renames over SMB2

[Jeremy Allison]

On Tue, Feb 07, 2012 at 06:47:22PM +0100, Christian Ambach wrote:
> Hi Jeremy,
> 
> I was looking at the fix for Bug #8412 and was wondering if the fix
> is really correct. I would have expected that Samba needs to check
> if any other opener of the parent directory had not set
> FILE_SHARE_DELETE instead of looking at the access_mask that was
> used to open the parent directory.
> 
> So I wrote a set of rename specific tests that all pass against
> Win2k8r2 but some of them fail against Samba 3.6.3 (didn't check
> master yet). I hope that at least one of them mimics the Word saving
> issue that was reported as BSO 8412.
> 
> Would you have a look at my torture test to see if they make sense?
> If they do, we'll need to check why Samba fails on some of them.

Did you read this comment on #8412 ?

"Looking at the share mode record give 100% correctness, but at a
correspondingly higher cost per rename. I'm ok with this for now, as it covers
most of the actual usage. You'd need to have 2 separate clients doing
operations in the same directory - where one of them has a directory open for
delete (something a Windows client never does to my knowledge) in order to run
into this actual case."

I'm happy with any expanded test coverage, but we may have
to judge if we want to take the extra cost it will take to
get a 100% pass. Remember, in order to get 100% compatibility
you need to open and check the share mode on every directory
in every path traversal. Previously we've decided not to take
that cost.

Jeremy.