MPC5748G EVB is dead after updating linker scripts by extending user code space starting from 0x01000000 to 0x00F94000.

[keen8825]

Target board: DEVKIT, SPC5748GMKU6;
IDE: S32DS for Power Architecture, Version: 2017.R1 Build id: 171018;
Debugger: OpenSDA;

Issue description: I changed the default generated .ld file by extending the user code space:
MEMORY
{

flash_rchw : org = 0x00F90000, len = 0x4
cpu0_reset_vec : org = 0x00F90000+0x10, len = 0x4
cpu2_reset_vec : org = 0x00F90000+0x04, len = 0x4

m_text : org = 0x00F94000, len = 1328K // My new setting
/*m_text : org = 0x01000000, len = 1024K*/ // Default setting
m_data : org = 0x40000000, len = 192K
}

Result: right after the changing, the compling and debuging can all work, after stop debugging and re-power the board, the program do not run any more, and I tried to debug again, the board can no longer be connected, and S32DS always poped up:
1. The target may have entered reset escalation. Please power cycle the board.
2. Warning: Device is not detected or may be censored. Please check connections or please enter password in GDB server parameters within debug configurations using the flag -uncensorpassword=n, then restart the debug session.

For more details, pls. check the attached screenshot and .ld file.

Hope for response soon, appreciate.

P.S.

@Martin Kovar, Hi, just compared with your answer https://community.nxp.com/message/1046716?commentID=1046716?sr=search&searchId=111b3e54-fc68-41c7-9e..., the flash space starting from 0x00F94000 is applicable, right?

jirikral Hi, I saw your answer Unable to start debug session on MPC5748G board. download stops at 98%   but unfortunately PROGPPCNEXUS seem not downable unless pay for it. Any other solutions?

[keen8825]

Hi there, 
As the initiator of this ticket, think I'd better give a conclusion for nice ending here.

First, @Martin Kovar sent me a tool to unlock the board via erasing whole flash memory, as mentioned in https://community.nxp.com/thread/494855, unfortunately my board was always failed to be connected. I think if no connection failure, this problem may ended here. But I would like to share what I did next, just hope to be useful from another point of view.

Then I gave up this EV board and turned to my own board, here is what I responded to Martin by email:

Other things to share with you that I also flashed the same .elf to our company newly built board (with MPC5746C mounted) by iSystem, the program cannot run neither, but the board can be connect and downloaded again, thanks to this, I could be able to investigate the root reason. Finally I see, after changed the .ld file, my application code are linked starting from the new space, which means address of _start()  also changed, but in my boot header structure, CPU0 Reset Vector is statically assigned to 0x0100_0000, as the result, the BAF could never find my _start() on 0x0100_0000, when it comes to 0x0100_0000, the program ran into an unknown area. After assigning the name of _start() to CPU0 Reset Vector rather than a constant address, the problem was solved. BTW, this part of code is from S32DS MPC5746C sample code. I suppose, NXP wants us only use the code flash block from 0x0100_0000 to 0x010FFFFF, and never expected us using the data flash block for storing code, like starting from 0x00F94000?

That's it, good luck.

Pls. find me if you want more discussion by Wechat "uncleHaoQi".

[martin_kovar]

Hello,

I sent you answer through case 00210397. Because of software sending, please continue through the internal ticket.

Thank you for understanding.

Regards,

Martin

[sohyunjang]

Hi. I am experiencing the same problem.

Is there a solution?

I use P&E Multilink not OpenSDA.

[steven_zhang]

i got the similar massage as below when i tries to debug my SPC5604 BOARD:

 Warning: Device is not detected or may be censored. Please check connections or please enter password in GDB server parameters within debug configurations using the flag -uncensorpassword=n, then restart the debug session

can you send me the answer ？