[Lldb-commits] [PATCH] D157556: Replace the singleton "ShadowListener" with a primary and N secondary listeners

Alex Langford via Phabricator via lldb-commits lldb-commits at lists.llvm.org
Wed Aug 16 10:28:22 PDT 2023


bulbazord requested changes to this revision.
bulbazord added a comment.
This revision now requires changes to proceed.

Thanks for all the work to clean up some of those interfaces! I know that was not straightforward. :)

I left a few more comments about minor things I noticed while re-reading this. I'm going to request changes to put this back into your queue.



================
Comment at: lldb/include/lldb/Utility/Event.h:233
+    m_pending_listeners.push_back(pending_listener_sp);
+  };
+
----------------
nit: trailing semicolon


================
Comment at: lldb/source/Target/Process.cpp:620-621
 
-  if (m_listener_sp->GetEventForBroadcaster(this, event_sp,
+  if (GetPrimaryListener()->GetEventForBroadcaster(this, event_sp,
                                             std::chrono::seconds(0)) &&
       event_sp)
----------------
Do you need to check for the validity of the primary listener before using it? It can point to nullptr from what I can tell


================
Comment at: lldb/source/Utility/Broadcaster.cpp:59
+    max_count++;
+  listeners.reserve(max_count);
 
----------------
I wonder if we should be reserving at all considering we're using an `llvm::SmallVector<$type, 4>`.  The point of a SmallVector is that on average it should be small and only sometimes should we be exceeding that hardcoded size. This means that if there are 8 listeners but only 2 are relevant, we're taking a SmallVector of size 4 and allocating space for 8 listeners (always triggering a memory allocation) even if it will only ever have 2 listeners in it.

I know that's the existing behavior and you're just preserving it, but something to think about for the future.


================
Comment at: lldb/source/Utility/Broadcaster.cpp:179-181
+  // The primary listener listens for all event bits:
+  if (m_primary_listener_sp)
+    return true;
----------------
This check is a bit redundant since `HasListeners` performs the exact same check


================
Comment at: lldb/source/Utility/Broadcaster.cpp:198
+
   std::lock_guard<std::recursive_mutex> guard(m_listeners_mutex);
+  for (auto it = m_listeners.begin(); it != m_listeners.end();) {
----------------



Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D157556/new/

https://reviews.llvm.org/D157556



More information about the lldb-commits mailing list