Hi,

Microsoft is working on a fix. In the meantime I have compiled 3.0.28a for
RHEL4 (needs a tiny patch, attached), and this also removes the problem (i.e
it works then with SP2). MS support claims that it could be due to badly
aligned fields

>
> NT NOTIFY Parameters
>
>            Next Entry Offset: 54  (<-- not longWord aligned!)
>
>           Action: REMOVED (object was removed) (2)
>
 >           File Name Len: 42

One of the differences between 3.0.28a and 3.0.25 we and MS see is that
3.0.28 does not seem to chain responses any more. Next Entry Offset == 0
always. The suspicion is that if the NEO is not 32-bit aligned this causes a
Bugcheck in Windows. MS will follow this up further.

Regards,

Niko

On Sat, May 10, 2008 at 6:04 PM, Niko Neufeld <[EMAIL PROTECTED]>
wrote:

> Yes. We have reported this to MS. I'll report back when I hear from them.
> Since this does not happen, when the server is itself W2300 I thought they
> wouldn't be very motivated :-)
> What I find surprising though, is that nobody else seems to have seen this
> or a similar problem. AFAIK SP2 is out since quite some time.
>
> Niko
> On May 10, 2008, at 2:18 AM, Jeremy Allison wrote:
>
>  On Fri, May 09, 2008 at 07:01:57PM +0200, Niko Neufeld wrote:
>>
>>> Hi,
>>> [...]
>>> Removing SP2 solves the problem.
>>>
>>
>> This looks like a Microsoft bug in SP2. Have you reported it
>> to them ? They are much more proactive in working with us
>> now so it might be worth reporting it and letting them work
>> it out with us.
>>
>> Jeremy.
>>
>
>
--- smbd/dmapi.c	2008-05-16 10:10:50.000000000 +0200
+++ ../../../samba-3.0.28a/source/smbd/dmapi.c	2008-03-08 16:33:36.000000000 +0100
@@ -33,7 +33,6 @@
 #else /* USE_DMAPI */
 
 #ifdef HAVE_XFS_DMAPI_H
-typedef signed long __s64;
 #include <xfs/dmapi.h>
 #elif defined(HAVE_SYS_DMI_H)
 #include <sys/dmi.h>
-- 
To unsubscribe from this list go to the following URL and read the
instructions:  https://lists.samba.org/mailman/listinfo/samba

Reply via email to