[pnfs] [PATCH 1/12] more ifdef CONFIG_PNFS for pnfs only code

[Benny Halevy]

J. Bruce Fields wrote:
> On Mon, May 07, 2007 at 10:32:02PM +0300, Benny Halevy wrote:
>> @@ -1182,6 +1184,7 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
>>  		case OP_RELEASE_LOCKOWNER:
>>  			op->status = nfsd4_release_lockowner(rqstp, &op->u.release_lockowner);
>>  			break;
>> +#ifdef CONFIG_PNFS
>>  		case OP_GETDEVICELIST:
>>  			op->status = nfsd4_getdevlist(rqstp, current_fh, &op->u.pnfs_getdevlist);
>>  			break;
>> @@ -1197,6 +1200,7 @@ nfsd4_proc_compound(struct svc_rqst *rqstp,
>>  		case OP_LAYOUTRETURN:
>>  			op->status = nfsd4_layoutreturn(rqstp, current_ses, current_fh, &op->u.pnfs_layoutreturn);
>>  			break;
>> +#endif
> 
> Might want to define these operations as stubs that just return
> nfs4err_opnotsupp (or whatever it is) in the non-PNFS case.  That might
> also port better to more recent kernels (where I replace this big switch
> by a function call table.  Hope that's OK....)

Absolutely.

> 
> Is the current plan to use both nfsdv4.1 and pnfs config options, or
> include 4.1 by default and only make PNFS optional?

I'm not sure what the plan is.
Since 4.1 is still experimental I think it makes sense to
use both config options so that v4.0 users can protect themselves
from the bleeding edge as much as possible.

> 
> --b.