Based on the comments, I'd say you're correct. However, only PSEUDO and
GPFS modify the buffer in any way other than setting length in
extract_handle(), and those 2 only byte swap them. All the protocols
get a handle by running the pair extract_handle() create_handle(). In
the majority of FS
Take for example nfs3_FhandleToCache(). It gets wire handle and calls
extract_handle() method. What is the job of extract_handle method?
Assuming that it is supposed to return "key" handle as the comments
there say. Then nfs3_FhandleToCache calls create_handle() with the
"key" handle. This is unex
> On 04/10/2017 11:57 AM, Frank Filz wrote:
> >> Hi All, there is usually a 1:1 relationship (ignoring handles across
> > architectures
> >> and versions) between nfs handle and object handle. One thing that is
> >> not clear is the "key" which is used for hashing the objects
> > (mdcache_entry_t).
On 04/10/2017 11:57 AM, Frank Filz wrote:
>> Hi All, there is usually a 1:1 relationship (ignoring handles across
> architectures
>> and versions) between nfs handle and object handle. One thing that is not
>> clear is the "key" which is used for hashing the objects
> (mdcache_entry_t).
>> Ganesha
> Hi All, there is usually a 1:1 relationship (ignoring handles across
architectures
> and versions) between nfs handle and object handle. One thing that is not
> clear is the "key" which is used for hashing the objects
(mdcache_entry_t).
> Ganesha 2.5 has handle_to_key() method to take unique bits
