_old_threads = old_thread_list.GetSize(false);
           for (size_t i = 0; i < num_old_threads; ++i)
             old_thread_list.GetThreadAtIndex(i, false)->ClearBackingThread();
+          // See if the OS plugin reports all threads.  If it does, then
+          // it is safe to clear unseen thread's plans here.  Otherwise we 
+          // should preserve them in case they show up again:
+          clear_unused_threads = GetTarget().GetOSPluginReportsAllThreads();
 
           // Turn off dynamic types to ensure we don't run any expressions.
           // Objective-C can run an expression to determine if a SBValue is a
           // dynamic type or not and we need to avoid this. OperatingSystem
           // plug-ins can't run expressions that require running code...
 
           Target &target = GetTarget();
           const lldb::DynamicValueType saved_prefer_dynamic =
if (StateIsStoppedState(state, true)) {
                                 // subclass
               new_thread_list); // The new thread list that we will show to the
                                 // user that gets filled in
 
           if (saved_prefer_dynamic != lldb::eNoDynamicValues)
             target.SetPreferDynamicValue(saved_prefer_dynamic);
         } else {
           // No OS plug-in, the new thread list is the same as the real thread
-          // list
+          // list.
           new_thread_list = real_thread_list;
         }
 
         m_thread_list_real.Update(real_thread_list);
         m_thread_list.Update(new_thread_list);
         m_thread_list.SetStopID(stop_id);
 
         if (GetLastNaturalStopID() != m_extended_thread_stop_id) {
           // Clear any extended threads that we may have accumulated previously
           m_extended_thread_list.Clear();
           m_extended_thread_stop_id = GetLastNaturalStopID();
 
           m_queue_list.Clear();
           m_queue_list_stop_id = GetLastNaturalStopID();
         }
       }
+      // Now update the plan stack map.
+      // If we do have an OS plugin, any absent real threads in the
+      // m_thread_list have already been removed from the ThreadPlanStackMap.
+      // So any remaining threads are OS Plugin threads, and those we want to
+      // preserve in case they show up again.
+      m_thread_plans.Update(m_thread_list, clear_unused_threads);
     }
   }
 }
 
 ThreadPlanStack *Process::FindThreadPlans(lldb::tid_t tid) {
   return m_thread_plans.Find(tid);
 }
 
-void Process::AddThreadPlansForThread(Thread &thread) {
-  if (m_thread_plans.Find(thread.GetID()))
-    return;
-  m_thread_plans.AddThread(thread);
+bool Process::PruneThreadPlansForTID(lldb::tid_t tid) {
+  return m_thread_plans.PrunePlansForTID(tid);
+}
+
+void Process::PruneThreadPlans() {
+  m_thread_plans.Update(GetThreadList(), true, false);
 }
 
-void Process::RemoveThreadPlansForTID(lldb::tid_t tid) {
-  m_thread_plans.RemoveTID(tid);
+bool Process::DumpThreadPlansForTID(Stream &strm, lldb::tid_t tid,
+                                    lldb::DescriptionLevel desc_level,
+                                    bool internal, bool condense_trivial,
+                                    bool skip_unreported_plans) {
+  return m_thread_plans.DumpPlansForTID(
+      strm, tid, desc_level, internal, condense_trivial, skip_unreported_plans);
+}
+void Process::DumpThreadPlans(Stream &strm, lldb::DescriptionLevel desc_level,
+                              bool internal, bool condense_trivial,
+                              bool skip_unreported_plans) {
+  m_thread_plans.DumpPlans(strm, desc_level, internal, condense_trivial,
+                           skip_unreported_plans);
 }
 
 void Process::UpdateQueueListIfNeeded() {
   if (m_system_runtime_up) {
     if (m_queue_list.GetSize() == 0 ||
         m_queue_list_stop_id != GetLastNaturalStopID()) {
       const StateType state = GetPrivateState();
       if (StateIsStoppedState(state, true)) {

lldb/source/Target/Target.cpp

const Property *exp_property =
       m_collection_sp->GetPropertyAtIndex(exe_ctx, true, ePropertyExperimental);
   OptionValueProperties *exp_values =
       exp_property->GetValue()->GetAsProperties();
   if (exp_values)
     exp_values->SetPropertyAtIndexAsBoolean(exe_ctx, ePropertyInjectLocalVars,
                                             true);
 }
 
+bool TargetProperties::GetOSPluginReportsAllThreads() const {
+  const bool fail_value = true;
+  const Property *exp_property =
+      m_collection_sp->GetPropertyAtIndex(nullptr, true, ePropertyExperimental);
+  OptionValueProperties *exp_values =
+      exp_property->GetValue()->GetAsProperties();
+  if (!exp_values)
+    return fail_value;
+    
+  return 
+      exp_values->GetPropertyAtIndexAsBoolean(nullptr, 
+                                              ePropertyOSPluginReportsAllThreads,
+                                              fail_value);
+}
+
+void TargetProperties::SetOSPluginReportsAllThreads(bool does_report) {
+  const Property *exp_property =
+      m_collection_sp->GetPropertyAtIndex(nullptr, true, ePropertyExperimental);
+  OptionValueProperties *exp_values =
+      exp_property->GetValue()->GetAsProperties();
+  if (exp_values)
+    exp_values->SetPropertyAtIndexAsBoolean(nullptr, 
+                                            ePropertyOSPluginReportsAllThreads,
+                                            does_report);
+}
+
+
+
 ArchSpec TargetProperties::GetDefaultArchitecture() const {
   OptionValueArch *value = m_collection_sp->GetPropertyAtIndexAsOptionValueArch(
       nullptr, ePropertyDefaultArch);
   if (value)
     return value->GetCurrentValue();
   return ArchSpec();
 }
 

lldb/source/Target/TargetProperties.td

 include "../../include/lldb/Core/PropertiesBase.td"
 
 let Definition = "experimental" in {
   def InjectLocalVars : Property<"inject-local-vars", "Boolean">,
     Global, DefaultTrue,
     Desc<"If true, inject local variables explicitly into the expression text. This will fix symbol resolution when there are name collisions between ivars and local variables. But it can make expressions run much more slowly.">;
+  def OSPluginReportsAllThreads: Property<"os-plugin-reports-all-threads", "Boolean">,
+    Global,
+    DefaultTrue,
+    Desc<"Set to False if your OS Plugins doesn't report all threads on each stop.">;
 }
 
 let Definition = "target" in {
   def DefaultArch: Property<"default-arch", "Arch">,
     Global,
     DefaultStringValue<"">,
     Desc<"Default architecture to choose, when there's a choice.">;
   def MoveToNearestCode: Property<"move-to-nearest-code", "Boolean">,
def IgnoreBreakpointsInExpressions: Property<"ignore-breakpoints-in-expressions", "Boolean">,
     Desc<"If true, breakpoints will be ignored during expression evaluation.">;
   def UnwindOnErrorInExpressions: Property<"unwind-on-error-in-expressions", "Boolean">,
     Global,
     DefaultTrue,
     Desc<"If true, errors in expression evaluation will unwind the stack back to the state before the call.">;
   def PythonOSPluginPath: Property<"python-os-plugin-path", "FileSpec">,
     DefaultUnsignedValue<1>,
     Desc<"A path to a python OS plug-in module file that contains a OperatingSystemPlugIn class.">;
   def StopOnSharedLibraryEvents: Property<"stop-on-sharedlibrary-events", "Boolean">,
     Global,
     DefaultFalse,
     Desc<"If true, stop when a shared library is loaded or unloaded.">;
   def DetachKeepsStopped: Property<"detach-keeps-stopped", "Boolean">,
     Global,
     DefaultFalse,
     Desc<"If true, detach will attempt to keep the process stopped.">;
   def MemCacheLineSize: Property<"memory-cache-line-size", "UInt64">,

lldb/source/Target/Thread.cpp

: ThreadProperties(false), UserID(tid),
       m_unwinder_up(), m_destroy_called(false),
       m_override_should_notify(eLazyBoolCalculate),
       m_extended_info_fetched(false), m_extended_info() {
   Log *log(lldb_private::GetLogIfAllCategoriesSet(LIBLLDB_LOG_OBJECT));
   LLDB_LOGF(log, "%p Thread::Thread(tid = 0x%4.4" PRIx64 ")",
             static_cast<void *>(this), GetID());
 
   CheckInWithManager();
-  
-  process.AddThreadPlansForThread(*this);
-  
-  QueueFundamentalPlan(true);
 }
 
 Thread::~Thread() {
   Log *log(lldb_private::GetLogIfAllCategoriesSet(LIBLLDB_LOG_OBJECT));
   LLDB_LOGF(log, "%p Thread::~Thread(tid = 0x%4.4" PRIx64 ")",
             static_cast<void *>(this), GetID());
   /// If you hit this assert, it means your derived class forgot to call
   /// DoDestroy in its destructor.
LLDB_LOGF(log,
               "Thread::%s(%p) for tid = 0x%4.4" PRIx64 " 0x%4.4" PRIx64
               ", pc = 0x%16.16" PRIx64,
               __FUNCTION__, static_cast<void *>(this), GetID(), GetProtocolID(),
               GetRegisterContext() ? GetRegisterContext()->GetPC()
                                    : LLDB_INVALID_ADDRESS);
     LLDB_LOGF(log, "^^^^^^^^ Thread::ShouldStop Begin ^^^^^^^^");
     StreamString s;
     s.IndentMore();
-    DumpThreadPlans(&s);
+    GetProcess()->DumpThreadPlansForTID(
+        s, GetID(), eDescriptionLevelVerbose, true /* internal */,
+        false /* condense_trivial */, true /* skip_unreported */);
     LLDB_LOGF(log, "Plan stack initial state:\n%s", s.GetData());
   }
 
   // The top most plan always gets to do the trace log...
   current_plan->DoTraceLog();
 
   // First query the stop info's ShouldStopSynchronous.  This handles
   // "synchronous" stop reasons, for example the breakpoint command on internal
while (!plan_ptr->IsBasePlan()) {
           DiscardPlan();
       }
     }
   }
 
   if (log) {
     StreamString s;
     s.IndentMore();
-    DumpThreadPlans(&s);
+    GetProcess()->DumpThreadPlansForTID(
+        s, GetID(), eDescriptionLevelVerbose, true /* internal */,
+        false /* condense_trivial */, true /* skip_unreported */);
     LLDB_LOGF(log, "Plan stack final state:\n%s", s.GetData());
     LLDB_LOGF(log, "vvvvvvvv Thread::ShouldStop End (returning %i) vvvvvvvv",
               should_stop);
   }
   return should_stop;
 }
 
 Vote Thread::ShouldReportStop(Event *event_ptr) {

 }
 
 bool Thread::MatchesSpec(const ThreadSpec *spec) {
   return (spec == nullptr) ? true : spec->ThreadPassesBasicTests(*this);
 }
 
 ThreadPlanStack &Thread::GetPlans() const {
   ThreadPlanStack *plans = GetProcess()->FindThreadPlans(GetID());
-  assert(plans && "Can't have a thread with no plans");
+  if (plans)
     return *plans;
+
+  // History threads don't have a thread plan, but they do ask get asked to
+  // describe themselves, which usually involves pulling out the stop reason.
+  // That in turn will check for a completed plan on the ThreadPlanStack.
+  // Instead of special-casing at that point, we return a Stack with a
+  // ThreadPlanNull as its base plan.  That will give the right answers to the
+  // queries GetDescription makes, and only assert if you try to run the thread.
+  if (!m_null_plan_stack_up)
+    m_null_plan_stack_up.reset(new ThreadPlanStack(*this, true));
+  return *(m_null_plan_stack_up.get());
 }
 
 void Thread::PushPlan(ThreadPlanSP thread_plan_sp) {
   assert(thread_plan_sp && "Don't push an empty thread plan.");
 
   Log *log(lldb_private::GetLogIfAllCategoriesSet(LIBLLDB_LOG_STEP));
   if (log) {
     StreamString s;
ThreadPlanSP thread_plan_sp(new ThreadPlanPython(*this, class_name,
                                                    extra_args_impl));
 
   status = QueueThreadPlan(thread_plan_sp, abort_other_plans);
   return thread_plan_sp;
 }
 
 uint32_t Thread::GetIndexID() const { return m_index_id; }
 
-void Thread::DumpThreadPlans(Stream *s, lldb::DescriptionLevel desc_level,
-                             bool include_internal,
-                             bool ignore_boring_threads) const {
-  if (ignore_boring_threads) {
-    if (!GetPlans().AnyPlans() && !GetPlans().AnyCompletedPlans()
-        && !GetPlans().AnyDiscardedPlans()) {
-      s->Printf("thread #%u: tid = 0x%4.4" PRIx64 "\n", GetIndexID(), GetID());
-      s->IndentMore();
-      s->Indent();
-      s->Printf("No active thread plans\n");
-      s->IndentLess();
-      return;
-    }
-  }
-  
-  s->Indent();
-  s->Printf("thread #%u: tid = 0x%4.4" PRIx64 ":\n", GetIndexID(), GetID());
-  GetPlans().DumpThreadPlans(s, desc_level, include_internal);
-}
-
 TargetSP Thread::CalculateTarget() {
   TargetSP target_sp;
   ProcessSP process_sp(GetProcess());
   if (process_sp)
     target_sp = process_sp->CalculateTarget();
   return target_sp;
 }
 

lldb/source/Target/ThreadList.cpp

if (this != &rhs) {
     // Now we look for threads that we are done with and make sure to clear
     // them up as much as possible so anyone with a shared pointer will still
     // have a reference, but the thread won't be of much use. Using
     // std::weak_ptr for all backward references (such as a thread to a
     // process) will eventually solve this issue for us, but for now, we need
     // to work around the issue
     collection::iterator rhs_pos, rhs_end = rhs.m_threads.end();
     for (rhs_pos = rhs.m_threads.begin(); rhs_pos != rhs_end; ++rhs_pos) {
+      // If this thread has already been destroyed, we don't need to look for
+      // it to destroy it again.
+      if (!(*rhs_pos)->IsValid())
+        continue;
+
       const lldb::tid_t tid = (*rhs_pos)->GetID();
       bool thread_is_alive = false;
       const uint32_t num_threads = m_threads.size();
       for (uint32_t idx = 0; idx < num_threads; ++idx) {
         ThreadSP backing_thread = m_threads[idx]->GetBackingThread();
         if (m_threads[idx]->GetID() == tid ||
             (backing_thread && backing_thread->GetID() == tid)) {
           thread_is_alive = true;
           break;
         }
       }
       if (!thread_is_alive) {
         (*rhs_pos)->DestroyThread();
-        m_process->RemoveThreadPlansForTID((*rhs_pos)->GetID());
       }
     }
   }
 }
 
 void ThreadList::Flush() {
   std::lock_guard<std::recursive_mutex> guard(GetMutex());
   collection::iterator pos, end = m_threads.end();

lldb/source/Target/ThreadPlan.cpp

 #include "lldb/Utility/State.h"
 
 using namespace lldb;
 using namespace lldb_private;
 
 // ThreadPlan constructor
 ThreadPlan::ThreadPlan(ThreadPlanKind kind, const char *name, Thread &thread,
                        Vote stop_vote, Vote run_vote)
-    : m_process(*thread.GetProcess().get()), m_tid(thread.GetID()), 
+    : m_process(*thread.GetProcess().get()), m_tid(thread.GetID()),
       m_stop_vote(stop_vote), m_run_vote(run_vote),
       m_takes_iteration_count(false), m_could_not_resolve_hw_bp(false),
       m_kind(kind), m_thread(&thread), m_name(name), m_plan_complete_mutex(),
       m_cached_plan_explains_stop(eLazyBoolCalculate), m_plan_complete(false),
       m_plan_private(false), m_okay_to_discard(true), m_is_master_plan(false),
       m_plan_succeeded(true) {
   SetID(GetNextID());
 }
 
 // Destructor
 ThreadPlan::~ThreadPlan() = default;
 
 Target &ThreadPlan::GetTarget() { return m_process.GetTarget(); }
 
 const Target &ThreadPlan::GetTarget() const { return m_process.GetTarget(); }
 
 Thread &ThreadPlan::GetThread() {
   if (m_thread)
     return *m_thread;
-    
   ThreadSP thread_sp = m_process.GetThreadList().FindThreadByID(m_tid);
   m_thread = thread_sp.get();
   return *m_thread;
 }
 
 bool ThreadPlan::PlanExplainsStop(Event *event_ptr) {
   if (m_cached_plan_explains_stop == eLazyBoolCalculate) {
     bool actual_value = DoPlanExplainsStop(event_ptr);
if (log) {
       addr_t pc = reg_ctx->GetPC();
       addr_t sp = reg_ctx->GetSP();
       addr_t fp = reg_ctx->GetFP();
       LLDB_LOGF(
           log,
           "%s Thread #%u (0x%p): tid = 0x%4.4" PRIx64 ", pc = 0x%8.8" PRIx64
           ", sp = 0x%8.8" PRIx64 ", fp = 0x%8.8" PRIx64 ", "
           "plan = '%s', state = %s, stop others = %d",
-          __FUNCTION__, GetThread().GetIndexID(), 
+          __FUNCTION__, GetThread().GetIndexID(),
           static_cast<void *>(&GetThread()), m_tid, static_cast<uint64_t>(pc),
           static_cast<uint64_t>(sp), static_cast<uint64_t>(fp), m_name.c_str(),
           StateAsCString(resume_state), StopOthers());
     }
   }
-  return DoWillResume(resume_state, current_plan);
+  bool success = DoWillResume(resume_state, current_plan);
+  m_thread = nullptr; // We don't cache the thread pointer over resumes.  This
+                      // Thread might go away, and another Thread represent
+                      // the same underlying object on a later stop.
+  return success;
 }
 
 lldb::user_id_t ThreadPlan::GetNextID() {
   static uint32_t g_nextPlanID = 0;
   return ++g_nextPlanID;
 }
 
 void ThreadPlan::DidPush() {}

lldb/source/Target/ThreadPlanStack.cpp

 #include "lldb/Target/Target.h"
 #include "lldb/Target/Thread.h"
 #include "lldb/Target/ThreadPlan.h"
 #include "lldb/Utility/Log.h"
 
 using namespace lldb;
 using namespace lldb_private;
 
-static void PrintPlanElement(Stream *s, const ThreadPlanSP &plan,
+static void PrintPlanElement(Stream &s, const ThreadPlanSP &plan,
                              lldb::DescriptionLevel desc_level,
                              int32_t elem_idx) {
-  s->IndentMore();
-  s->Indent();
-  s->Printf("Element %d: ", elem_idx);
-  plan->GetDescription(s, desc_level);
-  s->EOL();
-  s->IndentLess();
+  s.IndentMore();
+  s.Indent();
+  s.Printf("Element %d: ", elem_idx);
+  plan->GetDescription(&s, desc_level);
+  s.EOL();
+  s.IndentLess();
 }
 
-void ThreadPlanStack::DumpThreadPlans(Stream *s,
+ThreadPlanStack::ThreadPlanStack(const Thread &thread, bool make_null) {
+  if (make_null) {
+    // The ThreadPlanNull doesn't do anything to the Thread, so this is actually
+    // still a const operation.
+    m_plans.push_back(
+        ThreadPlanSP(new ThreadPlanNull(const_cast<Thread &>(thread))));
+  }
+}
+
+void ThreadPlanStack::DumpThreadPlans(Stream &s,
                                       lldb::DescriptionLevel desc_level,
                                       bool include_internal) const {
 
   uint32_t stack_size;
 
-  s->IndentMore();
-  s->Indent();
-  s->Printf("Active plan stack:\n");
-  int32_t print_idx = 0;
-  for (auto plan : m_plans) {
-    PrintPlanElement(s, plan, desc_level, print_idx++);
-  }
+  s.IndentMore();
+  PrintOneStack(s, "Active plan stack", m_plans, desc_level, include_internal);
+  PrintOneStack(s, "Completed plan stack", m_completed_plans, desc_level,
+                include_internal);
+  PrintOneStack(s, "Discarded plan stack", m_discarded_plans, desc_level,
+                include_internal);
+  s.IndentLess();
 
-  if (AnyCompletedPlans()) {
-    print_idx = 0;
-    s->Indent();
-    s->Printf("Completed Plan Stack:\n");
-    for (auto plan : m_completed_plans)
-      PrintPlanElement(s, plan, desc_level, print_idx++);
+void ThreadPlanStack::PrintOneStack(Stream &s, llvm::StringRef stack_name,
+                                    const PlanStack &stack,
+                                    lldb::DescriptionLevel desc_level,
+                                    bool include_internal) const {
+  // If the stack is empty, just exit:
+  if (stack.empty())
+    return;
+
+  // Make sure there are public completed plans:
+  bool any_public = false;
+  if (!include_internal) {
+    for (auto plan : stack) {
+      if (!plan->GetPrivate()) {
+        any_public = true;
+        break;
+      }
+    }
   }
 
-  if (AnyDiscardedPlans()) {
-    print_idx = 0;
-    s->Indent();
-    s->Printf("Discarded Plan Stack:\n");
-    for (auto plan : m_discarded_plans)
+  if (include_internal || any_public) {
+    int print_idx = 0;
+    s.Indent();
+    s.Printf("%s:\n", stack_name);
+    for (auto plan : stack) {
+      if (!include_internal && plan->GetPrivate())
+        continue;
       PrintPlanElement(s, plan, desc_level, print_idx++);
     }
-
-  s->IndentLess();
+  }
 }
 
 size_t ThreadPlanStack::CheckpointCompletedPlans() {
   m_completed_plan_checkpoint++;
   m_completed_plan_store.insert(
       std::make_pair(m_completed_plan_checkpoint, m_completed_plans));
   return m_completed_plan_checkpoint;
 }
case ePlans:
     return m_plans;
   case eCompletedPlans:
     return m_completed_plans;
   case eDiscardedPlans:
     return m_discarded_plans;
   }
   llvm_unreachable("Invalid StackKind value");
 }
+
+void ThreadPlanStackMap::Update(ThreadList &current_threads,
+                                bool delete_missing,
+                                bool check_for_new) {
+
+  // Now find all the new threads and add them to the map:
+  if (check_for_new) {
+    for (auto thread : current_threads.Threads()) {
+      lldb::tid_t cur_tid = thread->GetID();
+      if (!Find(cur_tid)) {
+        AddThread(*thread.get());
+        thread->QueueFundamentalPlan(true);
+      }
+    }
+  }
+
+  // If we aren't reaping missing threads at this point,
+  // we are done.
+  if (!delete_missing)
+    return;
+  // Otherwise scan for absent TID's.
+  std::vector<lldb::tid_t> missing_threads;
+  // If we are going to delete plans from the plan stack,
+  // then scan for absent TID's:
+  for (auto thread_plans : m_plans_list) {
+    lldb::tid_t cur_tid = thread_plans.first;
+    ThreadSP thread_sp = current_threads.FindThreadByID(cur_tid);
+    if (!thread_sp)
+      missing_threads.push_back(cur_tid);
+  }
+  for (lldb::tid_t tid : missing_threads) {
+    RemoveTID(tid);
+  }
+}
+
+void ThreadPlanStackMap::DumpPlans(Stream &strm,
+                                   lldb::DescriptionLevel desc_level,
+                                   bool internal, bool condense_if_trivial,
+                                   bool skip_unreported) {
+  for (auto elem : m_plans_list) {
+    lldb::tid_t tid = elem.first;
+    uint32_t index_id = 0;
+    ThreadSP thread_sp = m_process.GetThreadList().FindThreadByID(tid);
+
+    if (skip_unreported) {
+      if (!thread_sp)
+        continue;
+    }
+    if (thread_sp)
+      index_id = thread_sp->GetIndexID();
+
+    if (condense_if_trivial) {
+      if (!elem.second.AnyPlans() && !elem.second.AnyCompletedPlans() &&
+          !elem.second.AnyDiscardedPlans()) {
+        strm.Printf("thread #%u: tid = 0x%4.4" PRIx64 "\n", index_id, tid);
+        strm.IndentMore();
+        strm.Indent();
+        strm.Printf("No active thread plans\n");
+        strm.IndentLess();
+        return;
+      }
+    }
+
+    strm.Indent();
+    strm.Printf("thread #%u: tid = 0x%4.4" PRIx64 ":\n", index_id, tid);
+
+    elem.second.DumpThreadPlans(strm, desc_level, internal);
+  }
+}
+
+bool ThreadPlanStackMap::DumpPlansForTID(Stream &strm, lldb::tid_t tid,
+                                         lldb::DescriptionLevel desc_level,
+                                         bool internal,
+                                         bool condense_if_trivial,
+                                         bool skip_unreported) {
+  uint32_t index_id = 0;
+  ThreadSP thread_sp = m_process.GetThreadList().FindThreadByID(tid);
+
+  if (skip_unreported) {
+    if (!thread_sp) {
+      strm.Format("Unknown TID: {0}", tid);
+      return false;
+    }
+  }
+
+  if (thread_sp)
+    index_id = thread_sp->GetIndexID();
+  ThreadPlanStack *stack = Find(tid);
+  if (!stack) {
+    strm.Format("Unknown TID: {0}\n", tid);
+    return false;
+  }
+
+  if (condense_if_trivial) {
+    if (!stack->AnyPlans() && !stack->AnyCompletedPlans() &&
+        !stack->AnyDiscardedPlans()) {
+      strm.Printf("thread #%u: tid = 0x%4.4" PRIx64 "\n", index_id, tid);
+      strm.IndentMore();
+      strm.Indent();
+      strm.Printf("No active thread plans\n");
+      strm.IndentLess();
+      return true;
+    }
+  }
+
+  strm.Indent();
+  strm.Printf("thread #%u: tid = 0x%4.4" PRIx64 ":\n", index_id, tid);
+
+  stack->DumpThreadPlans(strm, desc_level, internal);
+  return true;
+}
+
+bool ThreadPlanStackMap::PrunePlansForTID(lldb::tid_t tid) {
+  // We only remove the plans for unreported TID's.
+  ThreadSP thread_sp = m_process.GetThreadList().FindThreadByID(tid);
+  if (thread_sp)
+    return false;
+
+  return RemoveTID(tid);
+}

lldb/source/Target/ThreadPlanStepOut.cpp

void ThreadPlanStepOut::DidPush() {
   if (m_step_out_to_inline_plan_sp)
     thread.QueueThreadPlan(m_step_out_to_inline_plan_sp, false);
   else if (m_step_through_inline_plan_sp)
     thread.QueueThreadPlan(m_step_through_inline_plan_sp, false);
 }
 
 ThreadPlanStepOut::~ThreadPlanStepOut() {
   if (m_return_bp_id != LLDB_INVALID_BREAK_ID)
-    GetThread().CalculateTarget()->RemoveBreakpointByID(m_return_bp_id);
+    GetTarget().RemoveBreakpointByID(m_return_bp_id);
 }
 
 void ThreadPlanStepOut::GetDescription(Stream *s,
                                        lldb::DescriptionLevel level) {
   if (level == lldb::eDescriptionLevelBrief)
     s->Printf("step out");
   else {
     if (m_step_out_to_inline_plan_sp)
       s->Printf("Stepping out to inlined frame so we can walk through it.");
     else if (m_step_through_inline_plan_sp)
       s->Printf("Stepping out by stepping through inlined function.");
     else {
       s->Printf("Stepping out from ");
       Address tmp_address;
       if (tmp_address.SetLoadAddress(m_step_from_insn, &GetTarget())) {
-        tmp_address.Dump(s, &GetThread(), Address::DumpStyleResolvedDescription,
+        tmp_address.Dump(s, &m_process, Address::DumpStyleResolvedDescription,
                          Address::DumpStyleLoadAddress);
       } else {
         s->Printf("address 0x%" PRIx64 "", (uint64_t)m_step_from_insn);
       }
 
       // FIXME: find some useful way to present the m_return_id, since there may
       // be multiple copies of the
       // same function on the stack.
 
       s->Printf(" returning to frame at ");
       if (tmp_address.SetLoadAddress(m_return_addr, &GetTarget())) {
-        tmp_address.Dump(s, &GetThread(), Address::DumpStyleResolvedDescription,
+        tmp_address.Dump(s, &m_process, Address::DumpStyleResolvedDescription,
                          Address::DumpStyleLoadAddress);
       } else {
         s->Printf("address 0x%" PRIx64 "", (uint64_t)m_return_addr);
       }
 
       if (level == eDescriptionLevelVerbose)
         s->Printf(" using breakpoint site %d", m_return_bp_id);
     }
   }
 
+  if (m_stepped_past_frames.empty())
+    return;
+
   s->Printf("\n");
   for (StackFrameSP frame_sp : m_stepped_past_frames) {
     s->Printf("Stepped out past: ");
     frame_sp->DumpUsingSettingsFormat(s);
   }
 }
 
 bool ThreadPlanStepOut::ValidatePlan(Stream *error) {

lldb/test/API/functionalities/plugins/python_os_plugin/stepping_plugin_threads/Makefile

+CXX_SOURCES := main.cpp
+ENABLE_THREADS := YES
+
+include Makefile.rules

lldb/test/API/functionalities/plugins/python_os_plugin/stepping_plugin_threads/TestOSPluginStepping.py

+"""
+Test that stepping works even when the OS Plugin doesn't report
+all threads at every stop.
+"""
+
+from __future__ import print_function
+
+
+import os
+import lldb
+from lldbsuite.test.lldbtest import *
+import lldbsuite.test.lldbutil as lldbutil
+
+
+class TestOSPluginStepping(TestBase):
+
+    mydir = TestBase.compute_mydir(__file__)
+    NO_DEBUG_INFO_TESTCASE = True
+
+    def test_python_os_plugin(self):
+        """Test that stepping works when the OS Plugin doesn't report all
+           threads at every stop"""
+        self.build()
+        self.main_file = lldb.SBFileSpec('main.cpp')
+        self.run_python_os_step_missing_thread(False)
+
+    def test_python_os_plugin_prune(self):
+        """Test that pruning the unreported PlanStacks works"""
+        self.build()
+        self.main_file = lldb.SBFileSpec('main.cpp')
+        self.run_python_os_step_missing_thread(True)
+
+    def get_os_thread(self):
+        return self.process.GetThreadByID(0x111111111)
+
+    def is_os_thread(self, thread):
+        id = thread.GetID()
+        return id == 0x111111111
+    
+    def run_python_os_step_missing_thread(self, do_prune):
+        """Test that the Python operating system plugin works correctly"""
+
+        # Our OS plugin does NOT report all threads:
+        result = self.dbg.HandleCommand("settings set target.experimental.os-plugin-reports-all-threads false")
+
+        python_os_plugin_path = os.path.join(self.getSourceDir(),
+                                             "operating_system.py")
+        (target, self.process, thread, thread_bkpt) = lldbutil.run_to_source_breakpoint(
+            self, "first stop in thread - do a step out", self.main_file)
+
+        main_bkpt = target.BreakpointCreateBySourceRegex('Stop here and do not make a memory thread for thread_1',
+                                                         self.main_file)
+        self.assertEqual(main_bkpt.GetNumLocations(), 1, "Main breakpoint has one location")
+
+        # There should not be an os thread before we load the plugin:
+        self.assertFalse(self.get_os_thread().IsValid(), "No OS thread before loading plugin")
+        
+        # Now load the python OS plug-in which should update the thread list and we should have
+        # an OS plug-in thread overlaying thread_1 with id 0x111111111
+        command = "settings set target.process.python-os-plugin-path '%s'" % python_os_plugin_path
+        self.dbg.HandleCommand(command)
+
+        # Verify our OS plug-in threads showed up
+        os_thread = self.get_os_thread()
+        self.assertTrue(
+            os_thread.IsValid(),
+            "Make sure we added the thread 0x111111111 after we load the python OS plug-in")
+        
+        # Now we are going to step-out.  This should get interrupted by main_bkpt.  We've
+        # set up the OS plugin so at this stop, we have lost the OS thread 0x111111111.
+        # Make sure both of these are true:
+        os_thread.StepOut()
+        
+        stopped_threads = lldbutil.get_threads_stopped_at_breakpoint(self.process, main_bkpt)
+        self.assertEqual(len(stopped_threads), 1, "Stopped at main_bkpt")
+        thread = self.process.GetThreadByID(0x111111111)
+        self.assertFalse(thread.IsValid(), "No thread 0x111111111 on second stop.")
+        
+        # Make sure we still have the thread plans for this thread:
+        # First, don't show unreported threads, that should fail:
+        command = "thread plan list -t 0x111111111"
+        result = lldb.SBCommandReturnObject()
+        interp = self.dbg.GetCommandInterpreter() 
+        interp.HandleCommand(command, result)
+        self.assertFalse(result.Succeeded(), "We found no plans for the unreported thread.")
+        # Now do it again but with the -u flag:
+        command = "thread plan list -u -t 0x111111111"
+        result = lldb.SBCommandReturnObject()
+        interp.HandleCommand(command, result)
+        self.assertTrue(result.Succeeded(), "We found plans for the unreported thread.")
+        
+        if do_prune:
+            # Prune the thread plan and continue, and we will run to exit.
+            interp.HandleCommand("thread plan prune 0x111111111", result)
+            self.assertTrue(result.Succeeded(), "Found the plan for 0x111111111 and pruned it")
+
+            # List again, make sure it doesn't work:
+            command = "thread plan list -u -t 0x111111111"
+            interp.HandleCommand(command, result)
+            self.assertFalse(result.Succeeded(), "We still found plans for the unreported thread.")
+            
+            self.process.Continue()
+            self.assertEqual(self.process.GetState(), lldb.eStateExited, "We exited.")
+        else:
+            # Now we are going to continue, and when we hit the step-out breakpoint, we will
+            # put the OS plugin thread back, lldb will recover its ThreadPlanStack, and
+            # we will stop with a "step-out" reason.
+            self.process.Continue()
+            os_thread = self.get_os_thread()
+            self.assertTrue(os_thread.IsValid(), "The OS thread is back after continue")
+            self.assertTrue("step out" in os_thread.GetStopDescription(100), "Completed step out plan")
+        
+        

lldb/test/API/functionalities/plugins/python_os_plugin/stepping_plugin_threads/main.cpp

+// This test will present lldb with two threads one of which the test will
+// overlay with an OSPlugin thread.  Then we'll do a step out on the thread_1,
+// but arrange to hit a breakpoint in main before the step out completes. At
+// that point we will not report an OS plugin thread for thread_1. Then we'll
+// run again and hit the step out breakpoint.  Make sure we haven't deleted
+// that, and recognize it.
+
+#include <condition_variable>
+#include <mutex>
+#include <stdio.h>
+#include <thread>
+
+static int g_value = 0; // I don't have access to the real threads in the
+                        // OS Plugin, and I don't want to have to count
+                        // StopID's. So I'm using this value to tell me which
+                        // stop point the program has reached.
+std::mutex g_mutex;
+std::condition_variable g_cv;
+static int g_condition = 0; // Using this as the variable backing g_cv
+                            // to prevent spurious wakeups.
+
+void step_out_of_here() {
+  std::unique_lock<std::mutex> func_lock(g_mutex);
+  // Set a breakpoint:first stop in thread - do a step out.
+  g_condition = 1;
+  g_cv.notify_one();
+  g_cv.wait(func_lock, [&] { return g_condition == 2; });
+}
+
+void *thread_func() {
+  // Do something
+  step_out_of_here();
+
+  // Return
+  return NULL;
+}
+
+int main() {
+  // Lock the mutex so we can block the thread:
+  std::unique_lock<std::mutex> main_lock(g_mutex);
+  // Create the thread
+  std::thread thread_1(thread_func);
+  g_cv.wait(main_lock, [&] { return g_condition == 1; });
+  g_value = 1;
+  g_condition = 2;
+  // Stop here and do not make a memory thread for thread_1.
+  g_cv.notify_one();
+  g_value = 2;
+  main_lock.unlock();
+
+  // Wait for the threads to finish
+  thread_1.join();
+
+  return 0;
+}

lldb/test/API/functionalities/plugins/python_os_plugin/stepping_plugin_threads/operating_system.py

+#!/usr/bin/python
+
+import lldb
+import struct
+
+
+class OperatingSystemPlugIn(object):
+    """Class that provides a OS plugin that along with the particular code in main.cpp
+       emulates the following scenario:
+             a) We stop in an OS Plugin created thread - which should be thread index 1
+             b) We step-out from that thread
+             c) We hit a breakpoint in another thread, and DON'T produce the OS Plugin thread.
+             d) We continue, and when we hit the step out breakpoint, we again produce the same
+                OS Plugin thread.
+             main.cpp sets values into the global variable g_value, which we use to tell the OS
+             plugin whether to produce the OS plugin thread or not.
+             Since we are always producing an OS plugin thread with a backing thread, we don't
+             need to implement get_register_info or get_register_data.
+    """
+
+    def __init__(self, process):
+        '''Initialization needs a valid.SBProcess object.
+
+        This plug-in will get created after a live process is valid and has stopped for the
+        first time.'''
+        print("Plugin initialized.")
+        self.process = None
+        self.start_stop_id = 0
+        self.g_value = lldb.SBValue()
+        
+        if isinstance(process, lldb.SBProcess) and process.IsValid():
+            self.process = process
+            self.g_value = process.GetTarget().FindFirstGlobalVariable("g_value")
+            if not self.g_value.IsValid():
+                print("Could not find g_value")
+            
+    def create_thread(self, tid, context):
+        print("Called create thread with tid: ", tid)
+        return None
+
+    def get_thread_info(self):
+        g_value = self.g_value.GetValueAsUnsigned()
+        print("Called get_thread_info: g_value: %d"%(g_value))
+        if g_value == 0 or g_value == 2:
+            return [{'tid': 0x111111111,
+                             'name': 'one',
+                             'queue': 'queue1',
+                             'state': 'stopped',
+                             'stop_reason': 'breakpoint',
+                             'core' : 1 }]
+        else:
+            return []
+
+    def get_register_info(self):
+        print ("called get_register_info")
+        return None
+
+    
+    def get_register_data(self, tid):
+        print("Get register data called for tid: %d"%(tid))
+        return None
+

lldb/test/API/functionalities/thread_plan/Makefile

+C_SOURCES := main.c
+CFLAGS_EXTRAS := -std=c99
+
+include Makefile.rules

lldb/test/API/functionalities/thread_plan/TestThreadPlanCommands.py

+"""
+Test that thread plan listing, and deleting works.
+"""
+
+
+
+import lldb
+import lldbsuite.test.lldbutil as lldbutil
+from lldbsuite.test.lldbtest import *
+
+
+class TestThreadPlanCommands(TestBase):
+
+    mydir = TestBase.compute_mydir(__file__)
+
+    NO_DEBUG_INFO_TESTCASE = True
+
+    def test_thread_plan_actions(self):
+        self.build()
+        self.main_source_file = lldb.SBFileSpec("main.c")
+        self.thread_plan_test()
+
+    def check_list_output(self, command, active_plans = [], completed_plans = [], discarded_plans = []):
+        # Check the "thread plan list" output against a list of active & completed and discarded plans.
+        # If all three check arrays are empty, that means the command is expected to fail. 
+
+        interp = self.dbg.GetCommandInterpreter()
+        result = lldb.SBCommandReturnObject()
+
+        num_active = len(active_plans)
+        num_completed = len(completed_plans)
+        num_discarded = len(discarded_plans)
+
+        interp.HandleCommand(command, result)
+        if self.TraceOn():
+            print("Command: %s"%(command))
+            print(result.GetOutput())
+
+        if num_active == 0 and num_completed == 0 and num_discarded == 0:
+            self.assertFalse(result.Succeeded(), "command: '%s' succeeded when it should have failed: '%s'"%
+                             (command, result.GetError()))
+            return
+
+        self.assertTrue(result.Succeeded(), "command: '%s' failed: '%s'"%(command, result.GetError()))
+        result_arr = result.GetOutput().splitlines()
+        num_results = len(result_arr)
+
+        # Match the expected number of elements.
+        # Adjust the count for the number of header lines we aren't matching:
+        fudge = 0
+        
+        if num_completed == 0 and num_discarded == 0:
+            # The fudge is 3: Thread header, Active Plan header and base plan
+            fudge = 3
+        elif num_completed == 0 or num_discarded == 0:
+            # The fudge is 4: The above plus either the Completed or Discarded Plan header:
+            fudge = 4
+        else:
+            # The fudge is 5 since we have both headers:
+            fudge = 5
+
+        self.assertEqual(num_results, num_active + num_completed + num_discarded + fudge,
+                             "Too many elements in match arrays")
+            
+        # Now iterate through the results array and pick out the results.
+        result_idx = 0
+        self.assertIn("thread #", result_arr[result_idx], "Found thread header") ; result_idx += 1
+        self.assertIn("Active plan stack", result_arr[result_idx], "Found active header") ; result_idx += 1
+        self.assertIn("Element 0: Base thread plan", result_arr[result_idx], "Found base plan") ; result_idx += 1
+
+        for text in active_plans:
+            self.assertFalse("Completed plan stack" in result_arr[result_idx], "Found Completed header too early.")
+            self.assertIn(text, result_arr[result_idx], "Didn't find active plan: %s"%(text)) ; result_idx += 1
+
+        if len(completed_plans) > 0:
+            self.assertIn("Completed plan stack:", result_arr[result_idx], "Found completed plan stack header") ; result_idx += 1
+            for text in completed_plans:
+                self.assertIn(text, result_arr[result_idx], "Didn't find completed plan: %s"%(text)) ; result_idx += 1
+
+        if len(discarded_plans) > 0:
+            self.assertIn("Discarded plan stack:", result_arr[result_idx], "Found discarded plan stack header") ; result_idx += 1
+            for text in discarded_plans:
+                self.assertIn(text, result_arr[result_idx], "Didn't find completed plan: %s"%(text)) ; result_idx += 1
+
+
+    def thread_plan_test(self):
+        (target, process, thread, bkpt) = lldbutil.run_to_source_breakpoint(self,
+                                   "Set a breakpoint here", self.main_source_file)
+
+        # Now set a breakpoint in call_me and step over.  We should have
+        # two public thread plans
+        call_me_bkpt = target.BreakpointCreateBySourceRegex("Set another here", self.main_source_file)
+        self.assertTrue(call_me_bkpt.GetNumLocations() > 0, "Set the breakpoint successfully")
+        thread.StepOver()
+        threads = lldbutil.get_threads_stopped_at_breakpoint(process, call_me_bkpt)
+        self.assertEqual(len(threads), 1, "Hit my breakpoint while stepping over")
+
+        current_id = threads[0].GetIndexID()
+        current_tid = threads[0].GetThreadID()
+        # Run thread plan list without the -i flag:
+        command = "thread plan list %d"%(current_id)
+        self.check_list_output (command, ["Stepping over line main.c"], [])
+
+        # Run thread plan list with the -i flag:
+        command = "thread plan list -i %d"%(current_id)
+        self.check_list_output(command, ["Stepping over line main.c", "Stepping out from"])
+
+        # Run thread plan list providing TID, output should be the same:
+        command = "thread plan list -t %d"%(current_tid)
+        self.check_list_output(command, ["Stepping over line main.c"])
+
+        # Provide both index & tid, and make sure we only print once:
+        command = "thread plan list -t %d %d"%(current_tid, current_id)
+        self.check_list_output(command, ["Stepping over line main.c"])
+
+        # Try a fake TID, and make sure that fails:
+        fake_tid = 0
+        for i in range(100, 10000, 100):
+            fake_tid = current_tid + i
+            thread = process.GetThreadByID(fake_tid)
+            if not thread:
+                break
+        
+        command = "thread plan list -t %d"%(fake_tid)
+        self.check_list_output(command)
+
+        # Now continue, and make sure we printed the completed plan:
+        process.Continue()
+        threads = lldbutil.get_stopped_threads(process, lldb.eStopReasonPlanComplete)
+        self.assertEqual(len(threads), 1, "One thread completed a step")
+        
+        # Run thread plan list - there aren't any private plans at this point:
+        command = "thread plan list %d"%(current_id)
+        self.check_list_output(command, [], ["Stepping over line main.c"])
+
+        # Set another breakpoint that we can run to, to try deleting thread plans.
+        second_step_bkpt = target.BreakpointCreateBySourceRegex("Run here to step over again",
+                                                                self.main_source_file)
+        self.assertTrue(second_step_bkpt.GetNumLocations() > 0, "Set the breakpoint successfully")
+        final_bkpt = target.BreakpointCreateBySourceRegex("Make sure we get here on last continue",
+                                                          self.main_source_file)
+        self.assertTrue(final_bkpt.GetNumLocations() > 0, "Set the breakpoint successfully")
+
+        threads = lldbutil.continue_to_breakpoint(process, second_step_bkpt)
+        self.assertEqual(len(threads), 1, "Hit the second step breakpoint")
+
+        threads[0].StepOver()
+        threads = lldbutil.get_threads_stopped_at_breakpoint(process, call_me_bkpt)
+
+        result = lldb.SBCommandReturnObject()
+        interp = self.dbg.GetCommandInterpreter()
+        interp.HandleCommand("thread plan discard 1", result)
+        self.assertTrue(result.Succeeded(), "Deleted the step over plan: %s"%(result.GetOutput()))
+
+        # Make sure the plan gets listed in the discarded plans:
+        command = "thread plan list %d"%(current_id)
+        self.check_list_output(command, [], [], ["Stepping over line main.c:"])
+
+        process.Continue()
+        threads = lldbutil.get_threads_stopped_at_breakpoint(process, final_bkpt)
+        self.assertEqual(len(threads), 1, "Ran to final breakpoint")
+        threads = lldbutil.get_stopped_threads(process, lldb.eStopReasonPlanComplete)
+        self.assertEqual(len(threads), 0, "Did NOT complete the step over plan")
+

lldb/test/API/functionalities/thread_plan/main.c

+#include <stdio.h>
+
+void
+call_me(int value) {
+  printf("called with %d\n", value); // Set another here.
+}
+
+int
+main(int argc, char **argv)
+{
+  call_me(argc); // Set a breakpoint here.
+  printf("This just spaces the two calls\n");
+  call_me(argc); // Run here to step over again.
+  printf("More spacing\n");
+  return 0; // Make sure we get here on last continue
+}