packages/linux-drivers: patch RTL8733BU for kernel 6.x and WPA3/iwd

[Zetarancio]

Patches the existing RTL8733BU package (WiFi + BT, USB). Scope: projects/ROCKNIX/packages/linux-drivers/RTL8733BU/ only (patches + modprobe.d). No device options or DTS. Can cohexist with this.

---

001 – 6.19 / build

• Makefile: EXTRA_CFLAGS / EXTRA_LDFLAGS for distro build.
• Rename hmac_sha256_* → rtw_hmac_sha256_* (kernel symbol clash).
• Force wireless_mode / max_bss_cnt when zero (6.19 init order) so wiphy registers.

002 – MAC address

• Use dev_addr_set() instead of writing netdev->dev_addr (required since 5.17).

003 – WPA3 (SAE) and iwd

• SAE key_mgmt_suite host byte order on 6.9+; copy cfg80211_external_auth_params fields (layout changed); mgmt_tx accept NULL channel; trigger association after SAE success; remove duplicate start_clnt_assoc in set_pmksa; MODULE_IMPORT_NS and timer compat.

004 – USB autosuspend

• Enable autosuspend and wire autopm in open/close; fix bt-coexist when CONFIG_USB_AUTOSUSPEND is set.
• Note: USB autosuspend may not be used in ROCKNIX (e.g. if devices rely on full power-down via a platform/rfkill driver). The patch is included for builds that enable it; behavior is correct when autosuspend is disabled.

modprobe.d/8733bu.conf
rtw_ips_mode=1, rtw_power_mgnt=2, rtw_lps_level=1 (IPS/LPS).

[aenertia]

See PR: regard wpa3 handshake issue when rk_crypto is present that needs resolving. Likewise the PK8 built-in needs to be present. This is an out of tree module so it 'MAY' bypass iwd for some handshake negotiation - but i'm dubious.

Issue also mentions connman, which we DO NOT USE, at all for Wifi. iwd fully handles both l2 and l3 addressing and connman has explicit blacklist/compiled without iwd support at all. It's something I want to remove as connman is just bloat for how we use it currently.

#2354

I would split this up. Remove any iwd bits / patches. Just focus on the driver packaging and inclusion.

[sydarn]

I would split this up. Remove any iwd bits / patches. Just focus on the driver packaging and inclusion.

It does only change the driver.

[spycat88]

I would split this up. Remove any iwd bits / patches. Just focus on the driver packaging and inclusion.

Eh? This PR doesn't touch iwd, it only touches the driver.

I see no issues here.

[aenertia]

I would split this up. Remove any iwd bits / patches. Just focus on the driver packaging and inclusion.

Eh? This PR doesn't touch iwd, it only touches the driver.

I see no issues here.

Yeah sorry I worded that badly - the patch just mentions a bunch of IWD specific handling bugs which I thought was worthwhile to investigate.

[Zetarancio]

No it's my bad. I originally developed it for my own buildroot and then I did copy and paste because I am lazy