On Sat, 11 Aug 2007 16:00:53 +0200
"Kay Sievers" [off-list ref] wrote:

On 8/8/07, Kristen Carlson Accardi [off-list ref] wrote:

quoted

Here is an updated set of patches that implement Asynchronous Notification
support for ATAPI devices.  In this version I no longer export the AN
capability through genhd, and the uevent is sent by the scsi_device
instead of gendisk.  I added a generic SCSI device event mechanism so
that it can be expanded in the future with other types of events.  Please
let me know what you think.

Does that mean:
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=8ce7ad7b2d11fae2c3d285a6a0caea9322c0b8fc

will go away?

Yes - this method was rejected by James, so I had to change away from
using the genhd to transmit events.

Then we will need to come up with an idea to propagate that to the
blockdev, it gets a bit nasty for userspace to find out which
blockdevice it should update. It's certainly possible, but just not as
easy as it is now, because userspace just doesn't really care in most
cases what kind of bus device a block device is coming from.

Is it possible to use the linkage that is in the scsi_device directory
of the disk that is sending the event to find the relevant block devices?
For example, /sys/class/scsi_device/0:0:0:0/device/block:sda
on my system tells me which block devices belong to this device.