[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <c5c17989-4c1e-35d2-5a75-a27e58cf6673@linux.alibaba.com>
Date: Mon, 20 Dec 2021 11:39:26 +0800
From: Bixuan Cui <cuibixuan@...ux.alibaba.com>
To: linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
linux-nfs@...r.kernel.org
Cc: bfields@...ldses.org, chuck.lever@...cle.com,
trond.myklebust@...merspace.com, anna.schumaker@...app.com,
davem@...emloft.net, kuba@...nel.org, pete.wl@...baba-inc.com,
wenan.mwa@...baba-inc.com, xiaoh.peixh@...baba-inc.com,
weipu.zy@...baba-inc.com
Subject: Re: [PATCH -next] SUNRPC: Clean XPRT_CONGESTED of xprt->state when
rpc task is killed
ping~
在 2021/12/14 下午9:53, Bixuan Cui 写道:
> When the values of tcp_max_slot_table_entries and
> sunrpc.tcp_slot_table_entries are lower than the number of rpc tasks,
> xprt_dynamic_alloc_slot() in xprt_alloc_slot() will return -EAGAIN, and
> then set xprt->state to XPRT_CONGESTED:
> xprt_retry_reserve
> ->xprt_do_reserve
> ->xprt_alloc_slot
> ->xprt_dynamic_alloc_slot // return -EAGAIN and task->tk_rqstp is NULL
> ->xprt_add_backlog // set_bit(XPRT_CONGESTED, &xprt->state);
>
> When rpc task is killed, XPRT_CONGESTED bit of xprt->state will not be
> cleaned up and nfs hangs:
> rpc_exit_task
> ->xprt_release // if (req == NULL) is true, then XPRT_CONGESTED
> // bit not clean
>
> Add xprt_wake_up_backlog(xprt) to clean XPRT_CONGESTED bit in
> xprt_release().
>
> Signed-off-by: Bixuan Cui <cuibixuan@...ux.alibaba.com>
> Signed-off-by: Xiaohui Pei <xiaoh.peixh@...baba-inc.com>
> ---
> net/sunrpc/xprt.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
> index a02de2b..70d11ae 100644
> --- a/net/sunrpc/xprt.c
> +++ b/net/sunrpc/xprt.c
> @@ -1952,6 +1952,7 @@ void xprt_release(struct rpc_task *task)
> if (req == NULL) {
> if (task->tk_client) {
> xprt = task->tk_xprt;
> + xprt_wake_up_backlog(xprt);
> xprt_release_write(xprt, task);
> }
> return;
Powered by blists - more mailing lists