ath11k: WCN6855 on Lenovo (17aa:9309) uses generic regdb, 5 GHz passive-only

alexiscalin alexiscalin at protonmail.com
Sun Nov 15 05:16:02 PST 2026


Hardware: ThinkPad P14s Gen 5 AMD (21ME), Qualcomm QCNFA765 / WCN6855
hw2.1 [17cb:1103], subsystem 17aa:9309, chip_id 0x12, board_id 0xff.
Firmware: WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.41
board-2.bin: atheros-firmware-20261019 (Fedora 44)
Kernel: 7.2.13 (ath11k identical to v7.2.13)

ath11k finds the board data for the Lenovo variant (SMBIOS type 0xf8
"LE"):

  boot found board data for 'bus=pci,vendor=17cb,device=1103,
  subsystem-vendor=17aa,subsystem-device=9309,qmi-chip-id=18,
  qmi-board-id=255,variant=LE'

but there is no regdb for it, so it falls back to the generic one:

  boot failed to fetch regdb data for bus=pci,vendor=17cb,device=1103,
  subsystem-vendor=17aa,subsystem-device=9309,qmi-chip-id=18,
  qmi-board-id=255,variant=LE from ath11k/WCN6855/hw2.1/board-2.bin
  boot found match regdb data for name 'bus=pci'

board-2.bin has 16 board entries for Lenovo (subsystem-device 9308,
9309, 930a and e0be) but no Lenovo regdb entry. Its 35 regdb entries are
the generic 'bus=pci' one and 34 for HP (subsystem-vendor 105b).

With the generic regdb, the firmware's default country is BR, and every
5 GHz rule it reports has REGULATORY_CHAN_NO_IR (FLAGS 2), also after
11d has switched the country to US or FR:

  wmi cc_ext BR dfs 0 ...
  wmi reg rule 1: (5170 - 5330 @ 160) (0, 20) (FLAGS 2)
  wmi reg rule 2: (5490 - 5730 @ 160) (0, 20) (FLAGS 2)
  wmi reg rule 3: (5735 - 5895 @ 160) (0, 20) (FLAGS 2)

So all 28 5 GHz channels are scanned passively, including the non-DFS
ones. For FR, wireless-regdb (iw reg get, global) has no NO-IR on these
ranges, but the device reports:

  global: (5150 - 5250 @ 80), (N/A, 23), (N/A), NO-OUTDOOR, AUTO-BW
          (5725 - 5875 @ 80), (N/A, 13), (N/A)
  phy:    (5170 - 5250 @ 80), (N/A, 20), (N/A), AUTO-BW, PASSIVE-SCAN
          (5735 - 5855 @ 80), (N/A, 14), (N/A), AUTO-BW, PASSIVE-SCAN

A full scan takes about 5 s, of which the 28 passive 5 GHz channels at a
150 ms dwell time account for about 4.2 s. This delays every connection
that needs a scan, e.g., Wi-Fi after resume: with NetworkManager, Wi-Fi
is back 6.6-7.7 s after the wake-up alarm. For comparison, an Intel 8265
laptop measured the same way takes 5.4-5.5 s. (An additional scan done
by wpa_supplicant after resume, which made it 12.6-12.7 s, is a separate
wpa_supplicant issue and has been reported on the hostap list:
"wpa_supplicant: Initialize is_6ghz_enabled when adding an interface".)

Could a regdb for the Lenovo WCN6855 variants be added to board-2.bin,
as was done for HP? I can test a candidate file on this machine.




More information about the ath11k mailing list