fix(ble/bluedroid): allow osi_event re-post during POSTING window

Do not reject osi_thread_post_event() when only POSTING is set.
QUEUED already prevents double-queueing; rejecting POSTING caused
HCI downstream lost wakeup. Add generic osi_event and hci downstream
diagnostics for post failures.

components/bt/host/bluedroid/hci/hci_layer.c

@@ -155,11 +155,21 @@ void hci_shut_down(void)
 
 bool hci_downstream_data_post(uint32_t timeout)
 {
+    bool ret;
+
     if (hci_host_env.downstream_data_ready == NULL) {
         HCI_TRACE_WARNING("%s downstream_data_ready event not created", __func__);
         return false;
     }
-    return osi_thread_post_event(hci_host_env.downstream_data_ready, timeout);
+
+    ret = osi_thread_post_event(hci_host_env.downstream_data_ready, timeout);
+    if (!ret) {
+        HCI_TRACE_DEBUG("%s post fail credits=%d cmdq=%u pktq=%u",
+                        __func__, hci_host_env.command_credits,
+                        (unsigned)fixed_pkt_queue_length(hci_host_env.command_queue),
+                        (unsigned)fixed_queue_length(hci_host_env.packet_queue));
+    }
+    return ret;
 }
 
 static int hci_layer_init_env(void)
@@ -249,6 +259,8 @@ static void hci_downstream_data_handler(void *arg)
      * All packets will be directly copied to single queue in driver layer with
      * H4 type header added (1 byte).
      */
+    UNUSED(arg);
+
     while (hci_host_check_send_available()) {
         /*Now Target only allowed one packet per TX*/
         BT_HDR *pkt = packet_fragmenter->fragment_current_packet();
@@ -264,6 +276,11 @@ static void hci_downstream_data_handler(void *arg)
             break;
         }
     }
+
+    HCI_TRACE_DEBUG("%s done credits=%d cmdq=%u pktq=%u",
+                    __func__, hci_host_env.command_credits,
+                    (unsigned)fixed_pkt_queue_length(hci_host_env.command_queue),
+                    (unsigned)fixed_queue_length(hci_host_env.packet_queue));
 }
 
 static void transmit_command(