================
@@ -623,8 +630,20 @@ void Debugger::Terminate() {
}
if (g_thread_pool) {
- // The destructor will wait for all the threads to complete.
- delete g_thread_pool;
+ // Delete the thread pool in a different thread so we can set a timeout.
----------------
JDevlieghere wrote:
> Or is it too late and no one would receive events?
Yup, that's exactly the problem. The debugger has already been destroyed at
this point, otherwise I could've written to the debugger's output stream. By
the time we terminate, there's pretty much nothing left at this point. Before
I had the callback I just wrote to stderr directly, but that's not a very nice
thing to do as a library.
> Another option would be to have the thread pool detach from all threads so
> that we don't need to wait until they complete and just go ahead and exit?
That's what we did in the past and that caused a bunch of crashes. If you want
to do that without crashing, you need to be very careful about what you do in a
task running on the thread pool, because the debugger might have been destroyed
and terminated while you were running.
https://github.com/llvm/llvm-project/pull/82799
_______________________________________________
lldb-commits mailing list
lldb-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
