[PATCH 0/5] Dynamic Pseudo Root

[Gabriel Barazer]

On 02/19/2008 1:27:53 AM +0100, Trond Myklebust 
<Trond.Myklebust at netapp.com> wrote:
> On Mon, 2008-02-18 at 19:01 -0500, Trond Myklebust wrote:
>>>    I suspect that our NFS server doesn't currently flag TMPFS
>>>    filehandles as being volatile.  If it doesn't, it really should, no
>>>    matter what the outcome of the above.  ditto for FAT filehandles
>>>    and probably others.
>> Why? TMPFS files never survive a reboot anyway, and otherwise they are
>> quite stable for the duration of their lifetime.
>>
>> The problem we have is matching TMPFS semantics to the semantics of a
>> pseudo-fs that demands changing.
> 
> Sorry, that didn't come out right.
> 
> The problem your scheme has that pseudo-fs semantics really require the
> ability to re-create the same _filehandles_ upon reboot in order to work
> correctly with existing clients.

Except if that particular pseudo-fs is capable of creating the same 
filehandles, because it is designed to do so (by not using inode for 
creating filehandles, or by generating inode numbers with a specific 
algorithm). This is not possible with the generic tmpfs and this doesn't 
require using volatiles FH.

Gabriel