| Message ID | 20220403102144.1679700-3-richard.purdie@linuxfoundation.org |
|---|---|
| State | Accepted, archived |
| Commit | 96a6303949cefd469bcf5ed250ff512271354357 |
| Headers | show |
| Series | [1/5] parse: Ensure any existing siggen is closed down first | expand |
diff --git a/lib/bb/server/process.py b/lib/bb/server/process.py index ce53fdc678..19ef83980f 100644 --- a/lib/bb/server/process.py +++ b/lib/bb/server/process.py @@ -28,6 +28,7 @@ import re import datetime import pickle import traceback +import gc import bb.server.xmlrpcserver from bb import daemonize from multiprocessing import queues @@ -739,8 +740,10 @@ class ConnectionWriter(object): self.event = self def _send(self, obj): + gc.disable() with self.wlock: self.writer.send_bytes(obj) + gc.enable() def send(self, obj): obj = multiprocessing.reduction.ForkingPickler.dumps(obj)
The python gc can trigger whilst we're holding the event stream lock and when cleaning up objects, they can trigger warnings. This translates into a new event which would then need the lock and we can deadlock. Disable gc whilst we hold that lock to avoid this unfortunate and problematic situation. Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org> --- lib/bb/server/process.py | 3 +++ 1 file changed, 3 insertions(+)