[PATCH] CIFS: Prevent error log on spurious oplock break

[Steve French]

I lean toward reducing or skipping the logging of the 'normsl' (or at least
possible) race between close and oplock break.

I see this eg spamming the log running xfstest 524

Can you repro it as well running that?

On Tue, Mar 9, 2021, 07:42 Vincent Whitchurch via samba-technical <
samba-technical at lists.samba.org> wrote:

> On Tue, Mar 09, 2021 at 01:05:11AM +0100, ronnie sahlberg wrote:
> > On Sun, Mar 7, 2021 at 8:52 PM Shyam Prasad N via samba-technical
> > <samba-technical at lists.samba.org> wrote:
> > > The reason for rejecting the request maybe a number of things like:
> > > corrupted request, stale request (for some old session), or for a
> > > wrong handle.
> > > I don't think we should treat any of these cases as a success.
> >
> > I agree with Shyam here.
> > We shouldn't change the return value to pretend success just to
> > suppress a warning.
>
> Thank you all for your comments.  I see that everyone agrees that the
> error print is useful for SMB2, so I will drop this patch.
>
> > However, if it is common to trigger with false positives we might want
> > to something to prevent it from
> > spamming the logs.
> > These messages could be useful if we encounter bugs in our leasing
> > code, or bugs in server
> > lease code, so we should't throw them away completely. But if false
> > positives are common ...
> >
> > Some thoughts I and Stever brainstormed about could be to change the
> code in the
> > demiltiplex thread where we currently dump the packets that were
> "invalid"
> > to maybe:
> > *  log once as VFS and then log any future ones as FYI
> > * log once as VFS and then only make the others available via dynamic
> > trace points
> > * rate limit it so we only log it once every n minutes?  (this is
> overkill?)
>
> Thank you for the suggestions.  In my case, I've only received some
> reports of this error being emitted very rarely (couple of times a month
> in our stability tests).  Right now it looks like the problem may only
> be with a particular NAS, and we're looking into triggering oplock
> breaks more often and catching the problem with some more logging.
>
>