Re: BUG in: Driver core: convert block from raw kobjects to core devices (fwd)

[James Bottomley <James.Bottomley@SteelEye.com>]

On Wed, 2007-10-31 at 12:44 -0400, Alan Stern wrote:
> On Wed, 31 Oct 2007, James Bottomley wrote:
> 
> > OK, light beginning to go on now.
> > 
> > The problem is that you've fallen into the conceptual trap we tried very
> > hard to avoid in the initial go around of joining SCSI upper layer
> > drivers to gendisks.  That's why no gendisk references are held by the
> > mid-layer, only by the entities that represent the objects created by
> > upper layer drivers.
> > 
> > Doesn't this circularity now exist for everything?  Every device that
> > creates a queue has a reference to the queue, every queue has a
> > reference to its attached gendisk and now every gendisk has a reference
> > to the device creating the queue?  This doesn't look to be a SCSI
> > specific problem.
> 
> It probably isn't.  I haven't looked at other subsystems but the 
> scenario you described could well be the common case.
> 
> Dropping a kobject's reference to its parent when the kobject is
> removed rather than when it is deleted will solve all these problems.  
> The queue's reference to the gendisk and the gendisk's reference to the
> device will both be dropped when the device is unregistered, allowing
> the device and gendisk to be released.  When the device is released it
> can then drop its reference to the queue, allowing the queue to be
> released.
> 
> In other words, reverting that 4-year-old patch would fix everything.  
> I'd still like to get your acknowledgement for my patch, posted here:
> 
> 	http://marc.info/?l=linux-scsi&m=119368368904151&w=2

It sounds plausible, but I'm not really up to speed on all the changes
going on in sysfs at the moment, so it would be dangerous to rely on
what I think sounds reasonable.

James