Re: [PATCH v3 00/13] This series adds support for the MediaTek MT7927 (Filogic 380) combo

[Javier Tia <floss@jetm.me>]

Hi Sean,

On Wed, Mar 26, 2026 Sean Wang wrote:
> This part patch 1-8 looks good overall. My only concern was a possible
> regression on mt7925, but now that the incorrect mt7925 320 MHz
> support has been fixed  in the version, this should be fine.

Thank you. Patch 5 also had a build error (is_320mhz_supported()
called is_mt7927() before it was defined in patch 6). Fixed in v4
by using mt76_chip() directly.

> I am still thinking a bit more about patches 9-11. My current
> preference is to introduce the generic layer first, and then migrate
> the mt7925 and mt7927-specific parts on top of it.
> I will handle this part on my side, since I want to carefully compare
> the vendor driver with your changes first, and make sure the mt7925
> side is solid before moving the mt7927-related changes forward. For
> the mt7927-related work, I will make sure your contribution, as well
> as the work from the other volunteers, is properly preserved when I
> carry this forward.

Understood. v4 drops patches 9-13 so you have full control over
the DMA, HW init, band_idx, and PCI device ID enablement.

Regarding your feedback on v3 patch 13 (CNM quirk) - you suggested
folding the MT792x_FW_CAP_CNM force into mt792x_get_mac80211_ops()
instead of the post-hoc memcpy. Since the PCI IDs patch is now
part of your series, would you prefer to handle the CNM fix there
as well, or should I send a standalone patch for it ahead of your
DMA/init work?

> Similar to the BT side, I would prefer to have a dedicated Linux
> firmware for mt7927 WiFi.

Noted. The driver currently requests mediatek/mt7927/ paths. Once
a dedicated firmware is available we can adjust if needed.

v4 sent with patches 1-8 + ASPM/PM disable (9 patches):
https://lore.kernel.org/linux-wireless/20260326-mt7927-wifi-support-v4-v4-0-8ab465addcfe@jetm.me/T/#u

Best,
Javier