Debugging NuMaker-M2351 not working

[danielRep]

Hi,

I recently started to familiarize with the NuMaker-M2351 board. Since I normally prioritize open-source tools, I was glad to see that Nuvoton had a fork of OpenOCD that I could use to flash and debug the platform. However, I have been facing some issues which I hope you can help me to solve.

I start the openocd bridge by executing the following command:
./src/openocd -f tcl/interface/nulink.cfg -f tcl/target/numicroM23_NS.cfg

Then I start the gdb session, connect to the openocd port, specify the binary and perform a reset
(gdb) target remote :3333
(gdb) file /path/to/file.elf
(gdb) monitor reset init

After I query OpenOCD to list the flash module:
(gdb) monitor flash banks

Ouput:

0 : NuMicro.flash_aprom (numicro) at 0x00000000, size 0x10080000, buswidth 0, chipwidth 0
1 : NuMicro.flash_data (numicro) at 0x0001f000, size 0x00000000, buswidth 0, chipwidth 0
2 : NuMicro.flash_ldrom (numicro) at 0x00100000, size 0x00001000, buswidth 0, chipwidth 0
3 : NuMicro.flash_config (numicro) at 0x00300000, size 0x00000010, buswidth 0, chipwidth 0

0 APROM: 512KB
1 actualy I don't know what region is this.
2 LDROM: 4KB
3 CONFIG-BLK: 32B

This are my binary file to be loaded section headers:

After I erase the flash:
(gdb) monitor numicro M2351_erase

numicro M2351_erase complete

Then I try to load the binary:
(gdb) load

Loading section .text, size 0x10da0 lma 0x10000
Loading section .rodata, size 0x7014 lma 0x20da0
Loading section .ARM.exidx, size 0x8 lma 0x27db4
Loading section .gnu.sgstubs, size 0x80 lma 0x28000
Loading section .data, size 0x9e8 lma 0x28800
Start address 0x102bc, load size 100388
Transfer rate: 49017 KB/sec, 10038 bytes/write.

I place a hardware breakpoint on the address of the first instruction of the Reset_Handler and execute the command continue
(gdb) hb *0x000102bd

What I would expected now is that the program to be debugged from now on, but it seems that the CPU is locked at least from the ouput from gdb:

Can you help me to figure out what is going on?

[ZaleYu]

Hi,
The numicroM23_NS.cfg file is used by non-secure program. If you tried to develop a non-secure program, a secure program must be downloaded in advance. Have you prepared any secure program and download it into your M2351 target chip before developing your non-secure program?

Sincerely,
Zale Yu

[danielRep]

Hi,
The numicroM23_NS.cfg file is used by non-secure program. If you tried to develop a non-secure program, a secure program must be downloaded in advance. Have you prepared any secure program and download it into your M2351 target chip before developing your non-secure program?

Sincerely,
Zale Yu

Hi Mr Yu

I add a typo, and I am using the numicroM23.cfg file since I am trying to flash my secure program. Meanwhile I have investigate a little bit further and I did the following:

Started openocd:
./src/openocd -f tcl/interface/nulink.cfg -f tcl/target/numicroM23.cfg

Started GDB and executed the following commands:
target remote :3333
monitor reset init
monitor numicro M2351_erase
monitor flash write_image /path/to/file.elf 0x00010000

Output:

However, if I try to read the flash memory, it seems it wasn't flashed:

Any idea Mr Yu?

[ZaleYu]

Hi Mr. Oliveira:
Could you download the NuEclipse IDE, as follows, and read chapter 3.2 in the user manual.
https://www.nuvoton.com/support/tool-and-software/software/dirver/?__locale=zh_TW

Try to create a M2351 secure project from New Project Wizard. Then build it and download the binary to the target chip.
I want to make sure whether your environment is fine or not. Thanks.

Sincerely,
Zale Yu

[danielRep]

Well, Mr Yu after writing the last reply I had an idea regarding the offset that I have passed during the flash command. Since I am using an elf file I think I don't need that offset, and now the code appears in the expected address:

Previously it was flashed on 0x00020000.

Let me continue to explore this setup, and if I don't find any more problems, I will close this issue.

[ZaleYu]

Hi Mr. Oliveira:
That is great. Don't mind and keep going.
Wish you have a good development experience with M2351 chip.

Sincerely,
Zale Yu

[danielRep]

Everything running smoothly.
More details here.
Closed.