Hi Ben,
On Thu, Jan 4, 2024 at 9:58 AM Benjamin Coddington <bcodding@xxxxxxxxxx> wrote:
>
> For backchannel requests that lookup the appropriate nfs_client, use the
> state-management rpc_clnt's rpc_timeout parameters for the backchannel's
> response. When the nfs_client cannot be found, fall back to using the
> xprt's default timeout parameters.
Thanks for sending the v4, it fixes the problem I was seeing yesterday!
Anna
>
> Signed-off-by: Benjamin Coddington <bcodding@xxxxxxxxxx>
> ---
> fs/nfs/callback_xdr.c | 5 +++++
> include/linux/sunrpc/bc_xprt.h | 3 ++-
> include/linux/sunrpc/sched.h | 14 +++++++++++++-
> include/linux/sunrpc/svc.h | 2 ++
> include/linux/sunrpc/xprt.h | 11 -----------
> net/sunrpc/clnt.c | 6 ++++--
> net/sunrpc/svc.c | 11 ++++++++++-
> net/sunrpc/xprt.c | 12 +++++++++---
> 8 files changed, 45 insertions(+), 19 deletions(-)
>
> diff --git a/fs/nfs/callback_xdr.c b/fs/nfs/callback_xdr.c
> index 321af81c456e..9369488f2ed4 100644
> --- a/fs/nfs/callback_xdr.c
> +++ b/fs/nfs/callback_xdr.c
> @@ -967,6 +967,11 @@ static __be32 nfs4_callback_compound(struct svc_rqst *rqstp)
> nops--;
> }
>
> + if (svc_is_backchannel(rqstp) && cps.clp) {
> + rqstp->bc_to_initval = cps.clp->cl_rpcclient->cl_timeout->to_initval;
> + rqstp->bc_to_retries = cps.clp->cl_rpcclient->cl_timeout->to_retries;
> + }
> +
> *hdr_res.status = status;
> *hdr_res.nops = htonl(nops);
> nfs4_cb_free_slot(&cps);
> diff --git a/include/linux/sunrpc/bc_xprt.h b/include/linux/sunrpc/bc_xprt.h
> index db30a159f9d5..f22bf915dcf6 100644
> --- a/include/linux/sunrpc/bc_xprt.h
> +++ b/include/linux/sunrpc/bc_xprt.h
> @@ -20,7 +20,8 @@
> #ifdef CONFIG_SUNRPC_BACKCHANNEL
> struct rpc_rqst *xprt_lookup_bc_request(struct rpc_xprt *xprt, __be32 xid);
> void xprt_complete_bc_request(struct rpc_rqst *req, uint32_t copied);
> -void xprt_init_bc_request(struct rpc_rqst *req, struct rpc_task *task);
> +void xprt_init_bc_request(struct rpc_rqst *req, struct rpc_task *task,
> + const struct rpc_timeout *to);
> void xprt_free_bc_request(struct rpc_rqst *req);
> int xprt_setup_backchannel(struct rpc_xprt *, unsigned int min_reqs);
> void xprt_destroy_backchannel(struct rpc_xprt *, unsigned int max_reqs);
> diff --git a/include/linux/sunrpc/sched.h b/include/linux/sunrpc/sched.h
> index 8ada7dc802d3..2d61987b3545 100644
> --- a/include/linux/sunrpc/sched.h
> +++ b/include/linux/sunrpc/sched.h
> @@ -37,6 +37,17 @@ struct rpc_wait {
> struct list_head timer_list; /* Timer list */
> };
>
> +/*
> + * This describes a timeout strategy
> + */
> +struct rpc_timeout {
> + unsigned long to_initval, /* initial timeout */
> + to_maxval, /* max timeout */
> + to_increment; /* if !exponential */
> + unsigned int to_retries; /* max # of retries */
> + unsigned char to_exponential;
> +};
> +
> /*
> * This is the RPC task struct
> */
> @@ -205,7 +216,8 @@ struct rpc_wait_queue {
> */
> struct rpc_task *rpc_new_task(const struct rpc_task_setup *);
> struct rpc_task *rpc_run_task(const struct rpc_task_setup *);
> -struct rpc_task *rpc_run_bc_task(struct rpc_rqst *req);
> +struct rpc_task *rpc_run_bc_task(struct rpc_rqst *req,
> + struct rpc_timeout *timeout);
> void rpc_put_task(struct rpc_task *);
> void rpc_put_task_async(struct rpc_task *);
> bool rpc_task_set_rpc_status(struct rpc_task *task, int rpc_status);
> diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
> index b10f987509cc..3331a1c2b47e 100644
> --- a/include/linux/sunrpc/svc.h
> +++ b/include/linux/sunrpc/svc.h
> @@ -250,6 +250,8 @@ struct svc_rqst {
> struct net *rq_bc_net; /* pointer to backchannel's
> * net namespace
> */
> + unsigned long bc_to_initval;
> + unsigned int bc_to_retries;
> void ** rq_lease_breaker; /* The v4 client breaking a lease */
> unsigned int rq_status_counter; /* RPC processing counter */
> };
> diff --git a/include/linux/sunrpc/xprt.h b/include/linux/sunrpc/xprt.h
> index f85d3a0daca2..464f6a9492ab 100644
> --- a/include/linux/sunrpc/xprt.h
> +++ b/include/linux/sunrpc/xprt.h
> @@ -30,17 +30,6 @@
> #define RPC_MAXCWND(xprt) ((xprt)->max_reqs << RPC_CWNDSHIFT)
> #define RPCXPRT_CONGESTED(xprt) ((xprt)->cong >= (xprt)->cwnd)
>
> -/*
> - * This describes a timeout strategy
> - */
> -struct rpc_timeout {
> - unsigned long to_initval, /* initial timeout */
> - to_maxval, /* max timeout */
> - to_increment; /* if !exponential */
> - unsigned int to_retries; /* max # of retries */
> - unsigned char to_exponential;
> -};
> -
> enum rpc_display_format_t {
> RPC_DISPLAY_ADDR = 0,
> RPC_DISPLAY_PORT,
> diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
> index daa9582ec861..886fc4c76558 100644
> --- a/net/sunrpc/clnt.c
> +++ b/net/sunrpc/clnt.c
> @@ -1302,8 +1302,10 @@ static void call_bc_encode(struct rpc_task *task);
> * rpc_run_bc_task - Allocate a new RPC task for backchannel use, then run
> * rpc_execute against it
> * @req: RPC request
> + * @timeout: timeout values to use for this task
> */
> -struct rpc_task *rpc_run_bc_task(struct rpc_rqst *req)
> +struct rpc_task *rpc_run_bc_task(struct rpc_rqst *req,
> + struct rpc_timeout *timeout)
> {
> struct rpc_task *task;
> struct rpc_task_setup task_setup_data = {
> @@ -1322,7 +1324,7 @@ struct rpc_task *rpc_run_bc_task(struct rpc_rqst *req)
> return task;
> }
>
> - xprt_init_bc_request(req, task);
> + xprt_init_bc_request(req, task, timeout);
>
> task->tk_action = call_bc_encode;
> atomic_inc(&task->tk_count);
> diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
> index 3f2ea7a0496f..3f714d33624b 100644
> --- a/net/sunrpc/svc.c
> +++ b/net/sunrpc/svc.c
> @@ -1557,6 +1557,7 @@ void svc_process_bc(struct rpc_rqst *req, struct svc_rqst *rqstp)
> {
> struct rpc_task *task;
> int proc_error;
> + struct rpc_timeout timeout;
>
> /* Build the svc_rqst used by the common processing routine */
> rqstp->rq_xid = req->rq_xid;
> @@ -1602,8 +1603,16 @@ void svc_process_bc(struct rpc_rqst *req, struct svc_rqst *rqstp)
> return;
> }
> /* Finally, send the reply synchronously */
> + if (rqstp->bc_to_initval > 0) {
> + timeout.to_initval = rqstp->bc_to_initval;
> + timeout.to_retries = rqstp->bc_to_initval;
> + } else {
> + timeout.to_initval = req->rq_xprt->timeout->to_initval;
> + timeout.to_initval = req->rq_xprt->timeout->to_retries;
> + }
> memcpy(&req->rq_snd_buf, &rqstp->rq_res, sizeof(req->rq_snd_buf));
> - task = rpc_run_bc_task(req);
> + task = rpc_run_bc_task(req, &timeout);
> +
> if (IS_ERR(task))
> return;
>
> diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
> index 6cc9ffac962d..af13fdfa6672 100644
> --- a/net/sunrpc/xprt.c
> +++ b/net/sunrpc/xprt.c
> @@ -1986,7 +1986,8 @@ void xprt_release(struct rpc_task *task)
>
> #ifdef CONFIG_SUNRPC_BACKCHANNEL
> void
> -xprt_init_bc_request(struct rpc_rqst *req, struct rpc_task *task)
> +xprt_init_bc_request(struct rpc_rqst *req, struct rpc_task *task,
> + const struct rpc_timeout *to)
> {
> struct xdr_buf *xbufp = &req->rq_snd_buf;
>
> @@ -1999,8 +2000,13 @@ xprt_init_bc_request(struct rpc_rqst *req, struct rpc_task *task)
> */
> xbufp->len = xbufp->head[0].iov_len + xbufp->page_len +
> xbufp->tail[0].iov_len;
> -
> - xprt_init_majortimeo(task, req, req->rq_xprt->timeout);
> + /*
> + * Backchannel Replies are sent with !RPC_TASK_SOFT and
> + * RPC_TASK_NO_RETRANS_TIMEOUT. The major timeout setting
> + * affects only how long each Reply waits to be sent when
> + * a transport connection cannot be established.
> + */
> + xprt_init_majortimeo(task, req, to);
> }
> #endif
>
> --
> 2.43.0
>
