| Message ID | 20180301084438.13594-5-peterx@redhat.com |
|---|---|
| State | New |
| Headers | show |
| Series | qio: general non-default GMainContext support | expand |
On Thu, Mar 01, 2018 at 04:44:27PM +0800, Peter Xu wrote: > The old incoming migration is running in main thread and default > gcontext. With the new qio_channel_add_watch_full() we can now let it > run in the thread's own gcontext (if there is one). > > Currently this patch does nothing alone. But when any of the incoming > migration is run in another iothread (e.g., the upcoming migrate-recover > command), this patch will bind the incoming logic to the iothread > instead of the main thread (which may already get page faulted and > hanged). > > RDMA is not considered for now since it's not even using the QIO watch > framework at all. > > CC: Juan Quintela <quintela@redhat.com> > CC: Dr. David Alan Gilbert <dgilbert@redhat.com> > CC: Laurent Vivier <lvivier@redhat.com> > Signed-off-by: Peter Xu <peterx@redhat.com> > --- > migration/exec.c | 9 ++++----- > migration/fd.c | 9 ++++----- > migration/socket.c | 10 +++++----- > 3 files changed, 13 insertions(+), 15 deletions(-) This should probably just be in a separate series, since it does nothing on its own, and nothing following in this series touches migration at all. Regards, Daniel
On Thu, Mar 01, 2018 at 04:03:44PM +0000, Daniel P. Berrangé wrote: > On Thu, Mar 01, 2018 at 04:44:27PM +0800, Peter Xu wrote: > > The old incoming migration is running in main thread and default > > gcontext. With the new qio_channel_add_watch_full() we can now let it > > run in the thread's own gcontext (if there is one). > > > > Currently this patch does nothing alone. But when any of the incoming > > migration is run in another iothread (e.g., the upcoming migrate-recover > > command), this patch will bind the incoming logic to the iothread > > instead of the main thread (which may already get page faulted and > > hanged). > > > > RDMA is not considered for now since it's not even using the QIO watch > > framework at all. > > > > CC: Juan Quintela <quintela@redhat.com> > > CC: Dr. David Alan Gilbert <dgilbert@redhat.com> > > CC: Laurent Vivier <lvivier@redhat.com> > > Signed-off-by: Peter Xu <peterx@redhat.com> > > --- > > migration/exec.c | 9 ++++----- > > migration/fd.c | 9 ++++----- > > migration/socket.c | 10 +++++----- > > 3 files changed, 13 insertions(+), 15 deletions(-) > > This should probably just be in a separate series, since it does nothing > on its own, and nothing following in this series touches migration at all. It was trying to solve all problems related to QIO+context, and migration is just one user of it. But sure I can postpone this patch to the postcopy recovery series. Thanks,
diff --git a/migration/exec.c b/migration/exec.c index 0bc5a427dd..9d0f82f1f0 100644 --- a/migration/exec.c +++ b/migration/exec.c @@ -65,9 +65,8 @@ void exec_start_incoming_migration(const char *command, Error **errp) } qio_channel_set_name(ioc, "migration-exec-incoming"); - qio_channel_add_watch(ioc, - G_IO_IN, - exec_accept_incoming_migration, - NULL, - NULL); + qio_channel_add_watch_full(ioc, G_IO_IN, + exec_accept_incoming_migration, + NULL, NULL, + g_main_context_get_thread_default()); } diff --git a/migration/fd.c b/migration/fd.c index cd06182d1e..9a380bbbc4 100644 --- a/migration/fd.c +++ b/migration/fd.c @@ -66,9 +66,8 @@ void fd_start_incoming_migration(const char *infd, Error **errp) } qio_channel_set_name(QIO_CHANNEL(ioc), "migration-fd-incoming"); - qio_channel_add_watch(ioc, - G_IO_IN, - fd_accept_incoming_migration, - NULL, - NULL); + qio_channel_add_watch_full(ioc, G_IO_IN, + fd_accept_incoming_migration, + NULL, NULL, + g_main_context_get_thread_default()); } diff --git a/migration/socket.c b/migration/socket.c index e090097077..60d732535c 100644 --- a/migration/socket.c +++ b/migration/socket.c @@ -173,11 +173,11 @@ static void socket_start_incoming_migration(SocketAddress *saddr, return; } - qio_channel_add_watch(QIO_CHANNEL(listen_ioc), - G_IO_IN, - socket_accept_incoming_migration, - listen_ioc, - (GDestroyNotify)object_unref); + qio_channel_add_watch_full(QIO_CHANNEL(listen_ioc), G_IO_IN, + socket_accept_incoming_migration, + listen_ioc, + (GDestroyNotify)object_unref, + g_main_context_get_thread_default()); } void tcp_start_incoming_migration(const char *host_port, Error **errp)
The old incoming migration is running in main thread and default gcontext. With the new qio_channel_add_watch_full() we can now let it run in the thread's own gcontext (if there is one). Currently this patch does nothing alone. But when any of the incoming migration is run in another iothread (e.g., the upcoming migrate-recover command), this patch will bind the incoming logic to the iothread instead of the main thread (which may already get page faulted and hanged). RDMA is not considered for now since it's not even using the QIO watch framework at all. CC: Juan Quintela <quintela@redhat.com> CC: Dr. David Alan Gilbert <dgilbert@redhat.com> CC: Laurent Vivier <lvivier@redhat.com> Signed-off-by: Peter Xu <peterx@redhat.com> --- migration/exec.c | 9 ++++----- migration/fd.c | 9 ++++----- migration/socket.c | 10 +++++----- 3 files changed, 13 insertions(+), 15 deletions(-)