Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   "Benjamin Coddington" <bcodding@redhat.com>
To:     bfields@fieldses.org, chuck.lever@oracle.com
Cc:     linux-nfs@vger.kernel.org
Subject: Re: [PATCH] NFSD: Keep existing listners on portlist error
Date:   Wed, 06 Oct 2021 10:22:20 -0400
Message-ID: <7A49A6B0-8E89-4123-9A2D-71FCC4B6EEED@redhat.com>
In-Reply-To: <45b916f1aa3fb7c059a574f61188a8f2f615410e.1633529847.git.bcodding@redhat.com>
References: <45b916f1aa3fb7c059a574f61188a8f2f615410e.1633529847.git.bcodding@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; format=flowed
Precedence: bulk

On 6 Oct 2021, at 10:18, Benjamin Coddington wrote:

> If nfsd has existing listening sockets without any processes, then an 
> error
> returned from svc_create_xprt() for an additional transport will 
> remove
> those existing listeners.  We're seeing this in practice when 
> userspace
> attempts to create rpcrdma transports without having the rpcrdma 
> modules
> present before creating nfsd kernel processes.  Fix this by checking 
> for
> existing sockets before callingn nfsd_destroy().
>
> Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
> ---
>  fs/nfsd/nfsctl.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
> index c2c3d9077dc5..df4613a4924c 100644
> --- a/fs/nfsd/nfsctl.c
> +++ b/fs/nfsd/nfsctl.c
> @@ -793,7 +793,10 @@ static ssize_t __write_ports_addxprt(char *buf, 
> struct net *net, const struct cr
>  		svc_xprt_put(xprt);
>  	}
>  out_err:
> -	nfsd_destroy(net);
> +	if (list_empty(&nn->nfsd_serv->sv_permsocks))
> +		nfsd_destroy(net);
> +	else
> +		nn->nfsd_serv->sv_nrthreads--;

Eh -- ignore this one, needs a v2 since we might decrement sv_nrthreads 
twice for the INET6 case..

Ben