δΊ 2014/5/2 6:09, Eric Blake ει:
On 04/30/2014 10:26 PM, Wenchao Xia wrote:
Now monitor has been hooked on the new event mechanism, so the patches
later can convert event callers one by one. qmp_query_events() is also
switched to use new generated event defines. Note that old function
monitor_protocol_event() is kept for existing caller to avoid code break,
but rate limiting is bypassed to avoid too many duplicated code. After
convertion, the function would be removed.
s/convertion/conversion/
s/would/will/
Signed-off-by: Wenchao Xia <wenchaoq...@gmail.com>
---
monitor.c | 47 ++++++++++++++++++++++++++---------------------
1 files changed, 26 insertions(+), 21 deletions(-)
@@ -644,7 +648,8 @@ void monitor_protocol_event(MonitorEvent event, QObject
*data)
}
trace_monitor_protocol_event(event, event_name, qmp);
- monitor_protocol_event_queue(event, QOBJECT(qmp));
+ /* Bypass rate limiting for now */
+ monitor_protocol_event_emit(event, QOBJECT(qmp));
QDECREF(qmp);
I'm not quite sure I follow - is this comment evidence of something
temporarily broken for the duration of conversions, that gets fixed
later in the series; or are you breaking event rate limiting?
This change breaks rate limiting, but gets fixed after conversion.
Since I modified old rate limiting code to work in new event mechnism,
so old event mechnism can't work any more, unless I duplicate those
code, make things complicate.