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;
Date:   Tue, 18 May 2021 13:04:56 -0400
From:   "J. Bruce Fields" <bfields@fieldses.org>
To:     Dai Ngo <dai.ngo@oracle.com>
Cc:     olga.kornievskaia@gmail.com, linux-nfs@vger.kernel.org,
        trondmy@hammerspace.com, chuck.lever@oracle.com
Subject: Re: [PATCH v5 1/2] NFSD: delay unmount source's export after
 inter-server copy completed.
Message-ID: <20210518170456.GA25205@fieldses.org>
References: <20210517224330.9201-1-dai.ngo@oracle.com>
 <20210517224330.9201-2-dai.ngo@oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20210517224330.9201-2-dai.ngo@oracle.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
Precedence: bulk

On Mon, May 17, 2021 at 06:43:29PM -0400, Dai Ngo wrote:
> +struct nfsd4_ssc_umount;
>  
>  enum {
>  	/* cache misses due only to checksum comparison failures */
> @@ -176,6 +177,10 @@ struct nfsd_net {
>  	unsigned int             longest_chain_cachesize;
>  
>  	struct shrinker		nfsd_reply_cache_shrinker;
> +
> +	spinlock_t              nfsd_ssc_lock;
> +	struct nfsd4_ssc_umount	*nfsd_ssc_umount;
...

> +void nfsd4_ssc_init_umount_work(struct nfsd_net *nn)
> +{
> +	nn->nfsd_ssc_umount = kzalloc(sizeof(struct nfsd4_ssc_umount),
> +					GFP_KERNEL);
> +	if (!nn->nfsd_ssc_umount)
> +		return;

Is there any reason this needs to be allocated dynamically?  Let's just
embed it in nfsd_net.

Actually, I'm not convinced the separate structure definition's really
that helpful:

> +struct nfsd4_ssc_umount {
> +	struct list_head nsu_list;
> +	unsigned long nsu_expire;
> +	wait_queue_head_t nsu_waitq;
> +};

How about just:

	struct nfsd_net {
	...
	/* tracking server-to-server copy mounts: */
	spinlock_t		nfsd_ssc_lock;
	struct list_head	nfsd_ssc_mount_list;
	unsigned long		nfsd_ssc_mount_expire;
	wait_queeu_head_t	nfsd_ssc_mount_waitq;

or something along those lines?

--b.