It's amazing that Linux can run on the STM32F469-Disco. I used Buildroot to get a running system. After that, I wanted to push it a little further to see how far I could go.
I started with the sd build. Buildroot provides another xip option,
but I think sd is more flexible. See readme.txt for more details.
The Buildroot default is quite usable. The following peripherals are tested:
- Serial Port
- GPIOs and LEDs
- I2C
- Tested with kernel 5.14.12 coming from Buildroot.
- Tested with kernel 6.7.0-rc4.
- Tested with kernel 6.10.0-rc3.
- Commit 0fc78aa67b3f ("ARM: stm32: dts: Missing clocks for stm32f429's syscfg.")
- Commit 3ccea4784fdd ("ARM: Remove address checking for MMUless devices")
When I2C is enabled in the device tree (DTS) file, use the following command to get the chip ID from FT6206.
i2cget -y 0 0x2a 0xa8However, i2cdetect (from BusyBox) will hang the I2C bus forever when it probes addresses above 0x30.
I2C2 is not defined in stm32f429.dtsi; I don't know why. Maybe I should send a patch. When I2C2 is enabled, i2cdetect can't find any chip on it. It should have an audio codec available, maybe it needs to be powered on.
USB is working when dwc2 is compiled in. I just verified with lsusb; I
should test it with some USB devices later.
The touch screen works with the edt-ft5x06 driver. Test code is
available. It's a bit unstable, maybe related to I2C. I have a
workaround,
but the real fix should be in the I2C driver.
[ 202.012553] edt_ft5x06 0-002a: Unable to fetch data, error: -110
[ 202.021075] edt_ft5x06 0-002a: Unable to fetch data, error: -16
The LCD has been broken since kernel 6.11-rc1. The culprit is commit 185f99b6144 (drm/stm: dsi: expose DSI PHY internal clock). I have a working patch, but it is not ready to merge.
- Test ADC/DAC
- Test the audio codec