[pnfs] [Fwd: [PATCH] update exchange id reply to draft 9]

[Marc Eshel]

"J. Bruce Fields" <bfields at fieldses.org> wrote on 03/04/2007 10:07:47 PM:

> On Sun, Mar 04, 2007 at 07:36:46PM -0800, Marc Eshel wrote:
> > From: Marc Eshel <eshel at almaden.ibm.com>
> > 
> > Had to reduce NFS4_OPAQUE_LIMIT from 1024 to 256 since few structs the 

> > include it are defined on the stack. I am note sure why it needs to be 

> > 1024 but if it must we will have to malloc those structs to avoid 
stack 
> > overflow.
> 
> The NFS4_OPAQUE_LIMIT is defined as 1024 by the rfc, and it's the
> maximum length of client identifiers and stateowners.  So in the
> (probably unlikely) event that a client sent us e.g. a client identifier
> of length more then 256 but less than 1024 and we rejected it for being
> too long, we'd fail to interoperate and it'd be our fault.

yes, I know we will need to support 1024 if the rfc can not changed. Even 
if we solve the stack issue I wouln't want to support 1024 long names. 
Some of the struct have 4 of them and and we have those stuct for each MDS 
and DS.
 
> In any case, objects on the stack of size 256 bytes are still pushing
> it, so the solution is definitely to get those off the stack.  What are
> those structs exactly?
> 
static int nfs4_proc_exchange_id(struct nfs4_client *clp)
{
        nfs4_verifier verifier;
        struct nfs41_exchange_id_args args;
        struct nfs41_exchange_id_res res = {

struct nfs_impl_id4 {
        u32             domain_len;
        char            domain[NFS4_OPAQUE_LIMIT];
        u32             name_len;
        char            name[NFS4_OPAQUE_LIMIT];
        struct nfstime4 date;
};
struct nfs41_exchange_id_args {
        nfs4_verifier *                 verifier;
        unsigned int                    id_len;
        char                            id[NFS4_OPAQUE_LIMIT];
        u32                             flags;
        struct nfs_impl_id4             impl_id;
        clientid4                       clientid;
        u32                             seqid;
};