daemon/worker send_waiting(): be more defensive

We encountered non-recoverable assertions due to popping
from an empty queue here, but I see no reason to block recovery here.
I'm still keeping it as a soft assertion until it's better understood.

I *suspect* what happened is that:
 - multiple queries queued up before outgoing TCP handshake completed
 - the session got into closing state for some reason
   *before* processing this whole queue
 - during that the queue got emptied

diff --git a/NEWS b/NEWS
index fd589b89d6e2ca7fcf7c28ceea231f87cb050400..5ff5a7b9042f6433a03e81aeb190e39a7401740d 100644 (file)
--- a/NEWS
+++ b/NEWS
@@ -11,6 +11,8 @@ Bugfixes
 Improvements
 ------------
 - /local-data/rpz/*/watchdog: new configuration to enable watchdog for RPZ files (!1665)
+- daemon: fix rare crashes with either of the lines below
+  [system] requirement "h && h->end > h->begin" failed in queue_pop_impl
 
 
 Knot Resolver 6.0.11 (2025-02-26)

diff --git a/daemon/worker.c b/daemon/worker.c
index 70963afdd09c98613be4a25b2a1c4ee67dc4b7f9..a58db0cafca03c31a182549fd8b48049a58748bc 100644 (file)
--- a/daemon/worker.c
+++ b/daemon/worker.c
@@ -732,7 +732,14 @@ static int send_waiting(struct session2 *session)
                        session2_close(session);
                        break;
                }
-               session2_waitinglist_pop(session, true);
+               // Let's be a bit defensive and check that nothing's changed before _pop()
+               // and recover if it has, as qr_task_send() is rather complex.
+               // TODO: fully analyze why the waitinglist could get empty here.
+               if (session2_waitinglist_get(session) == t) {
+                       session2_waitinglist_pop(session, true);
+               } else { // a normal assertion could kr_error_log() too much in some rarer cases
+                       VERBOSE_MSG(NULL, "soft assertion: waitinglist mismatch in send_waiting()\n");
+               }
        } while (!session2_waitinglist_is_empty(session));
        defer_sample_stop(&defer_prev_sample_state, true);