diff --git a/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.cpp b/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.cpp
--- a/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.cpp
+++ b/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.cpp
@@ -3689,12 +3689,25 @@
             __FUNCTION__, arg, process->GetID());
 
   EventSP event_sp;
+
+  // We need to ignore any packets that come in after we have
+  // have decided the process has exited.  There are some
+  // situations, for instance when we try to interrupt a running
+  // process and the interrupt fails, where another packet might
+  // get delivered after we've decided to give up on the process.
+  // But once we've decided we are done with the process we will
+  // not be in a state to do anything useful with new packets.
+  // So it is safer to simply ignore any remaining packets by
+  // explicitly checking for eStateExited before reentering the
+  // fetch loop.
+  
   bool done = false;
-  while (!done) {
+  while (!done && process->GetPrivateState() != eStateExited) {
     LLDB_LOGF(log,
               "ProcessGDBRemote::%s (arg = %p, pid = %" PRIu64
               ") listener.WaitForEvent (NULL, event_sp)...",
               __FUNCTION__, arg, process->GetID());
+
     if (process->m_async_listener_sp->GetEvent(event_sp, llvm::None)) {
       const uint32_t event_type = event_sp->GetType();
       if (event_sp->BroadcasterIs(&process->m_async_broadcaster)) {
@@ -3793,6 +3806,7 @@
                 } else {
                   process->SetExitStatus(-1, "lost connection");
                 }
+                done = true;
                 break;
               }
 
diff --git a/lldb/test/API/functionalities/gdb_remote_client/TestHaltFails.py b/lldb/test/API/functionalities/gdb_remote_client/TestHaltFails.py
new file mode 100644
--- /dev/null
+++ b/lldb/test/API/functionalities/gdb_remote_client/TestHaltFails.py
@@ -0,0 +1,72 @@
+from __future__ import print_function
+import lldb
+from lldbsuite.test.lldbtest import *
+from lldbsuite.test.decorators import *
+from gdbclientutils import *
+
+
+class TestHaltFails(GDBRemoteTestBase):
+
+    class MyResponder(MockGDBServerResponder):
+        
+        def setBreakpoint(self, packet):
+            return "OK"
+        
+        def interrupt(self):
+            # Simulate process waiting longer than the interrupt
+            # timeout to stop, then sending the reply.
+            time.sleep(14)
+            return "T02reason:signal"
+        
+        def cont(self):
+            # No response, wait for the client to interrupt us.
+            return None
+        
+    def wait_for_and_check_event(self, wait_time, value):
+        event = lldb.SBEvent()
+        got_event = self.dbg.GetListener().WaitForEvent(wait_time, event)
+        self.assertTrue(got_event, "Failed to get event after wait")
+        self.assertTrue(lldb.SBProcess.EventIsProcessEvent(event), "Event was not a process event")
+        event_type = lldb.SBProcess.GetStateFromEvent(event)
+        self.assertEqual(event_type, value)
+        
+    def get_to_running(self):
+        self.server.responder = self.MyResponder()
+        self.target = self.createTarget("a.yaml")
+        process = self.connect(self.target)
+        self.dbg.SetAsync(True)
+
+        # There should be a stopped event, consume that:
+        self.wait_for_and_check_event(2, lldb.eStateStopped)
+        process.Continue()
+
+        # There should be a running event, consume that:
+        self.wait_for_and_check_event(2, lldb.eStateRunning)
+        return process
+
+    @skipIfReproducer # FIXME: Unexpected packet during (passive) replay
+    def test_destroy_while_running(self):
+        process = self.get_to_running()
+        process.Destroy()
+
+        # Again pretend that after failing to be interrupted, we delivered the stop
+        # and make sure we still exit properly.
+        self.wait_for_and_check_event(14, lldb.eStateExited)
+            
+    @skipIfReproducer # FIXME: Unexpected packet during (passive) replay
+    def test_async_interrupt(self):
+        """
+        Test that explicitly calling AsyncInterrupt, which then fails, leads
+        to an "eStateExited" state.
+        """
+        process = self.get_to_running()
+        # Now do the interrupt:
+        process.SendAsyncInterrupt()
+
+        # That should have caused the Halt to time out and we should
+        # be in eStateExited:
+        self.wait_for_and_check_event(15, lldb.eStateExited)
+
+        
+
+