2009/387 [Pathname Reparse Points]

[Alan.M.Wright]

Joerg Schilling wrote:
> Nicolas Williams <Nicolas.Williams at sun.com> wrote:
>
>   
>> On Fri, Jul 10, 2009 at 10:18:23AM -0700, Scott Rotondo wrote:
>>     
>>> Darren J Moffat wrote:
>>>       
>>>> This looks very cool but I haven't quite got my head around it 
>>>> completely yet.
>>>>
>>>> What happens if open(2) is called with O_NOFOLLOW set on one of these 
>>>> reparse points ? (Please answer for ZFS local access, NFS and CIFS).
>>>>         
>>> Since these reparse points are implemented with a special type of 
>>> symlink, open() with O_NOFOLLOW should fail with such an object.
>>>       
>> On the client side a server-side reparse point behaves like a mount
>> point, very much in the same way as mirror mounts.
>>
>> Locally (on the server) a reparse point is stored in a symlink, but it
>> isn't followed, and if it were then it'd behave like a mount, just as on
>> the client side.
>>
>> There's nothing quite like following a symlink here, therefore
>> O_NOFOLLOW shouldn't apply on the client side.
>>     
>
> If the feature is implemented as symlink, how will stat() vs. lstat() perform 
> on these objects? Will it confuse existing applications?
>   
They will behave exactly as they do today if you were to define
a symlink containing garbage text.

Alan

> Jörg
>
>