[Libvirt-cim] [PATCH] [TEST] Adding new tc to verify KVMRedirectionSAP

[Kaitlin Rupert]

Deepti B. Kalakeri wrote:
> # HG changeset patch
> # User Deepti B. Kalakeri <deeptik at linux.vnet.ibm.com>
> # Date 1232018404 28800
> # Node ID cc74023eeaf78970482e0a7a0c4c37a7ff997530
> # Parent  43bf01b82ce9f627af1b5db55922b0584cd0e3b4
> [TEST] Adding new tc to verify KVMRedirectionSAP.
> 
> Tested with KVM/Xen on current sources and KVM with F9 rpm.

I should have caught this earlier, but we're currently doing the following:

         self.add_sub_node(devices, 'graphics', type='vnc', port='5900',
                           keymap='en-us')

This adds the graphics device node to the XML, which is good - we need 
this when calling define() (i.e. using virsh).

However, when cim_define() is called, the XML isn't used at all. 
Instead, cim_define() passes RASD objects to the DefineSystem() call. 
Right now, cim_define() doesn't specify a graphics RASD.  When the 
DefineSystem() call is made, the provider is using default values since 
no graphics RASD is supplied.

If you have an existing guest that has a VNC port of 5900, this test 
will fail.

> diff -r 43bf01b82ce9 -r cc74023eeaf7 suites/libvirt-cim/cimtest/KVMRedirectionSAP/01_enum_KVMredSAP.py
> --- /dev/null	Thu Jan 01 00:00:00 1970 +0000
> +++ b/suites/libvirt-cim/cimtest/KVMRedirectionSAP/01_enum_KVMredSAP.py	Thu Jan 15 03:20:04 2009 -0800
> @@ -0,0 +1,166 @@


> +        g_type = vsxml.xml_get_graphics_type()

This call is getting the graphics value from the XML. However, that 
might not be an accurate representation of what was passed to the 
DefineSystem() call.

We need to do some cleanup here to ensure that the XML is a reflection 
of the actual guest that was defined.

> +
> +        lport = vsxml.xml_get_graphics_port()

Same comment here.

-- 
Kaitlin Rupert
IBM Linux Technology Center
kaitlin at linux.vnet.ibm.com