STM32CubeIDE M4 debugging timed out on custom image.

[KnarfB]

I have a custom built buildroot image on my STM32MP157C-DK2. Used STM32CubeIDE 1.5.0 to create a M4 blinky project. Can download and run the .elf manually on the board, LED blinks okay. Engineering Mode works too. In Production Mode, I can download the .elf from IDE. Have to enter my credentials and see the download progress and the resulting .elf on the board.

But, after download, I have to enter the same credentials again and .. finally get:

Error in final launch sequence:
 
Failed to execute MI command:
 
target remote 192.168.2.44:3333
 
Error message from debugger back end:
192.168.2.44:3333: Connection timed out.
Failed to execute MI command:
target remote 192.168.2.44:3333
 
Error message from debugger back end:
192.168.2.44:3333: Connection timed out.
192.168.2.44:3333: Connection timed out.

There serial connection says:

[ 2174.385112] remoteproc remoteproc0: stopped remote processor m4
[ 2179.484162] remoteproc remoteproc0: powering up m4
[ 2179.493251] remoteproc remoteproc0: Booting fw image Hello_stm32mp157c_dk2_CM4.elf, size 2489804
[ 2179.500701] remoteproc remoteproc0: header-less resource table
[ 2179.506452] remoteproc remoteproc0: no resource table found for this firmware
[ 2179.513765] remoteproc remoteproc0: header-less resource table
[ 2179.519465] remoteproc remoteproc0: remote processor m4 is now up
-ne 
-sh: -ne: not found

Don't get, what's missing. There is a local gdbserver in the image. Is there an option to increase verbosity, i.e. to find out what the credentials are needed for 2nd time etc..?

Thanks

KnarfB

[LudovicR]

Hello @KnarfB​ ,

I think your are missing the resource_table section needed by remoteProc Linux framework:

 .resource_table :

 {

  . = ALIGN(4);

  KEEP (*(.resource_table*))

  . = ALIGN(4);

 } >RAM1_data

Could you compare your .ld file with one generated by STM32CubeIDE ?

Hope this help,

Ludovic

[KnarfB]

Hi Ludovic,

thanks for your reply. No, I didn't change .ld and that section is present in .elf (but has size 0).

Used STM32CubeIDE 1.5.0 to create a new STM32MP157C-DK2 project with "all peripherals". Just assigned LED_Y GPIO to M4.

Same behavior also with OpenAMP_TTY_echo_CM4 stock example: can download it from IDE (scp?) but debugger connection times out.

KnarfB

[LudovicR · Accepted solution]

Hi @KnarfB​ ,

Sorry for the misunderstanding, I was confused by the "custom build" sentence.

Anyway we have an issue with remoteProc framework, you should get something like :

[1033742.136156] remoteproc remoteproc0: stopped remote processor m4

[1033744.183138] remoteproc remoteproc0: powering up m4

[1033744.193205] remoteproc remoteproc0: Booting fw image dk2-mx_CM4.elf, size 2660280

[1033744.199826] remoteproc0#vdev0buffer: assigned reserved memory node vdev0buffer@10042000

[1033744.208425] virtio_rpmsg_bus virtio0: rpmsg host is online

[1033744.213738] remoteproc0#vdev0buffer: registered virtio0 (type 7)

[1033744.219753] remoteproc remoteproc0: remote processor m4 is now up

Could you double-check the version of your Linux booted on the target:

root@stm32mp1:/usr/local/projects/dk2-mx_CM4# uname -a

Linux stm32mp1 5.4.56 #1 SMP PREEMPT Wed Aug 5 07:59:52 UTC 2020 armv7l armv7l armv7l GNU/Linux

You can try to load manually your elf via the Linux console:

root@stm32mp1:/# cd /usr/local/projects/Hello_stm32mp157c_dk2_CM4/

root@stm32mp1:/usr/local/projects/Hello_stm32mp157c_dk2_CM4# ./fw_cortex_m4.sh stop

root@stm32mp1:/usr/local/projects/Hello_stm32mp157c_dk2_CM4# ./fw_cortex_m4.sh start

Ludovic

[KnarfB]

Hi @LudovicR​ ,

thanks for your answer. I could solve my issues by taking a close look at the OpenAMP_TTY_echo_CM4 example

KnarfB