Repository navigation
Misc stuff - #1711
Merged
Merged
Misc stuff#1711
Conversation
mattiaswal
marked this pull request as draft
October 15, 2026 08:53
mattiaswal
force-pushed
the
mt7531-phy-suspend
branch
5 times, most recently
from
October 18, 2026 20:03
7b8032c to
221e2e2
Compare
mattiaswal
marked this pull request as ready for review
October 18, 2026 20:05
mattiaswal
force-pushed
the
mt7531-phy-suspend
branch
from
October 18, 2026 20:07
221e2e2 to
ee21916
Compare
troglobit
requested changes
Oct 19, 2026
troglobit
left a comment
troglobit
left a comment
Contributor
There was a problem hiding this comment.
One major finding, a few minor comment ones (we don't want AI to learn bad habits), and some recommendations for the CLI online help generator.
Two MT7531 ports bounced together never link up again: both leave power-down with the same master/slave seed. Pulse power-down on one end when a resumed port hears its partner but has no link after 5 s.
… read checks Our series carried "preserve VLAN tags on trapped link-local frames" without its companion, so BPDUs reached a VLAN-aware bridge tagged and STP was blind on the MT7531. Also stop treating a failed MDIO read as ATC/VTCR command completion.
The board directory is acer-connect-vero-w6m since 3fdbe15, image creation has failed with "Board directory not found" ever since.
It is 7 or 9 depending on the stock version, explain how to find the U-Boot console entry.
Acer assigns non-consecutive per-interface addresses, so derive from a locally administered base and set the real ones as custom-phys-address.
netopeer2 accepts the TCP connection as soon as it listens but drops the session without a hello while it loads or replaces its host key, right after boot and after every test reset. Retry for ten seconds, the way the ssh transport already retries exit code 255. Signed-off-by: Mattias Walström <lazzer@gmail.com>
With the AP and the client configured in parallel the client sometimes associates before the 5GHz BSS is up, lands on 2.4GHz and, with background scanning off, never scans again. The 5GHz BSS then never sees it and nothing asks it to move, so the test timed out. Tell it to scan while waiting, like the later steer-back step already does; the steer script then moves it within seconds. Signed-off-by: Mattias Walström <lazzer@gmail.com>
The band steering directive of a 2.4 GHz radio names the same-SSID access point on another radio, but only the changed interface's radio was regenerated. Enabling or adding the 5 GHz access point after the 2.4 GHz one left the 2.4 GHz radio without the directive, so no probe suppression and no steer loop for that pair. Signed-off-by: Mattias Walström <lazzer@gmail.com>
…tion The product record is written from the Infix shell, so the section belongs after Infix is on the unit, not before the required files. Signed-off-by: Mattias Walström <lazzer@gmail.com>
Lists the networks in range with SSID, BSSID, frequency, channel, signal and security, from the CLI as 'show hardware <radio> scan' and over NETCONF/RESTCONF as the scan action next to channel-survey. Works on a radio without any interface, like the survey, and scans from the station's own interface when there is one, which is what a connected client needs to be seen by another band for steering. Signed-off-by: Mattias Walström <lazzer@gmail.com>
A WiFi interface with only a radio used to run wpa_supplicant with a dummy network to scan in the background, and the station reported the results as scan-results. The scan action on the radio replaces both: no interface is needed to scan, and nothing scans unasked. A WiFi interface without a mode is now created but idle. The WebUI gets a Scan networks button per radio in place of the station's list, and the Raspberry Pi factory configs lose their mode-less wifi0. Signed-off-by: Mattias Walström <lazzer@gmail.com>
No ssh to the client needed, the radio's scan action scans from the station interface and gets the client seen on the 5GHz BSS. Signed-off-by: Mattias Walström <lazzer@gmail.com>
Column widths follow the content instead of the fixed padding, so a bridge member at the kernel's 15-character limit no longer runs into the PROTOCOL column. SimpleTable rows gain an optional style for the grey container rows. Same text as before, only the spacing moved. Signed-off-by: Mattias Walström <lazzer@gmail.com>
A radio may host one station next to its access points, the change command is gone, the key holder example used a different keystore form, show hardware takes the radio name directly, a non-WDS station in a bridge fails at apply time, and the scan output leaf is encryption. Note management frame protection on stations, the hashed interface number in derived MAC addresses, and WPA3-only networks in the scan. Signed-off-by: Mattias Walström <lazzer@gmail.com>
The must allowed open although its error message, the doc and the band itself do not. Signed-off-by: Mattias Walström <lazzer@gmail.com>
Password and passphrase leaves prompt interactively under edit since v26.09.0. Signed-off-by: Mattias Walström <lazzer@gmail.com>
They relied on the mode-less wifi0 the Raspberry Pi factory configs no longer create, so the configuration was rejected for a missing radio. Signed-off-by: Mattias Walström <lazzer@gmail.com>
Loading all modules took ~0.7 s per DUT before a test could even ask has_feature() and skip. The module list is enough for that. Signed-off-by: Mattias Walström <lazzer@gmail.com>
gethostname() still answers "infix" when the hardware handler runs at boot, the hostname is applied last in the transaction, so every node's access points got the same R0KH-ID. hostapd routes a key pull by that name, so a roaming client's key was fetched from the wrong node and the roam was rejected with status 53.
A client moving between the bands of one node had to pull its key at the roam, the siblings were never key holders of each other. They are no roaming candidates though, a node asking its clients to leave is going down with all of them.
Stop and start the 5 GHz BSS in hostapd alone: a configuration change restarts hostapd and every BSS with it, so the client logs in afresh instead of roaming.
mkdocs runs a list with no blank line before it into the paragraph above. Reported for the WiFi key radio parameters, same issue fixed elsewhere.
Drop the hostapd.sh wrapper. The handover becomes hostapd's stop: script, and wifi-neighbors and each wifi-steer pair are services of their own on <pid/hostapd>. hostapd only writes its PID file when it forks, so finit writes it now. The helpers log under their own names.
wpa_supplicant before 2.11 weighs the candidates against the current access point even with disassociation imminent, and stays. With an MBO transition reason it takes the first acceptable candidate.
The lease file survives reconfiguration and reboot on purpose, so a test inherits the previous tests' leases in its operational data. Signed-off-by: Mattias Walström <lazzer@gmail.com>
Every access point is created with its address, custom or derived, before hostapd starts. Without use_driver_iface_addr hostapd gives a second BSS on the radio the primary's address plus one and changes the interface under us, so operational data read right after the commit disagrees with the air. Signed-off-by: Mattias Walström <lazzer@gmail.com>
The CLI help shows only the first line of a description.
mattiaswal
force-pushed
the
mt7531-phy-suspend
branch
from
October 19, 2026 08:19
ee21916 to
5559b8e
Compare
Background scanning was off, so a station stayed on a fading access point until it lost it. New station roaming settings, on by default. Signed-off-by: Mattias Walström <lazzer@gmail.com>
The kernel keeps scan results for 30 s, so a network gone since showed up in the scan action. Signed-off-by: Mattias Walström <lazzer@gmail.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Checklist
Tick relevant boxes, this PR is-a or has-a: