Some deficiencies with the Samba VFS that I think need to e rectified ...

[Richard Sharpe]

On Wed, Dec 11, 2013 at 8:17 AM, Volker Lendecke
<Volker.Lendecke at sernet.de> wrote:
> On Wed, Dec 11, 2013 at 07:09:44AM -0800, Richard Sharpe wrote:
>> On Tue, Dec 10, 2013 at 10:14 PM, Volker Lendecke
>> <Volker.Lendecke at sernet.de> wrote:
>> > On Tue, Dec 10, 2013 at 02:33:52PM -0800, Richard Sharpe wrote:
>> >> 2. We are stitching together directories on different systems
>> >> separated by a WAN and need access to things like ACCESS_MASK and
>> >> SHARE_MODE in functions like SMB_VFS_MKDIR and some others but it is
>> >> not currently available, which constrains us mightily.
>> >
>> > This sounds like a cifs-proxy like thing. Or is it a file
>> > system that supports share modes and a "full" createfile
>> > call?
>>
>> Never go full createfile :-)
>>
>> In some sense it is a distributed file system that is starting to
>> support many of the required Windows semantics. See my email about the
>> usefulness of SHARING_VIOLATION from a couple of days ago as well.
>
> At least we are supposed to do an flock call that has the
> share mode parameters. That's what we are using with some
> proprietary cluster file system that has a set_share call
> which is compatible with NFSv4. NFSv4 share modes have their
> flaws and Samba integration might not be great due to
> semantic differences, but that does the job. Wouldn't that
> at least partially help you? See source3/smbd/open.c lines
> 2678ff.

OK, but to properly evaluate sharing conflicts you need both the
ACCESS_MASK and the SHARE_MODE at the time of evaluation ...

It is possible to do today but not clean.

-- 
Regards,
Richard Sharpe
(何以解憂？唯有杜康。--曹操)