Re: Stable backports for qla2xxx target mode

[Greg KH <gregkh@linuxfoundation.org>]

On Fri, Aug 14, 2015 at 11:37:03AM -0700, Roland Dreier wrote:
> On Thu, Aug 13, 2015 at 5:49 PM, Greg KH <gregkh@linuxfoundation.org> wrote:
> > What exactly are they fixing?  They look like they add a ton of new
> > functions to the driver, and other features.  What is so broken in the
> > driver today that warrants this type of exception to the rules?
> 
> Let me start by saying that I'm not sure these changes really belong
> in stable, and I totally understand why you're pushing back.
> 
> However the fact is that upstream qla2xxx is utterly broken in target
> mode — trying to use Linux as a real FC target, where we have to deal
> with fabric changes, initiators going away, and so on, is hopeless
> without these changes.  The target will crash, or data will be
> corrupted, or similarly severe issues.
> 
> We were definitely derelict in getting our fixes upstream for a while,
> but they are there now.  So using a new kernel has a chance at
> working.  I think it's totally legitimate to say that the changes are
> too big for -stable, and someone who wants to use qla2xxx target mode
> needs a recent kernel or their own backports.  But as I said, the bugs
> are pretty severe (kernel crash or worse) so I could see pulling the
> changes into -stable also.

Ok, so given that it seems this code has always been broken in this mode
of operation, and the patches are huge, and they aren't even in a
normally released kernel, I'm not going to take them into -stable at
this point in time.

Feel free to resend them after about 6 months or so, after the fallout
of these major changes has made their way to the distros and users, and
you have it all working correctly, and I'll be glad to revisit them.

thanks,

greg k-h