On 4/9/24 11:19 AM, Boris Fiuczynski wrote:
On 4/9/24 16:56, Cole Robinson wrote:
This was the implied default before nodedevs gained a notion of being inactive, and matches how we handle parsing other objects
Signed-off-by: Cole Robinson <crobinso@redhat.com> --- src/test/test_driver.c | 1 + 1 file changed, 1 insertion(+)
diff --git a/src/test/test_driver.c b/src/test/test_driver.c index 41828f86b6..9db7a44035 100644 --- a/src/test/test_driver.c +++ b/src/test/test_driver.c @@ -1269,6 +1269,7 @@ testParseNodedevs(testDriver *privconn, return -1; } + virNodeDeviceObjSetActive(obj, true);
This will actually render the mdev object to be transient which is an active mdev not having a persistent definition. The data using virNodeDeviceDefParseXML is stored in the mdevs defined_config only therefore the data is showing up when you use "virsh nodedev-dumpxml" as it defaults to the "current state" of the nodedev.
For data consistency of the node you should instead do virNodeDeviceObjSetPersistent(obj, true);
Now the mdev is inactive and persistent and the virsh command should be correct.
I guess this would also resolve the requirement for the next patch unless you have a requirement to mock dumping transient mdevs.
Oops I replied to patch 5 before seeing this one. Let me explore this a bit. Thanks, Cole