Based on the discussion we have had so far on xl disk configuration handling, I have made a fair amount of implementation changes.  As most of you might have suspected the changes are vast and has ripple effect all over the place due to interface changes.  But I think I got the go ahead from the maintainers to make these changes for 4.1 and I also believe we are better off making these changes for 4.1.  The plan is to send the patches in following order -

1) Patch 1/3 - Interface changes, parsing changes, frontend/endback specific disk parameters setup and other ripple effects caused by interface changes.  Ideally, I would like to keep this patch to just the interface and parsing changes but by doing so I might break the rest of the code from even compiling.  So, this change is going to be huge.
2) Patch 2/3 - Disk validation changes
3) Patch 3/3 - "xl Disk Configuration option" documentation

Let me know if you would prefer a different order for convenience or for some other reason.

I have attached the changes I have made so far for you to get an idea of the level of changes coming.  Note that this is NOT a patch submission.

Here are the parts of the patch I would like to get reviewed -

1) Interface changes - libxl.idl, libxl.h
2) Parsing changes - xl_cmdimpl.c.  I have basically divided each disk configuration information into three chunks - pdev, vdev and attribute and parsed it based on our discussion/documentation.  In specific, parse_disk_config, parse_disk_attrib, parse_disk_vdev_info and parse_disk_pdev_info implementation is the core of parsing code and that is what I would like to get reviewed to be sure I am not way off from what you would be willing to accept.

What not to review in this patch -

Anything other than interface and parsing changes, I would request that you ignore for now.  I am sending it only for the following reasons -

1) Since we are very close to 4.1, I would like you to get an idea of all the places this change touches.
2) I would want the patch to build!

Other than that, there are parts of the patch outside interface/parsing changes that might be incomplete or even incorrect!  I will be sending out follow up emails to better understand certain parts of disk hotplug code and disk frontend/backend parameters setup code and will make changes based on the input I get and the patch as a whole should be ready for review after that.  If you would rather I resend just the interface/parsing changes for review in that case, let me know.  Thanks.

Kamala