AM2434: Execute custom code from RAM and existing (running) application

Tommy Beckmann

Part Number: AM2434
Other Parts Discussed in Thread: SYSCONFIG

Dear TI-Experts,

on the AM243x series, is it possible to run an application (.appimage) that receives another application (e.g. via a Ethernet) - write this to internal RAM and then execute this "new" second application by:

a. Triggering a CPU reset (software reset) or

b. Switching the program counter (PC) to the new .appimage

and therefore start the second application?

(I am aware, that the sysconfig must fit and no reinitialize of hardware must occur)

Background: I need to inject a toolbox into RAM (only) within our production - I this environment it is only possible to connect the device via M12 plug (Industrial Ethernet).

Thanks in advance and best regards,

Tommy

over 2 years ago

0 Aakash Kedia over 2 years ago

TI__Mastermind 26145 points

Hi Tommy Beckmann,

You can use BOOTP TFTP to transfer the image (if you have a lwip stack) and program the same into the RAM. You need to copy the image to the required location on RAM and then assert CPU reset.

Your flow is correct and will work. Do you have any more updates from our side ?

Best Regards,
Aakash

0 Tommy Beckmann over 2 years ago in reply to Aakash Kedia

Intellectual 292 points

Hi Aakash,

thanks for your response.
For my issue: let's assume the file transfer via TFTP is working.

I reproduce all following steps using ccs (12.1) - together with AM243x Launcpad:

1. Build the gpio_led_blink example
2. Start a debug session and load the program into RAM.
3. Dump the complete MSRAM region to file using Memory Browser:

(File Type TI Data)
4. Modified the demo - replaced all code in main() through while(1){}
5. Rebuild the application, reset the board and loaded new application -> stucks in while(1) -> OK
6. Loaded the dump file from step 3. back into external RAM
7. Verified changed mem region with Memory Browser ->OK
6. Executed Processor Reset via CCS (Ctrl-Shift-R) - expect: The original application starts up.

Observed:
After reset I started up in _vecotrs():

after stepping through function _c_int00() I stop it seems that I stuck in HwiP_irq_handler() .

I also tied to dump the R5F_VECS region as well with same procedure - without success:

Am I missing any step?
Is there a software reset command for CPU only?

Thanks in advance and best regards,

Tommy

0 Aakash Kedia over 2 years ago in reply to Tommy Beckmann

TI__Mastermind 26145 points

Hi Tommy Beckmann,

I tried the same steps and I do see this to be working on my setup. Try giving the CPU reset twice. Let me know if you need any more help or send out an meeting invite on a-kedia@ti.com if you are looking at any debug calls.

Best Regards,
Aakash

0 Tommy Beckmann over 2 years ago in reply to Aakash Kedia

Intellectual 292 points

Hi Aakash,

unfortunatelly this does not resolve my issue.

I'll try to find a slot for a debug session and let you know via e-mail. You can close the thread here.

Best regards,

Tommy

0 Tommy Beckmann over 2 years ago in reply to Aakash Kedia

Intellectual 292 points

Hi Aakash,

unfortunately this does not resolve my issue.

I'll try to find a slot for a debug session and let you know via e-mail. You can close the thread here.

Best regards,

Tommy

0 Tommy Beckmann over 2 years ago in reply to Aakash Kedia

Intellectual 292 points

Hi Aakash,

unfortunately this does not resolve my issue.

I'll try to find a slot for a debug session and let you know via e-mail. You can close the thread here.

Best regards,

Tommy

0 Aakash Kedia over 2 years ago in reply to Tommy Beckmann

TI__Mastermind 26145 points

Hi Tommy,

Please send an invite suitable for both our time-zones at a-kedia@ti.com. We will close this issue together.

Best Regards,
Aakash

+1 Aakash Kedia over 2 years ago in reply to Aakash Kedia

TI__Mastermind 26145 points

Hi Tommy Beckmann,

Just for clarification. You took dump of all the sections in the linker right ? like MSRAM, TCM memory and so on..

Waiting for your invite.

Best Regards,
Aakash

0 Tommy Beckmann over 2 years ago in reply to Aakash Kedia

Intellectual 292 points

Hi Aakash,

to confirm the solution: dumping MSRAM and TCM memory solves the issue:

In combination with the function Bootloader_socCpuResetReleaseSelf(); a sw-reset is possible as expected.

Thanks for your support !

Best regards,

Tommy

Arm-based microcontrollers

Arm-based microcontrollers forum

AM2434: Execute custom code from RAM and existing (running) application