BUG/MEDIUM: mux-h2: immediately remove a failed connection from the idle list

If a connection faces an error or a timeout, it must be removed from its
idle list ASAP. We certainly don't want to risk sending new streams on
it.

This should be backported to 2.0 (replacing MT_LIST_DEL with LIST_DEL_LOCKED)
and 1.9 (there's no lock there, the idle lists are per-thread and per-server
however a LIST_DEL_INIT will be needed).
This commit is contained in:
Willy Tarreau 2019-10-31 15:36:30 +01:00
parent c61966f9b4
commit 4481e26e5d

View File

@ -3626,6 +3626,17 @@ static int h2_process(struct h2c *h2c)
TRACE_DEVEL("leaving after releasing the connection", H2_EV_H2C_WAKE);
return -1;
}
/* connections in error must be removed from the idle lists */
HA_SPIN_LOCK(OTHER_LOCK, &toremove_lock[tid]);
MT_LIST_DEL((struct mt_list *)&conn->list);
HA_SPIN_UNLOCK(OTHER_LOCK, &toremove_lock[tid]);
}
else if (h2c->st0 == H2_CS_ERROR) {
/* connections in error must be removed from the idle lists */
HA_SPIN_LOCK(OTHER_LOCK, &toremove_lock[tid]);
MT_LIST_DEL((struct mt_list *)&conn->list);
HA_SPIN_UNLOCK(OTHER_LOCK, &toremove_lock[tid]);
}
if (!b_data(&h2c->dbuf))
@ -3733,6 +3744,11 @@ static struct task *h2_timeout_task(struct task *t, void *context, unsigned shor
offer_buffers(NULL, tasks_run_queue);
}
/* in any case this connection must not be considered idle anymore */
HA_SPIN_LOCK(OTHER_LOCK, &toremove_lock[tid]);
MT_LIST_DEL((struct mt_list *)&h2c->conn->list);
HA_SPIN_UNLOCK(OTHER_LOCK, &toremove_lock[tid]);
/* either we can release everything now or it will be done later once
* the last stream closes.
*/