[Libvirt-cim] [PATCH 1/2] cimtest: Adding support for ConsoleRASD

[John Ferlan]

On 10/11/2013 07:09 AM, Viktor Mihajlovski wrote:
> The test domains are now defined with a ConsoleRASD if the
> tested libvirt-cim does support those. A few testcases had
> to be adjusted to properly handle the newly encountered instances
> and associations.
> 
> Signed-off-by: Viktor Mihajlovski <mihajlov at linux.vnet.ibm.com>


<...snip...>

> diff --git a/suites/libvirt-cim/lib/XenKvmLib/vxml.py b/suites/libvirt-cim/lib/XenKvmLib/vxml.py
> index 74c4f23..9489a5c 100644
> --- a/suites/libvirt-cim/lib/XenKvmLib/vxml.py
> +++ b/suites/libvirt-cim/lib/XenKvmLib/vxml.py
> @@ -50,9 +50,11 @@ from CimTest.ReturnCodes import SKIP, PASS, FAIL
>  from XenKvmLib.classes import virt_types, get_typed_class
>  from XenKvmLib.enumclass  import GetInstance
>  from XenKvmLib.const import get_provider_version
> +from XenKvmLib.xm_virt_util import host_cpu_model
>  
>  vsms_graphics_sup = 763
>  vsms_inputdev_sup = 771
> +vsms_console_sup = 1276
>  
>  class XMLClass:
>      xml_string = ""
> @@ -598,7 +600,8 @@ class VirtCIM:
>      def __init__(self, virt, dom_name, uuid, pae, acpi, apic, disk_dev, 
>                   disk_source, net_type, net_name, net_mac, vcpus, mem,
>                   mem_allocunits, emu_type, grstype, ip,
> -                 is_ipv6_only, port_num, kmap, irstype, btype, vnc_passwd):
> +                 is_ipv6_only, port_num, kmap, irstype, btype, vnc_passwd,
> +                 con_source_type, con_target_type):
>          self.virt = virt
>          self.domain_name = dom_name
>          self.err_rc = None
> @@ -623,6 +626,9 @@ class VirtCIM:
>                                                    ipv6_flag=is_ipv6_only, 
>                                                    lport=port_num, keymap=kmap, 
>                                                    vnc_passwd=vnc_passwd)
> +            self.casd = vsms.get_casd_class(virt)(name=dom_name,
> +                                                  source_type=con_source_type,
> +                                                  target_type=con_target_type)
>          self.masd = vsms.get_masd_class(virt)(megabytes=mem, 
>                                                mallocunits=mem_allocunits,
>                                                name=dom_name)
> @@ -655,9 +661,14 @@ class VirtCIM:
>                  res_settings.append(str(self.gasd))
>  
>          if curr_cim_rev >= vsms_inputdev_sup:
> -            if self.iasd is not None:
> +            if self.iasd is not None  and \
> +                    not str(host_cpu_model(ip, self.virt)).startswith('s390'):
>                  res_settings.append(str(self.iasd))
>  
> +        if curr_cim_rev >= vsms_console_sup:
> +            if ref_conf is None and self.casd is not None:
> +                res_settings.append(str(self.casd))
> +


I've been working on applying the "next set" of Xu Wang's fixes and
found that by just adding the four core XenKvmLib/*.py files to a test
branch I could core FilterList 03_create.py test rather reliably until I
commented out the above 3 lines.

The first change applied that caused the failure was the writing of the
console xml (#22 in the series).

The core was because for some reason the dumping of the xml didn't find
the "<filterref>" tag and thus the call to "filterref =
test.libvirt_applied_filter_lists(domain_name)[0]" would have an index
out of range exception.

Also a number of the VirtualSystemManagementService and VSSD tests
failed with the "Only the first console can be a serial port..." message.

Unfortunately I missed the notification that the system I had borrowed
had timed out its lease and thus I've lost usage for a bit.  However, I
do know without this particular patch, the other changes did run through
cimtest.

I'm going to try and revive the ability to run on my own f19 system, I
have to find a tog-pegasus that works for me....

John