[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <DB9PR05MB9078A431986292D93127682A88B8A@DB9PR05MB9078.eurprd05.prod.outlook.com>
Date: Fri, 24 Nov 2023 10:22:28 +0000
From: Tung Quang Nguyen <tung.q.nguyen@...tech.com.au>
To: xu <xu.xin.sc@...il.com>
CC: "davem@...emloft.net" <davem@...emloft.net>,
"jmaloy@...hat.com" <jmaloy@...hat.com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
"tipc-discussion@...ts.sourceforge.net"
<tipc-discussion@...ts.sourceforge.net>,
"xu.xin16@....com.cn" <xu.xin16@....com.cn>,
"yang.yang29@....com.cn" <yang.yang29@....com.cn>,
"ying.xue@...driver.com" <ying.xue@...driver.com>,
"zhang.yunkai@....com.cn" <zhang.yunkai@....com.cn>
Subject: RE: [RFC PATCH] net/tipc: reduce tipc_node lock holding time in
tipc_rcv
>Why can't use le->lock instead of node's lock to protect it in tipc_link_rcv.
>
I have already explained:
__tipc_node_link_down()
{
...
if (!l || tipc_link_is_reset(l)) <-- read link status
...
}
>>What I showed you were just 2 use cases (link reset/delete). There are more use cases (netlink, transmit path etc) that need proper
>locks.
>
>The same. We can also add spin_lock_bh(&le->lock) to protect the link in other places where it changes the link status in addition to
>'reset/delete'. Because using node lock to protect the link in tipc_link_rcv is really wasting CPU performance.
>
If you want to change current lock policy, you need to submit a complete/correct patch. I will acknowledge this patch if I can see a significant improvement in my test.
Powered by blists - more mailing lists