Re: [PATCH RFC 21/21] NFSD: Add tracepoints to observe clientID activity

[Chuck Lever III <chuck.lever@oracle.com>]

> On May 11, 2021, at 1:38 PM, Bruce Fields <bfields@fieldses.org> wrote:
> 
> On Tue, May 11, 2021 at 03:59:00PM +0000, Chuck Lever III wrote:
>> As Dave reported yesterday, this patch is unfinished and is probably
>> junk. But any thoughts on how the tracepoints should be organized
>> in this code would help.
>> 
>> So I was thinking we probably want a tracepoint to fire for each
>> case that is handled in this code (and in nfsd4_exchangeid).
>> However, this comment in nfsd4_setclientid:
>> 
>>   /* Cases below refer to rfc 3530 section 14.2.33: */
>> 
>> Is confusing.
>> 
>> - RFC 3530 is superceded by RFC 7530, and the section numbers have changed.
>> 
>> - The cases in this section in both RFCs aren't numbered, they are
>> bullet points.
> 
> Honestly I think those particular comments should just go.  The code
> doesn't even follow those bullet points very closely any more.

OK, I'll squash this in, and include similar changes to
nfsd4_setclientid_confirm().


> --b.
> 
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index f47f72bc871f..2aa5d15b08ed 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -3954,11 +3954,9 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
> 	new = create_client(clname, rqstp, &clverifier);
> 	if (new == NULL)
> 		return nfserr_jukebox;
> -	/* Cases below refer to rfc 3530 section 14.2.33: */
> 	spin_lock(&nn->client_lock);
> 	conf = find_confirmed_client_by_name(&clname, nn);
> 	if (conf && client_has_state(conf)) {
> -		/* case 0: */
> 		status = nfserr_clid_inuse;
> 		if (clp_used_exchangeid(conf))
> 			goto out;
> @@ -3970,7 +3968,6 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
> 	unconf = find_unconfirmed_client_by_name(&clname, nn);
> 	if (unconf)
> 		unhash_client_locked(unconf);
> -	/* We need to handle only case 1: probable callback update */
> 	if (conf && same_verf(&conf->cl_verifier, &clverifier)) {
> 		copy_clid(new, conf);
> 		gen_confirm(new, nn);

--
Chuck Lever