Skip to content

docs: powkiddy rgb10 max 2 port notes - #2

Closed
picaso1-cloud wants to merge 8 commits into
andr36oid:mainfrom
picaso1-cloud:rgb10max2-port
Closed

picaso1-cloud wants to merge 8 commits into
andr36oid:mainfrom
picaso1-cloud:rgb10max2-port

Conversation

@picaso1-cloud

Copy link
Copy Markdown

Adds porting notes for the powkiddy rgb10 max 2 (rk3326, og super class) next to the existing kernel-updates doc:

  • the console, its odroid-go-super board family and its usb-c ports
  • the profile folder, its line in andr36oid-profiles.txt and the BOOT files
  • the two boot.ini lines the console needs (orientation, usbcore autosuspend)
  • the device tree: panel, buttons, leds, battery, and removing the rk817 charger's dc-detect gpio (it floats high here and shows a false charging state)
  • the wifi power problem and the boot service that solves it, including why the rk817 otg rail behaves as a pmic latch that kernel writes never reach, and the raw i2c write that flips it
  • the two small static tools the service needs (rk817tool, wifitool)
  • what works, the known limits (2.4 ghz only, no wpa3/pmf) and the evidence

The service script and tool sources are included under docs/rgb10max2/ so the release image build can be updated from this folder.

Adds porting notes for the powkiddy rgb10 max 2 (rk3326, og super class)
next to the existing kernel-updates doc:

- the console, its odroid-go-super board family and its usb-c ports
- the profile folder, its line in andr36oid-profiles.txt and the BOOT files
- the two boot.ini lines the console needs (orientation, usbcore autosuspend)
- the device tree: panel, buttons, leds, battery, and removing the rk817
  charger's dc-detect gpio (it floats high here and shows a false charging
  state)
- the wifi power problem and the boot service that solves it, including why
  the rk817 otg rail behaves as a pmic latch that kernel writes never reach,
  and the raw i2c write that flips it
- the two small static tools the service needs (rk817tool, wifitool)
- what works, the known limits (2.4 ghz only, no wpa3/pmf) and the evidence

The service script and tool sources are included under docs/rgb10max2/ so the
release image build can be updated from this folder.
Corrects the power section after hardware verification: the stock topology
(otg phy port disabled) is what keeps the charger working - with the phy port
enabled the charger receives EXTCON_USB_VBUS_EN, sets otg_in and disables its
own plug interrupts, so a later charger is never detected. The rk817 OTG rail
is a PMIC-side latch flipped by a raw i2c write at boot. Verified on hardware:
charging at +869 mA with wlan0 connected at the same time.
Refresh the device tree and service files to the validated versions.
Documents the panel bug (SF/HWC never enables the CRTC for the Android
framebuffer; first vop modeset only happens on a sleep/wake cycle) and the
service fix: an automatic power-key cycle after sys.boot_completed brings
the screen up with no button press. Also sets
debug.sf.nobootanimation=1 in /vendor/build.prop, which cuts ~10s of boot.
Updates the service file to the validated v27.
…mode combo)

The port mapped the stick-push GPIOs to BTN_TRIGGER_HAPPY5/6, so Android never
saw an L3+R3 press and the mouse-mode feature never launched. Emit
BTN_THUMBL/BTN_THUMBR (317/318) directly: Generic.kl maps them to
BUTTON_THUMBL/THUMBR and emulators see standard L3/R3. Verified on hardware
via getevent after a live BOOT-partition update.
…ping)

This board wires the right stick pots to the opposite mux channels vs the R36S
(ch0 = R-X, ch1 = R-Y). The stock abs_*_channel props are inert in this driver;
use amux-channel-mapping = <1 0 2 3> with invert-absrx and no invert-absry.
Fixes the swapped/inverted right stick in mouse mode and emulators. Verified on
hardware via getevent and the joymouse cursor directions.
If /data/local/tmp/ts/{tailscaled,state/tailscaled.state} exist, the boot
service starts tailscaled (full VPN mode, /dev/tun works as root) and the
janitor keeps it alive. Without those files the block is a no-op, so the
service stays universal.
@itskenny0

Copy link
Copy Markdown
Member

Apologies, but these commits don't seem to pertain to the sd flasher. Would you mind opening pull requests in the correct repositories? Thanks lots!

@itskenny0 itskenny0 closed this Nov 8, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants