[samba-bugs at samba.org: [Bug 10593] PANIC: assert failed at ../source3/smbd/open.c(1582): ret]

[Volker Lendecke]

Hi!

Under the attachment find a patchset that for me reproduces
and fixes the panic from bug 10593. The problem is the
following:

Client sends a create for an oplocked file. We defer the
request for the oplock break, watching the locking.tdb
record. Client after very few milliseconds cancels the
request. Oplock break response comes in via dbwrap_watch,
but the mid is already gone. We panic where we should not.

I've uploaded two fixes: The quick one just removes that
paranoia check for this case. The real one is a bit more
intrusive: It shuffles the talloc hierarchy for deferred
records such that the dbwrap_watch_send request goes away
when the deferred record is cancelled.

Neither version has survived autobuild yet, so take a look
at this with care. However, asking for comments already,
this is a bit subtle to me :-)

Thanks,

Volker

----- Forwarded message from samba-bugs at samba.org -----

Date: Fri, 20 Jun 2014 14:28:20 +0000
From: samba-bugs at samba.org
To: samba-qa at samba.org
Subject: [Bug 10593] PANIC: assert failed at ../source3/smbd/open.c(1582): ret

https://bugzilla.samba.org/show_bug.cgi?id=10593

--- Comment #4 from Volker Lendecke <vl at samba.org> 2014-06-20 14:28:20 UTC ---
Created attachment 10042
  --> https://bugzilla.samba.org/attachment.cgi?id=10042
patch for master

This is a reproducer for the bug including a fix for master. This fix might be
a little intrusive and will need some patch cosmetics, but I'd like to present
it already now for review.

-- 
Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.

----- End forwarded message -----

-- 
SerNet GmbH, Bahnhofsallee 1b, 37081 Göttingen
phone: +49-551-370000-0, fax: +49-551-370000-9
AG Göttingen, HRB 2816, GF: Dr. Johannes Loxen
http://www.sernet.de, mailto:kontakt at sernet.de