Repository navigation
anyka: RTL8188FU, RDA5995 and the USB host on every Anyka image, no load_anyka error - #2640
Conversation
The TECKIN TC100 (AK3918EV300, no Ethernet) carries an RTL8188FU (0bda:f179). The image shipped only r8188eu, which binds it through new_id but never tunes, so the camera had no network at all. The Anyka USB host (otg-hs) was a module that nothing loaded: USB, and so the Wi-Fi dongle, came up only after a manual modprobe. It is now built in on every AK39 board; AV130 keeps the vendor's ak_hcd. Reported in #2639.
The AK39 boards build their sensor, ISP and encoder drivers into the kernel and ship no load_anyka, so every boot printed "load_anyka: not found" (#2639). Families with a load script run it exactly as before.
Ethernet-less Anyka cameras on any of the AK39 parts carry the same USB RTL8188FU as the TC100 (#2639), not only EV300.
Rdamicro's rdawfmac (USB 1e04:8888) with its firmware compiled in, from OpenIPC/rda-wlan: drivers/net/wireless/rdaw80211 of realme's open TechLife Robot Vacuum kernel, imported unmodified, then built for USB by default. RDA5995 is the other USB Wi-Fi chip common on Ethernet-less Anyka cameras.
dimerr
changed the title
dimerr
changed the title
PR Summary by QodoEnable USB Wi-Fi and built-in USB hosts across Anyka images
AI Description
Diagram
High-Level Assessment
Files changed (19)
|
Code Review by Qodo
1. Anyka camera changes lack hardware proof
|
![qodo-free-for-open-source-projects[bot]](https://avatars.githubusercontent.com/in/1068000?s=60&v=4){kind=small}
| # USB RTL8188FU (0bda:f179), which r8188eu binds but cannot tune. | ||
| BR2_PACKAGE_RTL8188FU_OPENIPC=y | ||
| BR2_PACKAGE_RDA5995_OPENIPC=y |
There was a problem hiding this comment.
1. Anyka camera changes lack hardware proof 📘 Rule violation ☼ Reliability
The Anyka defconfigs newly select two wireless drivers and build the USB host into the kernel without evidence from a real camera. The PR reports only a QEMU boot and explicitly says no camera was tested, so the changed firmware behavior on every Anyka image remains unverified on affected hardware.
Agent Prompt
## Issue description
The PR changes firmware behavior across Anyka images but provides no real-camera verification; the only reported boot evidence comes from QEMU, and Wi-Fi association is explicitly untested.
## Fix Focus Areas
- br-ext-chip-anyka/configs/ak3918ev300_lite_defconfig[66-68]
- general/overlay/etc/init.d/S70vendor[7-13]
## Recommended Fix
Test an affected Anyka camera with the changed image, recording boot logs, USB host initialization, driver loading, Wi-Fi behavior, and the vendor-module startup message. Add the before/after hardware output to the PR description, or narrow the change to code that does not alter camera firmware behavior.
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
There was a problem hiding this comment.
Agreed, and the PR stays a draft for that reason. Everything here has run only in the emulator; no Anyka camera has booted it. The reporter in #2639 can RAM-boot images over UART, and the hardware output will go in the description once they have. Until then the open questions are the ones listed there: RTL8188FU association, and whether rdawfmac attaches to a real RDA5995.
rda5995-openipc installed rdawfmac but no wlandev profile loaded it, so a camera set to its RDA adapter took S40network's failure path. The profile reports the module's status: rdawfmac fails to load when no chip attaches, and then there is no wlan0 to bring up.
1 participant
Problem
#2639: an AK3918EV300 camera (chip ID 0x20160100, RTL8188FU Wi-Fi, no Ethernet) boots
ak3918ev300_litefrom RAM but has no network:otg-hs(CONFIG_USB_AKOTG_HS_HCD) was a module that nothing loaded; the reporter needed a manualmodprobe otg-hs.r8188eu. Bound to0bda:f179throughnew_id, it createswlan0but finds no APs.load_anyka: not found. The AK39 boards build their sensor, ISP and encoder drivers into the kernel and ship no load script.Changes
BR2_PACKAGE_RTL8188FU_OPENIPC=ygeneral/package/rda5995-openipcrdawfmacfor the RDA5995 (USB1e04:8888), the other USB Wi-Fi chip common on Ethernet-less Anyka cameras; firmware compiled into the module. Source is OpenIPC/rda-wlan:drivers/net/wireless/rdaw80211of realme's open TechLife Robot Vacuum kernel, imported unmodified, then built for USB by default. Pinned to593c3184ae3f84c563b1704f6ad24879f122634b; enabled on every Anyka defconfigak_hcd)CONFIG_USB_AKOTG_HS_HCD=y;olddefconfigleaves each config otherwise unchangedgeneral/overlay/etc/wireless/usbrda5995-genericprofile:modprobe rdawfmac, exiting with its status (the driver fails to load when no chip attaches)general/overlay/etc/init.d/S70vendorload_<vendor>, printsNo load_<vendor> on this image, nothing to loadinstead of failing; families with a script run it exactly as beforeThe S70vendor change is in the shared overlay, so it reaches every image and widens the CI matrix to all boards. It's a separate commit, so it can be dropped if a per-family fix is preferred.
The kernel side of #2639 (the
gpio_keys_inithang, 1-bit SD) is OpenIPC/linux#112. This PR doesn't depend on it.fw_env.configfrom the issue isn't addressed here: the env had never been saved on that board.Hardware tested on
No camera. Built locally and booted in the qemu-hisilicon emulator (
ak3918ev300machine) only. The reporter can test RAM-booted images over UART; the Wi-Fi association itself is untested.Evidence
Local build of
ak3918ev300_liteagainst OpenIPC/linux#112:Before (nightly-20261019 build of master):
After (RTL8188FU and RDA5995):
Emulator boot of the built image:
rdawfmacresolves and runs its init, but with no RDA chip on the emulated bus it waits 9 s for a device to register and unloads with-ENODEV, by the driver's design. Attaching to a real RDA5995 is untested.test_shell_parse.shandSTRICT=1 test_strip_shell_comments.shpass.Scope
general/package/all-patches/linux/(those go to OpenIPC/linux)general/overlay/or in a sharedload_<vendor>script hardcodes a value specific to my boardLD_PRELOAD, and no binaries that cannot be rebuilt from source