On Wed, 2021-07-14 at 11:50 -0400, Chuck Lever wrote:
> In some rare failure modes, the server is actually reading the
> transport, but then just dropping the requests on the floor.
> TCP_USER_TIMEOUT cannot detect that case.
>
> Prevent such a stuck server from pinning client resources
> indefinitely by ensuring that async lease renewal requests can time
> out even if the connection is still operational.
>
> Signed-off-by: Chuck Lever <chuck.lever@xxxxxxxxxx>
> ---
> fs/nfs/nfs4proc.c | 9 +++++++++
> 1 file changed, 9 insertions(+)
>
> diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
> index e1214bb6b7ee..346217f6a00b 100644
> --- a/fs/nfs/nfs4proc.c
> +++ b/fs/nfs/nfs4proc.c
> @@ -5612,6 +5612,12 @@ struct nfs4_renewdata {
> * nfs4_proc_async_renew(): This is not one of the nfs_rpc_ops; it
> is a special
> * standalone procedure for queueing an asynchronous RENEW.
> */
> +static void nfs4_renew_prepare(struct rpc_task *task, void
> *calldata)
> +{
> + task->tk_flags &= ~RPC_TASK_NO_RETRANS_TIMEOUT;
> + rpc_call_start(task);
> +}
> +
> static void nfs4_renew_release(void *calldata)
> {
> struct nfs4_renewdata *data = calldata;
> @@ -5650,6 +5656,7 @@ static void nfs4_renew_done(struct rpc_task
> *task, void *calldata)
> }
>
> static const struct rpc_call_ops nfs4_renew_ops = {
> + .rpc_call_prepare = nfs4_renew_prepare,
> .rpc_call_done = nfs4_renew_done,
> .rpc_release = nfs4_renew_release,
> };
> @@ -9219,6 +9226,8 @@ static void nfs41_sequence_prepare(struct
> rpc_task *task, void *data)
> struct nfs4_sequence_args *args;
> struct nfs4_sequence_res *res;
>
> + task->tk_flags &= ~RPC_TASK_NO_RETRANS_TIMEOUT;
> +
> args = task->tk_msg.rpc_argp;
> res = task->tk_msg.rpc_resp;
>
>
>
This isn't necessary. The server isn't allowed to drop these calls on
the floor.
--
Trond Myklebust
Linux NFS client maintainer, Hammerspace
trond.myklebust@xxxxxxxxxxxxxxx
