Re: ath12k: QCN9274 hw2.0 single-band cards on IPQ9574 — firmware RDDM after WMI_INIT_CMD (WLAN.WBE.1.6-01243)

[Stefano Ursino]

Hi Jeff,

Thanks for the pointer to Nazar's patch and for pinging the team.

Bugzilla #221550 was filed shortly after your message:
https://bugzilla.kernel.org/show_bug.cgi?id=221550

I see you have since assigned it to Rameshkumar Sundaram -- thank you.


Fix B update
------------

I read your May 11 review of Nazar's patch [1] where you pointed out the
zeroed-slot problem and suggested a compact-array approach with a separate
write index. I have implemented exactly that:

  - write index j (separate from iteration index i)
  - unknown module_id entries skipped with ath12k_dbg()
  - ab->num_db_cap = j at loop exit (no zeroed slots, no free_dir_buff:)

The patch applies cleanly on top of current ath-next. I can post it as
[PATCH ath-next] referencing Nazar's series and your review comment if
that would help move it forward.

[1] https://lists.infradead.org/pipermail/ath12k/2026-May/009267.html


Firmware regression matrix (Fixes A + B applied, 3x QCN9274 hw2.0)
-------------------------------------------------------------------

  WLAN.WBE.1.3.1-00162: stalls at 'failed to receive default regd during
                         init' (regulatory timeout, no RDDM)

  WLAN.WBE.1.4.1-00199: no RDDM; 2.4 GHz card (board_id=0x1, OTP
                         programmed) reaches 'ieee80211 phy2' -- driver
                         initializes fully

  WLAN.WBE.1.6-01243:   MHI_CB_EE_RDDM ~316 ms after WMI_INIT_CMD on
                         all three cards simultaneously

WBE.1.4.1 is the clean reference point: the hardware and host driver
work correctly, the RDDM is a regression introduced between WBE.1.4.1
and WBE.1.6.

Thanks,
Stefano