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

[Alan Stern <stern@rowland.harvard.edu>]

On Tue, 23 Oct 2007, Kay Sievers wrote:

> > Yes; I haven't been able to figure out why we get different results.
> > 
> > > There must be something going wrong with the block patch in conjunction
> > > with the crazy SCSI release logic.

I found out why it works on my system -- which still leaves open the 
question of why it fails on yours.

Although the gendisk device is a child of the scsi_device, nevertheless 
disk->kobj is not a child of the scsi_device's embedded kobject.  
Instead it is a child of the static (!) block_depr kobject, which is 
defined in block/genhd.c.  Hence the disk's single reference to the 
scsi_device is indeed dropped when the disk is unregistered, which 
breaks the loop of circular references.

It works this way because I have CONFIG_SYSFS_DEPRECATED set; perhaps
you don't.  If you compare the two versions of get_device_parent() in
drivers/base/core.c you'll see the difference (it's the dev_type ==
&disk_type case).

Alan Stern

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/