memory/gmap: add test for shadow memory counters #5287
Conversation
|
smitterl
force-pushed
the
gmap_counters
branch
3 times, most recently
from
0d0f655
to
06f9575
Compare
smitterl
force-pushed
the
gmap_counters
branch
2 times, most recently
from
6e712ee
to
67213cd
Compare
|
@fbq815 Please can you help review? |
|
I have a question: is l2_guest in this code still a level 1 guest? Others LGTM |
smitterl
force-pushed
the
gmap_counters
branch
from
67213cd
to
a60332b
Compare
It is a level 1 kvm guest but L2 guest here because an LPAR is L1 as it's also virtualized. |
if I understand correctly, I prefer we assume the virtual machine as L1 in confgiure file case you also name a functiona as "l1_counters_go_up". There would be some mis-understanding for others? How about we add comment like "we assume the test scenario as virtual machine which is a L2 guest in LPAR" and we only have L1 and L2 as the name? |
Depends on the language, a mainframe LPAR is a virtualization L1, therefore a KVM guest on the LPAR is an L2. You could count the first KVM guest as L1 but that's not how it's usually counted on IBM side IIUC. |
smitterl
force-pushed
the
gmap_counters
branch
from
a60332b
to
4ad0e6d
Compare
I get your point. I updated the |
libvirt/tests/src/memory/gmap.py
Outdated
| }, | ||
| "target": {"dir": target_tag}, | ||
| "binary": {"flock": "off", "lock_posix": "off"} | ||
| }) |
There was a problem hiding this comment.
Just FYI.
libvirt_device_utils.create_fs_xml() is seldom used in current tp-libvirt. I did only find one occurrence.
Now the popular codes are like below. And the hardcoded 7 lines in python module are not recommended. I think it is better if we put the parameters and values into cfg.
fs_dict = {'accessmode':'passthrough', 'source':{'dir': '${source_dir}'}, "target": {'dir': 'mount_tag0'}, 'binary': {'lock_posix':'off','flock':'off'}}
dev_type = "filesystem"
fs_dict = eval(params.get('fs_dict', '{}'))
fs = libvirt_vmxml.create_vm_device_by_type(dev_type, fs_dict)
virsh.attach_device(vm_name, fs.xml, debug=True, ignore_status=False)
There was a problem hiding this comment.
Got it. Thanks for the example. Updated.
libvirt/tests/src/memory/gmap.py
Outdated
| test.fail("Gmap counters didn't go up.") | ||
| finally: | ||
| cleanup_actions.reverse() | ||
| for a in cleanup_actions: |
There was a problem hiding this comment.
This naming (a) seems informal. Could you give a meaningful name?
There was a problem hiding this comment.
Here the scope is really small just this and the next line so I think the name is ok. However, I updated it to 'action'.
There was a problem hiding this comment.
@dzhengfy I addressed all your comments, please have a look, thank you!
There was a problem hiding this comment.
JOB ID : 064246135562db85db932c63662d75fb237c078a
JOB LOG : /var/lib/avocado/job-results/job-2024-02-19T05.51-0642461/job.log
(1/1) type_specific.io-github-autotest-libvirt.gmap.l3_shadow_table_counters: PASS (27.17 s)
RESULTS : PASS 1 | ERROR 0 | FAIL 0 | SKIP 0 | WARN 0 | INTERRUPT 0 | CANCEL 0
JOB HTML : /var/lib/avocado/job-results/job-2024-02-19T05.51-0642461/results.html
JOB TIME : 27.49 s
KVM records shadow memory events for a nested guest. Confirm the counters go up. Signed-off-by: Sebastian Mitterle <smitterl@redhat.com>
smitterl
force-pushed
the
gmap_counters
branch
from
4ad0e6d
to
1879af8
Compare
KVM records shadow memory events for a nested guest. Confirm the counters go up.