Commit 05eac20f authored by Max Kellermann's avatar Max Kellermann

lib/nfs/Connection: detect libnfs reconnect

When rpc_reconnect_requeue() gets called from inside nfs_service(), the NfsInputStream can stall completely because the old socket has been unregistered from epoll automatically, but the new one has never been registered. Therefore, nfs_service() will never be called again. This kludge attempts to detect this condition by checking nfs_which_events()==POLLOUT. https://bugs.musicpd.org/view.php?id=4081
parent 38d263ac
ver 0.20.4 (not yet released)
* input
- nfs: fix freeze after reconnect
* output
- sndio: work around a libroar C++ incompatibility
* workaround for GCC 4.9 "constexpr" bug
......
......@@ -396,6 +396,17 @@ NfsConnection::ScheduleSocket()
assert(GetEventLoop().IsInside());
assert(context != nullptr);
const int which_events = nfs_which_events(context);
if (which_events == POLLOUT && SocketMonitor::IsDefined())
/* kludge: if libnfs asks only for POLLOUT, it means
that it is currently waiting for the connect() to
finish - rpc_reconnect_requeue() may have been
called from inside nfs_service(); we must now
unregister the old socket and register the new one
instead */
SocketMonitor::Steal();
if (!SocketMonitor::IsDefined()) {
int _fd = nfs_get_fd(context);
if (_fd < 0)
......@@ -405,7 +416,7 @@ NfsConnection::ScheduleSocket()
SocketMonitor::Open(_fd);
}
SocketMonitor::Schedule(libnfs_to_events(nfs_which_events(context)));
SocketMonitor::Schedule(libnfs_to_events(which_events));
}
inline int
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment