[pnfs] Server Sessions

[Iyer, Rahul]

Yes, we would. This value could be set in nfsd4_sequence and used in all
the pnfs operations.
Regards
Rahul


> -----Original Message-----
> From: William A. (Andy) Adamson [mailto:andros at citi.umich.edu] 
> Sent: Monday, January 29, 2007 2:32 PM
> To: Iyer, Rahul
> Cc: pnfs at linux-nfs.org
> Subject: Re: Server Sessions
> 
> won't we need the current_session for LAYOUT_GET, 
> LAYOUTCOMMIT, LAYOUT_RETURN?
> 
> -->Andy
> 
> 
> On 1/29/07, Iyer, Rahul < Rahul.Iyer at netapp.com 
> <mailto:Rahul.Iyer at netapp.com> > wrote:
> 
> 
> 
> 	> -----Original Message-----
> 	> From: William A. (Andy) Adamson [mailto: 
> andros at citi.umich.edu]
> 	> Sent: Monday, January 29, 2007 2:08 PM
> 	> To: Iyer, Rahul
> 	> Cc: pnfs at linux-nfs.org
> 	> Subject: Server Sessions 
> 	>
> 	> hi rahul
> 	>
> 	> i'm looking at the data servers use of sessions.
> 	>
> 	> do you have any code to create a 'current_session' in
> 	> fs/nfsd/nfs4proc.c?
> 	
> 	No, I have not. But it should be fairly simple... Much 
> like the putfh 
> 	code. nfsd4_sequence() will give you an idea of how to 
> get a nfs4_client
> 	struct given a sessionid.
> 	
> 	> have you written the clean-up code for __nfs4_state_shutdown?
> 	Nothing explicit. Because exchange_id and 
> create_session are similar in 
> 	state processing to setclientid and 
> setclientid_confirm. The additional
> 	part for sessions is that there is an extra hash table. 
> I do not do any
> 	other "state allocation" as there isn't any seqid 
> checking. The cleanup 
> 	for the hashtable stuff is in expire_client. So, I do 
> not believe
> 	anything extra is required in __nfs4_state_shutdown() 
> as it calls
> 	expire_client().
> 	
> 	Regards
> 	Rahul
> 	
> 	
> 
> 
>