SUNRPC: Improve observability in svc_tcp_accept()
authorChuck Lever <chuck.lever@oracle.com>
Mon, 15 May 2023 13:32:53 +0000 (09:32 -0400)
committerChuck Lever <chuck.lever@oracle.com>
Mon, 5 Jun 2023 13:01:42 +0000 (09:01 -0400)
The -ENOMEM arm could fire repeatedly if the system runs low on
memory, so remove it.

Don't bother to trace -EAGAIN error events, since those fire after
a listener is created (with no work done) and once again after an
accept has been handled successfully (again, with no work done).

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
net/sunrpc/svcsock.c

index a2d0bf4..fbe33c9 100644 (file)
@@ -886,13 +886,8 @@ static struct svc_xprt *svc_tcp_accept(struct svc_xprt *xprt)
        clear_bit(XPT_CONN, &svsk->sk_xprt.xpt_flags);
        err = kernel_accept(sock, &newsock, O_NONBLOCK);
        if (err < 0) {
-               if (err == -ENOMEM)
-                       printk(KERN_WARNING "%s: no more sockets!\n",
-                              serv->sv_name);
-               else if (err != -EAGAIN)
-                       net_warn_ratelimited("%s: accept failed (err %d)!\n",
-                                            serv->sv_name, -err);
-               trace_svcsock_accept_err(xprt, serv->sv_name, err);
+               if (err != -EAGAIN)
+                       trace_svcsock_accept_err(xprt, serv->sv_name, err);
                return NULL;
        }
        if (IS_ERR(sock_alloc_file(newsock, O_NONBLOCK, NULL)))