board: arduino: portenta_h7: Add support for various peripherals.

[rahul-arasikere]

I have been developing with the Arduino Portenta H7 with the machine controller shield and I had noticed some of the peripherals were not enabled in the device tree. As such I have enabled a few of them and done some basic testing to ensure they work.

I have opened this pull request to see if there was any interest in merging them upstream and plan to in the coming few weeks to add support for the additional features of the board in a seperate pull request.

This pull request adds support for the following:

• Enables the external LSE (32.768 KHz) and HSE (25 MHz) clocks
• Modifies the clock tree to change the SYSCLK from 96 MHz to 480 Mhz for the M7 core and 240 MHz for M4 Core
• Enables the external SDRAM
• Enables the external QuadSPI Flash
• Enables the ethernet port
• Enables the high speed USB peripheral over the full speed usb peripheral. This also enables the CDC ACM driver for serial output.
• Rearranges the flash partition to eventually support mcuboot

The remaining work required to fully support this board is the following:

• Support the NXP PF1550 PMIC
• Support the Murata 1DX Wi-FI / BLE Co-Processor
• Support the NXP SE050 Secure Element
• Support the Analogix ANX7625 DisplayPort over USB C peripheral

I would appreciate any feedback or comments on this pull request. Thanks.

[rahul-arasikere]

After further testing, num-bidir-endpoints of usbotg-hs node can be set to 9 on the latest zephyr version. I think they fixed the rx fifo issue I noticed on the previous versions.

[pillo79]

Hello, sorry for the delay. Tried your PR on a Portenta H7 and could not get the clock to stabilize. This is the backtrace from a west debug:

^C
Program received signal SIGINT, Interrupt.
0x08043046 in LL_RCC_HSE_IsReady () at modules/hal/stm32/stm32cube/stm32h7xx/drivers/include/stm32h7xx_ll_rcc.h:1688
1688      return ((READ_BIT(RCC->CR, RCC_CR_HSERDY) == (RCC_CR_HSERDY)) ? 1UL : 0UL);

(gdb) bt
#0  0x08043046 in LL_RCC_HSE_IsReady () at modules/hal/stm32/stm32cube/stm32h7xx/drivers/include/stm32h7xx_ll_rcc.h:1688
#1  set_up_fixed_clock_sources () at zephyr/drivers/clock_control/clock_stm32_ll_h7.c:587
#2  stm32_clock_control_init (dev=<optimized out>) at zephyr/drivers/clock_control/clock_stm32_ll_h7.c:850
#3  0x08044d08 in z_sys_init_run_level (level=level@entry=INIT_LEVEL_PRE_KERNEL_1) at zephyr/kernel/init.c:338
#4  0x08044efc in z_cstart () at zephyr/kernel/init.c:629
#5  0x08042320 in z_prep_c () at zephyr/arch/arm/core/cortex_m/prep_c.c:196
#6  0x0804218a in z_arm_reset () at zephyr/arch/arm/core/cortex_m/reset.S:169

This was after doing this:

west build -p -b arduino_portenta_h7//m7 samples/basic/blinky && west flash

The board has the Arduino bootloader which is unaffected. The Portenta H7 is on a Portenta Max Carrier, but I see the same behavior also if I don't connect anything.
Reverting the first commit ("enable 480M") does not build due to a SYSCLK definition error.
Any idea what I might do wrong or what should I test?

[rahul-arasikere]

Hello, sorry for the delay. Tried your PR on a Portenta H7 and could not get the clock to stabilize. This is the backtrace from a west debug:

^C
Program received signal SIGINT, Interrupt.
0x08043046 in LL_RCC_HSE_IsReady () at modules/hal/stm32/stm32cube/stm32h7xx/drivers/include/stm32h7xx_ll_rcc.h:1688
1688      return ((READ_BIT(RCC->CR, RCC_CR_HSERDY) == (RCC_CR_HSERDY)) ? 1UL : 0UL);

(gdb) bt
#0  0x08043046 in LL_RCC_HSE_IsReady () at modules/hal/stm32/stm32cube/stm32h7xx/drivers/include/stm32h7xx_ll_rcc.h:1688
#1  set_up_fixed_clock_sources () at zephyr/drivers/clock_control/clock_stm32_ll_h7.c:587
#2  stm32_clock_control_init (dev=<optimized out>) at zephyr/drivers/clock_control/clock_stm32_ll_h7.c:850
#3  0x08044d08 in z_sys_init_run_level (level=level@entry=INIT_LEVEL_PRE_KERNEL_1) at zephyr/kernel/init.c:338
#4  0x08044efc in z_cstart () at zephyr/kernel/init.c:629
#5  0x08042320 in z_prep_c () at zephyr/arch/arm/core/cortex_m/prep_c.c:196
#6  0x0804218a in z_arm_reset () at zephyr/arch/arm/core/cortex_m/reset.S:169

This was after doing this:

west build -p -b arduino_portenta_h7//m7 samples/basic/blinky && west flash

The board has the Arduino bootloader which is unaffected. The Portenta H7 is on a Portenta Max Carrier, but I see the same behavior also if I don't connect anything. Reverting the first commit ("enable 480M") does not build due to a SYSCLK definition error. Any idea what I might do wrong or what should I test?

So just tried running the sample against the commit and it is working:

LED state: OFF
LED state: ON
LED state: OFF
LED state: ON

During development I noticed that the oscillator enable pin & ethernet release pins would not getting toggled correctly. This is done via the device tree via the regulator enable pin. Can you try completely powering the board off and turning it on after flashing? @pillo79 Thanks.

[pillo79]

Can you try completely powering the board off and turning it on after flashing?

Already done, no difference.

• With main, the RGB LED blinks quickly green only once, then it blinks red slowly (as per the blinky sample).
  I believe the bootloader leaves the green LED on, and it gets reset by the Zephyr init code.
• With your PR, the green LED stays on constantly and I see that backtrace.

In either case, a double tap allows me to enter bootloader and reflash.
Do you have the Arduino bootloader installed? Just so we are in sync, can you update to the latest one (I'm using mbed_portenta 4.1.1) and test again?
Thanks for the feedback!

[rahul-arasikere]

Can you try completely powering the board off and turning it on after flashing?

Already done, no difference.

• With main, the RGB LED blinks quickly green only once, then it blinks red slowly (as per the blinky sample).
  I believe the bootloader leaves the green LED on, and it gets reset by the Zephyr init code.
• With your PR, the green LED stays on constantly and I see that backtrace.

In either case, a double tap allows me to enter bootloader and reflash. Do you have the Arduino bootloader installed? Just so we are in sync, can you update to the latest one (I'm using mbed_portenta 4.1.1) and test again? Thanks for the feedback!

I am using the arduino bootloader based out of the 4.1.1 sdk. I believe my bootloader version is v25. If you revert the first commit, you need to ensure the clock speed is set correctly in the rcc node I believe. Could you probe using an oscillator to see if you can see a 25 Mhz clock signal?

[rahul-arasikere]

@pillo79 what version of zephyr sdk are you using? Have you tried running the blinky sample without any carrier board? I only use the breakout board for ethernet port access.

[kartben]

@pillo79 would it make sense to add Arduino platforms to the MAINTAINERS files so that boards' PRs are correctly assigned? If you think it's a good idea, could you please take care of this? :)

[rahul-arasikere]

After looking into zephyr a little more, I realized the Murata 1DX chip is supported by hal-infineon: look at drivers/wifi/infineon/Kconfig.airoc:100 as well as driver/bluetooth/hci/Kconfig.infineon:75.

If there is interest in trying to get this PR merged, I can spend some time and try to get that working as well.

[erwango]

After looking into zephyr a little more, I realized the Murata 1DX chip is supported by hal-infineon: look at drivers/wifi/infineon/Kconfig.airoc:100 as well as driver/bluetooth/hci/Kconfig.infineon:75.

If there is interest in trying to get this PR merged, I can spend some time and try to get that working as well.

Sure, but let keep this for another PR. Let's complete review on this one first.

[pillo79]

Apologies @rahul-arasikere for dropping the ball on this.

I discussed this internally with the Portenta H7 folks and did some experiments this week on different boards. The issue I was having is related to the missing LSE clock on my board (see comment).

I can confirm that the PR works fine on a fully-populated board; however, there are a few changes to be made to clocks to keep in the design parameters.

[rahul-arasikere]

@pillo79 I have implemented the requested changes, let me know if I need to change anything.

[pillo79]

Thanks for the updates and your effort. Found some minor issues that are easily fixable.

[pillo79]

I would like to add a paragraph on the different board revisions and LSE. Something along these lines (not a native speaker so feel free to adapt!):

The high precision low speed external (LSE) clock is only fully supported on boards with hardware revision 4.10 or greater. By default the internal source is used; to enable the use of the external oscillator, manually specify the hardware revision at build time (see :ref:`application_board_version` for information on how to build for specific revisions of the board).

Also, probably an empty line is required after the table? CI doc builds are down currently so I'm not sure 🤷‍♂️

[romainreignier]

Hi, how to check the board revision we have on hand?

[pillo79]

That is easy using the Arduino IDE and mbed_portenta core: just print out the output from boardRevision() (it's a uint16-packed major.minor combo).

In Zephyr, though, this functionality is not yet available. To get this information directly, you have to read the contents of the revision field of the PortentaBoardInfo structure that is flashed at memory address 0x801F000.

[romainreignier]

Ok, thanks, this might be added to the documentation.
Interestingly, running the STM32H747_getBootloaderInfo Arduino sketch does print a lot of information but not the board revision if we are not in Secure mode (magic == 0x85)

[romainreignier]

I have tested this PR on my Arduino Portenta H7.
I was able to successful run some samples:

samples/hello_world
samples/basic/blinky
samples/subsys/usb/cdc_acm
samples/subsys/console/echo

But I have noticed this warning:

/home/user/zephyrproject/zephyr/build/zephyr/zephyr.dts:1372.29-1405.5: Warning (spi_bus_bridge): /soc/quadspi@52005000: node name for SPI buses should be 'spi'
<stdout>: Warning (spi_bus_reg): Failed prerequisite 'spi_bus_bridge'

warning: USB_DFU_CLASS (defined at subsys/usb/device/class/dfu/Kconfig:4) was assigned the value 'y'
but got the value 'n'. Check these unsatisfied dependencies: IMG_MANAGER (=n). See
http://docs.zephyrproject.org/latest/kconfig.html#CONFIG_USB_DFU_CLASS and/or look up USB_DFU_CLASS
in the menuconfig/guiconfig interface. The Application Development Primer, Setting Configuration
Values, and Kconfig - Tips and Best Practices sections of the manual might be helpful too.

With the Arduino Portenta Breakout board, I have managed to test Ethernet by adding:

if NETWORKING

config NET_L2_ETHERNET
       default y

endif # NETWORKING

and

&mac {
       status = "okay";
};

I could run the following samples:

samples/net/sockets/echo
samples/net/telnet

I am curious about this line in the defconfig:

CONFIG_USB_DFU_CLASS=y

What is it supposed to provide?

Isn't the CONFIG_USB_DEVICE_DFU_PID needed then?

An nice addition for a new PR would be to automatically go to the bootloader when the serial baudrate is set to 1200 bauds, as done by the Arduino IDE.

[rahul-arasikere]

I have tested this PR on my Arduino Portenta H7. I was able to successful run some samples:

samples/hello_world
samples/basic/blinky
samples/subsys/usb/cdc_acm
samples/subsys/console/echo

But I have noticed this warning:

/home/user/zephyrproject/zephyr/build/zephyr/zephyr.dts:1372.29-1405.5: Warning (spi_bus_bridge): /soc/quadspi@52005000: node name for SPI buses should be 'spi'
<stdout>: Warning (spi_bus_reg): Failed prerequisite 'spi_bus_bridge'

warning: USB_DFU_CLASS (defined at subsys/usb/device/class/dfu/Kconfig:4) was assigned the value 'y'
but got the value 'n'. Check these unsatisfied dependencies: IMG_MANAGER (=n). See
http://docs.zephyrproject.org/latest/kconfig.html#CONFIG_USB_DFU_CLASS and/or look up USB_DFU_CLASS
in the menuconfig/guiconfig interface. The Application Development Primer, Setting Configuration
Values, and Kconfig - Tips and Best Practices sections of the manual might be helpful too.

With the Arduino Portenta Breakout board, I have managed to test Ethernet by adding:

if NETWORKING

config NET_L2_ETHERNET
       default y

endif # NETWORKING

and

&mac {
       status = "okay";
};

I could run the following samples:

samples/net/sockets/echo
samples/net/telnet

I am curious about this line in the defconfig:

CONFIG_USB_DFU_CLASS=y

What is it supposed to provide?

Isn't the CONFIG_USB_DEVICE_DFU_PID needed then?

An nice addition for a new PR would be to automatically go to the bootloader when the serial baudrate is set to 1200 bauds, as done by the Arduino IDE.

That must be a holdover from when I was trying to build mcuboot. However I would first need to support the nxp pmic first as I learned from soft bricking one board.

[erwango]

Nit: Odd indentation

[erwango]

Nit: Odd indentation

[erwango]

Nit: Odd indentation

[erwango]

Nit: Odd indentation

[erwango]

Here, there is now a new usage of reg property to encode the actual memory address mapping to be taken into account.
See #68274

		reg = <0x90000000 DT_SIZE_M(16)>; /* 128 Mbits */

Note also the unit that might need to be fixed. I let you check this part

[erwango]

Enabling the USB stack is application responsibility. I propose to provide USB settings in .defconfig file under if USB_DEVICE_STACK condition.

[rahul-arasikere]

In the arduino IDE you would use the USB stack to talk to Serial port. Would it not make sense to have the USB enabled by default for this board?

[erwango]

Please fix this long line

[rahul-arasikere]

@romainreignier that is a good tip about the L2_ETHERNET config, I have added the if ETHERNET config to Kconfig.portenta_h7 file.

[pillo79]

Thanks @rahul-arasikere for this contribution! Tested on both board revisions with no issues.

[fabiobaltieri]

@erwango can you take another look?

[npal-cy]

After looking into zephyr a little more, I realized the Murata 1DX chip is supported by hal-infineon: look at drivers/wifi/infineon/Kconfig.airoc:100 as well as driver/bluetooth/hci/Kconfig.infineon:75.

If there is interest in trying to get this PR merged, I can spend some time and try to get that working as well.

For BLE enabling - you can refer to arduino/giga_r1:

zephyr/boards/arduino/giga_r1/Kconfig.defconfig

Line 6 in 7bf6e8b

if BT

and

zephyr/boards/arduino/giga_r1/arduino_giga_r1_stm32h747xx_m7.dts

Line 101 in 7bf6e8b

murata-1dx {

For Wi-Fi part, so far it is not supported for STM32H7 because of missing SDHC driver for this platform.

[rahul-arasikere]

@npal-cy I notice that the st chip has a sdmmc driver, would that not work with the wi-fi driver? I have not looked too hard at it yet.

[npal-cy]

@npal-cy I notice that the st chip has a sdmmc driver, would that not work with the wi-fi driver? I have not looked too hard at it yet.

It will not work. it should be sdhc driver. stm32 disk driver does not have needed API.

[rahul-arasikere]

Should have got all the formatting issues in the devicetree files. @erwango

[erwango]

This one was correct