@@ -93,6 +93,7 @@ struct rpc_clnt {
const struct cred *cl_cred;
unsigned int cl_max_connect; /* max number of transports not to the same IP */
struct super_block *pipefs_sb;
+ atomic_t cl_task_count;
};
/*
@@ -958,12 +958,17 @@ void rpc_shutdown_client(struct rpc_clnt *clnt)
trace_rpc_clnt_shutdown(clnt);
+ clnt->cl_shutdown = 1;
while (!list_empty(&clnt->cl_tasks)) {
rpc_killall_tasks(clnt);
wait_event_timeout(destroy_wait,
list_empty(&clnt->cl_tasks), 1*HZ);
}
+ /* wait for tasks still in workqueue or waitqueue */
+ wait_event_timeout(destroy_wait,
+ atomic_read(&clnt->cl_task_count) == 0, 1 * HZ);
+
rpc_release_client(clnt);
}
EXPORT_SYMBOL_GPL(rpc_shutdown_client);
@@ -1139,6 +1144,7 @@ void rpc_task_release_client(struct rpc_task *task)
list_del(&task->tk_task);
spin_unlock(&clnt->cl_lock);
task->tk_client = NULL;
+ atomic_dec(&clnt->cl_task_count);
rpc_release_client(clnt);
}
@@ -1189,10 +1195,7 @@ void rpc_task_set_client(struct rpc_task *task, struct rpc_clnt *clnt)
task->tk_flags |= RPC_TASK_TIMEOUT;
if (clnt->cl_noretranstimeo)
task->tk_flags |= RPC_TASK_NO_RETRANS_TIMEOUT;
- /* Add to the client's list of all tasks */
- spin_lock(&clnt->cl_lock);
- list_add_tail(&task->tk_task, &clnt->cl_tasks);
- spin_unlock(&clnt->cl_lock);
+ atomic_inc(&clnt->cl_task_count);
}
static void
@@ -1787,9 +1790,14 @@ call_reserveresult(struct rpc_task *task)
if (status >= 0) {
if (task->tk_rqstp) {
task->tk_action = call_refresh;
+
+ /* Add to the client's list of all tasks */
+ spin_lock(&task->tk_client->cl_lock);
+ if (list_empty(&task->tk_task))
+ list_add_tail(&task->tk_task, &task->tk_client->cl_tasks);
+ spin_unlock(&task->tk_client->cl_lock);
return;
}
-
rpc_call_rpcerror(task, -EIO);
return;
}
Under heavy write load, we've seen the cl_tasks list grows to millions of entries. Even though the list is extremely long, the system still runs fine until the user wants to get the information of all active RPC tasks by doing: When this happens, tasks_start acquires the cl_lock to walk the cl_tasks list, returning one entry at a time to the caller. The cl_lock is held until all tasks on this list have been processed. While the cl_lock is held, completed RPC tasks have to spin wait in rpc_task_release_client for the cl_lock. If there are millions of entries in the cl_tasks list it will take a long time before tasks_stop is called and the cl_lock is released. The spin wait tasks can use up all the available CPUs in the system, preventing other jobs to run, this causes the system to temporarily lock up. This patch fixes this problem by delaying inserting the RPC task on the cl_tasks list until the RPC call slot is reserved. This limits the length of the cl_tasks to the number of call slots available in the system. Signed-off-by: Dai Ngo <dai.ngo@oracle.com> --- include/linux/sunrpc/clnt.h | 1 + net/sunrpc/clnt.c | 18 +++++++++++++----- 2 files changed, 14 insertions(+), 5 deletions(-)