[Asterisk-code-review] res_pjsip_pubsub: Postpone destruction of old subscriptions on RLS up... (asterisk[20])
Friendly Automation
asteriskteam at digium.com
Fri Sep 9 08:35:57 CDT 2022
Friendly Automation has submitted this change. ( https://gerrit.asterisk.org/c/asterisk/+/19010 )
Change subject: res_pjsip_pubsub: Postpone destruction of old subscriptions on RLS update
......................................................................
res_pjsip_pubsub: Postpone destruction of old subscriptions on RLS update
Set termination state to old subscriptions to prevent queueing and sending
NOTIFY messages on exten/device state changes.
Postpone destruction of old subscriptions until all already queued tasks
that may be using old subscriptions have completed.
ASTERISK-29906
Change-Id: I96582aad3a26515ca73a8460ee6756f56f6ba23b
---
M res/res_pjsip_exten_state.c
M res/res_pjsip_pubsub.c
2 files changed, 47 insertions(+), 2 deletions(-)
Approvals:
Joshua Colp: Looks good to me, but someone else must approve
George Joseph: Looks good to me, approved
Friendly Automation: Approved for Submit
diff --git a/res/res_pjsip_exten_state.c b/res/res_pjsip_exten_state.c
index b51df87..3c7baee 100644
--- a/res/res_pjsip_exten_state.c
+++ b/res/res_pjsip_exten_state.c
@@ -296,7 +296,8 @@
.body_data = &task_data->exten_state_data,
};
- /* Terminated subscriptions are no longer associated with a valid tree, and sending
+ /* The subscription was terminated while notify_task was in queue.
+ Terminated subscriptions are no longer associated with a valid tree, and sending
* NOTIFY messages on a subscription which has already been terminated won't work.
*/
if (ast_sip_subscription_is_terminated(task_data->exten_state_sub->sip_sub)) {
@@ -339,6 +340,13 @@
struct notify_task_data *task_data;
struct exten_state_subscription *exten_state_sub = data;
+ /* Terminated subscriptions are no longer associated with a valid tree.
+ * Do not queue notify_task.
+ */
+ if (ast_sip_subscription_is_terminated(exten_state_sub->sip_sub)) {
+ return 0;
+ }
+
if (!(task_data = alloc_notify_task_data(exten, exten_state_sub, info))) {
return -1;
}
diff --git a/res/res_pjsip_pubsub.c b/res/res_pjsip_pubsub.c
index d0f3717..4086445 100644
--- a/res/res_pjsip_pubsub.c
+++ b/res/res_pjsip_pubsub.c
@@ -3994,6 +3994,15 @@
return 0;
}
+static int destroy_subscriptions_task(void *obj)
+{
+ struct ast_sip_subscription *sub = (struct ast_sip_subscription *) obj;
+
+ destroy_subscriptions(sub);
+
+ return 0;
+}
+
/*!
* \brief Called whenever an in-dialog SUBSCRIBE is received
*
@@ -4067,8 +4076,19 @@
AST_SCHED_DEL_UNREF(sched, sub_tree->notify_sched_id, ao2_ref(sub_tree, -1));
sub_tree->send_scheduled_notify = 0;
}
+
+ /* Terminate old subscriptions to stop sending NOTIFY messages on exten/device state changes */
+ set_state_terminated(old_root);
+
+ /* Shutdown old subscriptions to remove exten/device state change callbacks
+ that can queue tasks for old subscriptions */
shutdown_subscriptions(old_root);
- destroy_subscriptions(old_root);
+
+ /* Postpone destruction until all already queued tasks that may be using old subscriptions have completed */
+ if (ast_sip_push_task(sub_tree->serializer, destroy_subscriptions_task, old_root)) {
+ ast_log(LOG_ERROR, "Failed to push task to destroy old subscriptions for RLS '%s->%s'.\n",
+ ast_sorcery_object_get_id(endpoint), old_root->resource);
+ }
} else {
destroy_subscriptions(new_root);
}
--
To view, visit https://gerrit.asterisk.org/c/asterisk/+/19010
To unsubscribe, or for help writing mail filters, visit https://gerrit.asterisk.org/settings
Gerrit-Project: asterisk
Gerrit-Branch: 20
Gerrit-Change-Id: I96582aad3a26515ca73a8460ee6756f56f6ba23b
Gerrit-Change-Number: 19010
Gerrit-PatchSet: 2
Gerrit-Owner: Alexei Gradinari <alex2grad at gmail.com>
Gerrit-Reviewer: Friendly Automation
Gerrit-Reviewer: George Joseph <gjoseph at digium.com>
Gerrit-Reviewer: Joshua Colp <jcolp at sangoma.com>
Gerrit-MessageType: merged
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-code-review/attachments/20220909/32990843/attachment-0001.html>
More information about the asterisk-code-review
mailing list