From: Jeff Layton <jlayton@kernel.org>
To: Chuck Lever <cel@kernel.org>, NeilBrown <neil@brown.name>,
Olga Kornievskaia <okorniev@redhat.com>,
Dai Ngo <Dai.Ngo@oracle.com>, Tom Talpey <tom@talpey.com>,
Trond Myklebust <trondmy@kernel.org>,
Anna Schumaker <anna@kernel.org>,
"David S. Miller" <davem@davemloft.net>,
Eric Dumazet <edumazet@google.com>,
Jakub Kicinski <kuba@kernel.org>,
Paolo Abeni <pabeni@redhat.com>, Simon Horman <horms@kernel.org>,
Shuah Khan <shuah@kernel.org>
Cc: Slawomir Stepien <sst@poczta.fm>,
linux-nfs@vger.kernel.org, linux-kernel@vger.kernel.org,
netdev@vger.kernel.org, Trond Myklebust <trondmy@gmail.com>,
linux-kselftest@vger.kernel.org,
Jeff Layton <jlayton@kernel.org>
Subject: [PATCH v5 06/15] SUNRPC: report local rpcbind calls that get no answer
Date: Tue, 01 Sep 2026 09:39:44 -0400 [thread overview]
Message-ID: <20260901-nfsd-nl-hang-v5-6-a540d86668b0@kernel.org> (raw)
In-Reply-To: <20260901-nfsd-nl-hang-v5-0-a540d86668b0@kernel.org>
A caller that creates many listeners in one operation calls svc_register()
once for each of them. Every call waits for the local rpcbind on its own,
so a rpcbind that never answers costs the caller one timeout per listener.
The caller has no way to learn that the first call already failed.
Split rpcb_register_call() failures by whether rpcbind answered:
- answered: a FALSE reply, or an RPC-level rejection that
rpc_decode_header() derives from the reply. -EACCES (FALSE reply or
AUTH_ERROR), -EPROTONOSUPPORT, -EPFNOSUPPORT, -EOPNOTSUPP.
- never sent: -ENOMEM, -EMSGSIZE, -ERESTARTSYS.
- no answer: everything else, i.e. transport errors.
Any no-answer error gets represented by -EIO, which is already what the
RPC layer reports for most of it: rpc_check_timeout() returns -EIO for a
soft timeout without RPC_TASK_TIMEOUT, and call_status() documents -EIO
as "shutdown or soft timeout".
Keep a count of the number of rpcbind failures in the serv. Later
patches will use that to watch for hard rpcbind failures, and alter
their behavior accordingly.
Assisted-by: LLM
Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
include/linux/sunrpc/clnt.h | 3 ++-
include/linux/sunrpc/svc.h | 7 +++++--
net/sunrpc/rpcb_clnt.c | 22 +++++++++++++++++++---
net/sunrpc/svc.c | 42 +++++++++++++++++++++++++++++++++++++++++-
4 files changed, 67 insertions(+), 7 deletions(-)
diff --git a/include/linux/sunrpc/clnt.h b/include/linux/sunrpc/clnt.h
index 3c2b8c355ab3..30344c0d6a9d 100644
--- a/include/linux/sunrpc/clnt.h
+++ b/include/linux/sunrpc/clnt.h
@@ -199,7 +199,8 @@ struct rpc_xprt *rpc_task_get_xprt(struct rpc_clnt *clnt,
int rpcb_create_local(struct net *);
void rpcb_put_local(struct net *);
-int rpcb_register(struct net *, u32, u32, int, unsigned short);
+int rpcb_register(struct net *net, u32 prog, u32 vers, int prot,
+ unsigned short port);
int rpcb_v4_register(struct net *net, const u32 program,
const u32 version,
const struct sockaddr *address,
diff --git a/include/linux/sunrpc/svc.h b/include/linux/sunrpc/svc.h
index 2db1b9ec5658..5fa9417e034d 100644
--- a/include/linux/sunrpc/svc.h
+++ b/include/linux/sunrpc/svc.h
@@ -78,6 +78,7 @@ struct svc_serv {
unsigned int sv_max_payload; /* datagram payload size */
unsigned int sv_max_mesg; /* max_payload + 1 page for overheads */
unsigned int sv_xdrsize; /* XDR buffer size */
+ atomic_t sv_rpcb_failures; /* unanswered rpcbind calls */
struct list_head sv_permsocks; /* all permanent sockets */
struct list_head sv_tempsocks; /* all temporary sockets */
int sv_tmpcnt; /* count of temporary "valid" sockets */
@@ -451,6 +452,7 @@ int sunrpc_set_pool_mode(const char *val);
int sunrpc_get_pool_mode(char *val, size_t size);
void svc_rpcb_cleanup(struct svc_serv *serv, struct net *net);
int svc_bind(struct svc_serv *serv, struct net *net);
+unsigned int svc_rpcb_failure_count(struct svc_serv *serv);
struct svc_serv *svc_create(struct svc_program *, unsigned int,
int (*threadfn)(void *data));
bool svc_rqst_replace_page(struct svc_rqst *rqstp,
@@ -471,8 +473,9 @@ unsigned int svc_serv_maxthreads(const struct svc_serv *serv);
int svc_pool_stats_open(struct svc_info *si, struct file *file);
void svc_process(struct svc_rqst *rqstp);
void svc_process_bc(struct rpc_rqst *req, struct svc_rqst *rqstp);
-int svc_register(const struct svc_serv *, struct net *, const int,
- const unsigned short, const unsigned short);
+int svc_register(struct svc_serv *serv, struct net *net,
+ const int family, const unsigned short proto,
+ const unsigned short port);
void svc_wake_up(struct svc_serv *);
void svc_reserve(struct svc_rqst *rqstp, int space);
diff --git a/net/sunrpc/rpcb_clnt.c b/net/sunrpc/rpcb_clnt.c
index 0aa376b82a52..7255c1e07eec 100644
--- a/net/sunrpc/rpcb_clnt.c
+++ b/net/sunrpc/rpcb_clnt.c
@@ -412,7 +412,8 @@ static struct rpc_clnt *rpcb_create(struct net *net, const char *nodename,
return rpc_create(&args);
}
-static int rpcb_register_call(struct sunrpc_net *sn, struct rpc_clnt *clnt, struct rpc_message *msg, bool is_set)
+static int rpcb_register_call(struct sunrpc_net *sn, struct rpc_clnt *clnt,
+ struct rpc_message *msg, bool is_set)
{
int flags = RPC_TASK_NOCONNECT;
int error, result = 0;
@@ -422,8 +423,23 @@ static int rpcb_register_call(struct sunrpc_net *sn, struct rpc_clnt *clnt, stru
msg->rpc_resp = &result;
error = rpc_call_sync(clnt, msg, flags);
- if (error < 0)
- return error;
+ if (error < 0) {
+ switch (error) {
+ /* rpcbind answered; the reply itself carries the error */
+ case -EPROTONOSUPPORT:
+ case -EPFNOSUPPORT:
+ case -EOPNOTSUPP:
+ case -EACCES:
+ /* the call never made it onto the wire */
+ case -ENOMEM:
+ case -EMSGSIZE:
+ /* the caller is going away; this says nothing about rpcbind */
+ case -ERESTARTSYS:
+ return error;
+ }
+ /* anything else, we assume that rpcbind isn't functional */
+ return -EIO;
+ }
if (!result)
return -EACCES;
diff --git a/net/sunrpc/svc.c b/net/sunrpc/svc.c
index 4f402bbf97ba..ca6f90653327 100644
--- a/net/sunrpc/svc.c
+++ b/net/sunrpc/svc.c
@@ -1179,10 +1179,40 @@ int svc_generic_rpcbind_set(struct net *net,
error = svc_rpcbind_set_version(net, progp, version,
family, proto, port);
+ /* -EIO means no answer, not a refusal, so vs_rpcb_optnl must keep it. */
+ if (error == -EIO)
+ return error;
+
return (vers->vs_rpcb_optnl) ? 0 : error;
}
EXPORT_SYMBOL_GPL(svc_generic_rpcbind_set);
+/**
+ * svc_rpcb_failure_count - local rpcbind calls for @serv that got no answer
+ * @serv: RPC service to query
+ *
+ * svc_register() adds one for each of its calls that got no answer. A reply
+ * that refuses one entry does not count, because rpcbind answered and the
+ * next entry may still succeed.
+ *
+ * The count is kept per serv rather than per net. The local rpcbind client
+ * is per-net and lockd shares it, but a count that another service can move
+ * says nothing about this serv's own calls.
+ *
+ * This is for callers that cannot see the svc_register() return, because a
+ * transport class sits in between. Such a caller reads the count before it
+ * starts and compares as it goes, so there is no state to reset between
+ * operations. The count never resets, and callers must not attach meaning
+ * to the value itself.
+ *
+ * Return: the number of unanswered calls since this serv was created.
+ */
+unsigned int svc_rpcb_failure_count(struct svc_serv *serv)
+{
+ return atomic_read(&serv->sv_rpcb_failures);
+}
+EXPORT_SYMBOL_GPL(svc_rpcb_failure_count);
+
/**
* svc_register - register an RPC service with the local portmapper
* @serv: svc_serv struct for the service to register
@@ -1193,10 +1223,11 @@ EXPORT_SYMBOL_GPL(svc_generic_rpcbind_set);
*
* Service is registered for any address in the passed-in protocol family
*/
-int svc_register(const struct svc_serv *serv, struct net *net,
+int svc_register(struct svc_serv *serv, struct net *net,
const int family, const unsigned short proto,
const unsigned short port)
{
+ bool noanswer = false;
unsigned int p, i;
int error = 0;
@@ -1208,10 +1239,16 @@ int svc_register(const struct svc_serv *serv, struct net *net,
struct svc_program *progp = &serv->sv_programs[p];
for (i = 0; i < progp->pg_nvers; i++) {
+ const struct svc_version *vers = progp->pg_vers[i];
int ret;
ret = progp->pg_rpcbind_set(net, progp, i,
family, proto, port);
+ if (ret == -EIO) {
+ noanswer = true;
+ if (vers && vers->vs_rpcb_optnl)
+ ret = 0;
+ }
if (ret < 0) {
printk(KERN_WARNING "svc: failed to register "
"%sv%u RPC service (errno %d).\n",
@@ -1223,6 +1260,9 @@ int svc_register(const struct svc_serv *serv, struct net *net,
}
}
+ if (noanswer)
+ atomic_inc(&serv->sv_rpcb_failures);
+
return error;
}
--
2.55.0
next prev parent reply other threads:[~2026-09-01 13:40 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-01 13:39 [PATCH v5 00/15] nfsd/sunrpc: harden the netlink listener set interface Jeff Layton
2026-09-01 13:39 ` [PATCH v5 01/15] NFSD: cap the number of listeners accepted in listener_set Jeff Layton
2026-09-01 13:39 ` [PATCH v5 02/15] NFSD: validate transport name in listener_set before serv creation Jeff Layton
2026-09-01 13:39 ` [PATCH v5 03/15] SUNRPC: keep the first error in svc_register() Jeff Layton
2026-09-01 13:39 ` [PATCH v5 04/15] SUNRPC: bound the local rpcbind client timeout to 1s Jeff Layton
2026-09-01 13:39 ` [PATCH v5 05/15] NFSD: report listener creation failures through extack Jeff Layton
2026-09-01 13:39 ` Jeff Layton [this message]
2026-09-01 13:39 ` [PATCH v5 07/15] SUNRPC: stop svc_register() once rpcbind stops answering Jeff Layton
2026-09-01 13:39 ` [PATCH v5 08/15] SUNRPC: stop the svc_unregister() sweep " Jeff Layton
2026-09-01 13:39 ` [PATCH v5 09/15] SUNRPC: stop unregistering listeners " Jeff Layton
2026-09-01 13:39 ` [PATCH v5 10/15] NFSD: stop registering with rpcbind after a failure in listener_set Jeff Layton
2026-09-01 13:39 ` [PATCH v5 11/15] SUNRPC: check rpc_sockaddr2uaddr() for failure when registering Jeff Layton
2026-09-01 13:39 ` [PATCH v5 12/15] selftests/nfsd: exercise listener_set request validation Jeff Layton
2026-09-01 13:39 ` [PATCH v5 13/15] selftests/nfsd: add a per-netns rpcbind stub and the listener round-trips Jeff Layton
2026-09-01 13:39 ` [PATCH v5 14/15] selftests/nfsd: check that listener_set asks rpcbind once Jeff Layton
2026-09-01 13:39 ` [PATCH v5 15/15] selftests/nfsd: check that listener removal " Jeff Layton
2026-09-02 13:59 ` [PATCH v5 00/15] nfsd/sunrpc: harden the netlink listener set interface Chuck Lever
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260901-nfsd-nl-hang-v5-6-a540d86668b0@kernel.org \
--to=jlayton@kernel.org \
--cc=Dai.Ngo@oracle.com \
--cc=anna@kernel.org \
--cc=cel@kernel.org \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=horms@kernel.org \
--cc=kuba@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=linux-nfs@vger.kernel.org \
--cc=neil@brown.name \
--cc=netdev@vger.kernel.org \
--cc=okorniev@redhat.com \
--cc=pabeni@redhat.com \
--cc=shuah@kernel.org \
--cc=sst@poczta.fm \
--cc=tom@talpey.com \
--cc=trondmy@gmail.com \
--cc=trondmy@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox