Add a simple WiFi MLO test (New)

[tomli380576]

WARNING: This modifies com.canonical.certification::sru-server

This test is only intended for PCs running 24.04 and later. Please lmk if the test job should be placed somewhere else.

A new pair environment variables (MLO_SSID, MLO_PSK) has been added for this test. Default values for both of them have been added here: https://github.com/canonical/ce-oem-dut-checkbox-configuration/pull/278/

Caution

Since some of the labs don't have a MLO wifi AP yet, please don't merge this PR immediately after approval. I'll update in this thread once I can be sure that all labs have an MLO AP.
Update: If the mlo_ssid_specified workaround is acceptable, we can merge this without waiting for all the labs to be ready.
Update 2: safe to merge after approval. Labs that don't have a Wifi7 AP can set the manifest entry to false

Description

This PR adds a WiFi MLO (multi-link operation) test for the upcoming wifi-7 capable cards. For simplicity this PR only uses nmcli to connect/disconnect and does not take networkd into account.

Resolved issues

Documentation

What this test does is basically:

1. Connect to a wifi AP that supports MLO with nmcli
2. If connection worked, check if iw dev prints multiple links under the "MLD with links" section

Interface wlp0s20f3
	ifindex 2
	wdev 0x1
	addr c8:6e:08:17:85:4a
	ssid cert-be-wpa3-tel-l10
	type managed
	wiphy 0
	multicast TXQ:
		qsz-byt	qsz-pkt	flows	drops	marks	overlmt	hashcol	tx-bytes	tx-packets
		0	0	0	0	0	0	0	0		0
	MLD with links:
	 - link ID  1 link addr ea:ab:77:e3:d8:e7
	 - link ID  2 link addr 92:4f:25:02:8a:eb
	   channel 85 (6375 MHz), width: 320 MHz, center1: 6265 MHz

3. Disconnect and remove the connection

The file is called wifi 7 test because there are more features in wifi 7 than what's implemented here. MLO happens to be a high priority one so it's included first. MLO and connection type are the only checks that can fail the entire test case for now; the connection bandwidth and 4096QAM checks only print warnings because 320mhz is not a required feature and 4096QAM is too sensitive to the environment.

Manifest?

I think we don't need a new manifest such as "has_mlo_support" here because according to Cisco's guide, MLO is a required feature of wifi7. So if we have wireless_sta_protocol.interface_be == 'supported', then we can safely assume that MLO is supported by the DUT. If the test fails, either there's a real issue with the DUT's wifi stack or the environment isn't set up correctly

Tests

Unit tests

https://certification.canonical.com/hardware/202508-37862/submission/450483/
https://certification.canonical.com/hardware/202508-37862/submission/451747/

[codecov]

Codecov Report

❌ Patch coverage is 91.86047% with 14 lines in your changes missing coverage. Please review.
✅ Project coverage is 30.02%. Comparing base (1ad5e7a) to head (14e360d).

Files with missing lines	Patch %	Lines
providers/base/bin/wifi_7_test.py	91.08%	9 Missing and 5 partials ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #2109       +/-   ##
===========================================
- Coverage   52.86%   30.02%   -22.84%     
===========================================
  Files         395      149      -246     
  Lines       42636    16401    -26235     
  Branches     7906     2812     -5094     
===========================================
- Hits        22538     4924    -17614     
+ Misses      19295    11345     -7950     
+ Partials      803      132      -671     

Flag	Coverage Δ
checkbox-ng	?
checkbox-support	?
contrib-provider-ce-oem	?
provider-base	29.98% <91.86%> (+0.65%)	⬆️
provider-certification-client	?
provider-certification-server	?
provider-dss	?
provider-genio	?
provider-gpgpu	?
provider-iiotg	?
provider-resource	?
provider-sru	?
release-tools	?

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[seankingyang]

I left some suggestion.
Most of the script logics are good for me, and i think if we only grab one time (after 5 sec negotiation) may easy fail when the lab env is not stable.

To make the script more reliable, let's change how we check the connection.
Instead of checking only once, we should keep trying for 30 seconds. The first time the check passes, the script should exit and report success.
This will prevent the script from failing due to temporary network issues.

[tomli380576]

I left some suggestion. Most of the script logics are good for me, and i think if we only grab one time (after 5 sec negotiation) may easy fail when the lab env is not stable.

To make the script more reliable, let's change how we check the connection. Instead of checking only once, we should keep trying for 30 seconds. The first time the check passes, the script should exit and report success. This will prevent the script from failing due to temporary network issues.

Thanks for the catch! I encountered a "connection successful + only 1 link present" situation today and a simple disconnect + reconnect fixed it.

[seankingyang]

Thanks for adding the retry logic.
I see you use slugify so the iw output updates on each retry, even if the job is re-run. This is the great idea for debugging.

[clairlin53]

LGTM+1

[pieqq]

I left a few comments inline below.

One thing I have concerns with is how you do the connection. If I run this on my device, my current WiFi connection to my usual AP will be lost, no? Other jobs like wireless/wireless_connection_* use a different way to handle a test connection, it seems, by creating a specific connection name and using it:

checkbox/providers/base/bin/wifi_nmcli_test.py

Lines 77 to 115 in dde3844

	def turn_up_connection(uuid):
	# uuid can also be connection name
	print_head("Turn up NM connection")
	cmd = "nmcli c up {}".format(uuid)
	print("Turn up {}".format(uuid))
	activate_uuid = get_nm_activate_connection()
	if uuid == activate_uuid:
	print("{} state is already activated".format(uuid))
	return None
	try:
	print_cmd(cmd)
	sp.call(shlex.split(cmd))
	except Exception as e:
	print("Can't turn on {}: {}".format(uuid, str(e)))
	def turn_down_nm_connections():
	print_head("Turn off NM all connections")
	connections = _get_nm_wireless_connections()
	for name, value in connections.items():
	uuid = value["uuid"]
	print("Turn down connection", name)
	cmd = "nmcli c down {}".format(uuid)
	print_cmd(cmd)
	sp.check_call(shlex.split(cmd))
	print("{} {} is down now".format(name, uuid))
	print()
	def delete_test_ap_ssid_connection():
	print_head("Cleaning up TEST_CON connection")
	connections = _get_nm_wireless_connections()
	if "TEST_CON" not in connections:
	print("No TEST_CON connection found, nothing to delete")
	return
	cmd = "nmcli c delete TEST_CON"
	print_cmd(cmd)
	sp.check_call(shlex.split(cmd))
	print("TEST_CON is deleted")

[tomli380576]

Hi @pieqq, Thanks for the review! The length of the test grew quite a bit so here's a short summary of what changed:

1. The problem with existing connections being lost: This is fixed by adding a nmcli c up call for each existing wifi connection after all the tests are done. I didn't add a connection name here like the other nmcli test since the original connection will still be erased if it's the same AP as --mlo-ssid whether the new connection has a name or not. This connection will still be restored at the end, except it calls connect() again instead of nmcli c up. Let me know if you would still like to see a name just for readability.
2. Resource jobs and checking package versions: I moved the version comparison logic to the resource job which only produces True/False for the pxu engine to consume. One thing I would like to get feedback on is the mlo_ssid_specified condition as it is a temporary workaround for labs that don't have an MLO AP set up. For labs that have an AP ready, we can have them run this test early.
3. Parser changes: In the output of iw dev <iface> link, the number of words in the tx bitrate or rx bitrate is not guaranteed, so I replaced the fixed index accesses with list searches to handle different connection types. There's also a new connection type Legacy for pre-wifi-4 APs.

[tomli380576]

The CI failures are caused by missing py3.5 and py3.6 packages which should be fixable with a rebase since #2141 was merged. I'll rebase after the review to avoid completely wiping out the commit timeline.

[pieqq]

The Python code looks good to me.

2 things inline with regards to passing env variable to the Checkbox job, and maybe using a manifest instead of a resource.

[tomli380576]

Final run with the latest commit: https://certification.canonical.com/hardware/202503-36508/submission/455725/

[tomli380576]

Hi @pieqq, I've rebased the branch to make it ready to merge since the only big changes after the last review are in the pxu files. The commit to to diff against is abd2de8. Thanks!