Message-ID: <4DBA63D0.6060104@panasas.com>
Date: Fri, 29 Apr 2011 10:08:00 +0300
From: Benny Halevy <bhalevy@panasas.com>
To: Dros Adamson <dros@netapp.com>
CC: trond@netapp.com, linux-nfs@vger.kernel.org
Subject: Re: [PATCH] NFS: move pnfs layouts to nfs_server structure
References: <1303944050-29932-1-git-send-email-dros@netapp.com> <1303944050-29932-2-git-send-email-dros@netapp.com> <4DB8EA18.1030508@panasas.com> <77373F14-888D-4817-9388-44DE4CD16607@netapp.com>
In-Reply-To: <77373F14-888D-4817-9388-44DE4CD16607@netapp.com>
Content-Type: text/plain; charset=ISO-8859-1
Sender: linux-nfs-owner@vger.kernel.org
MIME-Version: 1.0

On 2011-04-28 21:21, Dros Adamson wrote:
> 
> On Apr 28, 2011, at 12:16 AM, Benny Halevy wrote:
> 
>> On 2011-04-28 01:40, Weston Andros Adamson wrote:
>> ...
>>
>>> @@ -167,18 +176,27 @@ static u32 initiate_bulk_draining(struct nfs_client *clp,
>>> 	};
>>>
>>> 	spin_lock(&clp->cl_lock);
>>> -	list_for_each_entry(lo, &clp->cl_layouts, plh_layouts) {
>>> +	rcu_read_lock();
>>> +	list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) {
>>> 		if ((args->cbl_recall_type == RETURN_FSID) &&
>>> -		    memcmp(&NFS_SERVER(lo->plh_inode)->fsid,
>>> -			   &args->cbl_fsid, sizeof(struct nfs_fsid)))
>>> -			continue;
>>> -		if (!igrab(lo->plh_inode))
>>> +		    memcmp(&server->fsid, &args->cbl_fsid,
>>> +			   sizeof(struct nfs_fsid)))
>>> 			continue;
>>> -		get_layout_hdr(lo);
>>> -		BUG_ON(!list_empty(&lo->plh_bulk_recall));
>>> -		list_add(&lo->plh_bulk_recall, &recall_list);
>>> +
>>> +		list_for_each_entry(lo, &server->layouts, plh_layouts) {
>>> +			if (!igrab(lo->plh_inode))
>>> +				continue;
>>> +			get_layout_hdr(lo);
>>> +			BUG_ON(!list_empty(&lo->plh_bulk_recall));
>>> +			list_add(&lo->plh_bulk_recall, &recall_list);
>>> +		}
>>> +
>>> +		if (args->cbl_recall_type == RETURN_FSID)
>>> +			break;
>>
>> I actually thought of suggesting that yesterday but
>> I'm not sure fsid is unique per nfs_server.  Is it?
>>
>> Benny
> 
> Ah!  I incorrectly assumed that fsids are unique.  client.c:nfs_clone_server() proves otherwise.
> 
> Thanks!

NP. I'm glad we caught that in time :)

Benny

> 
> -dros
> 
>