Debugging program in external flash XIP mode STM32N6

[sai2]

Hi,

I am trying to debug STM32N6 program loaded in external flash. When I started the debugger in the FSBL project I can put a breakpoint just before it jumps into application inside external flash. It is obvious I wont be able to see the symbols since it is not loaded. So I loaded my application elf file under debug_configuration/Startup/Load Image and Symbols (See attached image). When I start my debugging session, I encounter the following error and was unable to debug even FSBL code.

Board: STM32N6 DK

STM32CubeIDE: 1.17.0

Platform: Window

project: Template_XIP_FSBL (found under Stcube package for N6).

Here is the error:

STMicroelectronics ST-LINK GDB server. Version 7.9.0
Copyright (c) 2024, STMicroelectronics. All rights reserved.

Starting server with the following options:
 Persistent Mode : Disabled
 Logging Level : 1
 Listen Port Number : 61234
 Status Refresh Delay : 15s
 Verbose Mode : Disabled
 SWD Debug : Enabled
 InitWhile : Enabled

Waiting for debugger connection...
Debugger connected
Waiting for debugger connection...
Debugger connected
Waiting for debugger connection...
Failed to read all registers from target
Failed to read all registers from target
Failed to read all registers from target
Failed to read all registers from target

Thanks

[Ghofrane GSOURI]

Hello @sai2 

First make sure to add this 

Select ST-LINK (OpenOCD) as the [Debug probe] in the debug configuration of the Template_ XIP_FSBL 

Also let's try to generate the trusted binary, this could resolve the issue .

The binary file of the application that you want  to load in the memory need to be signed. 
This operation is done using the STM32_SigningTool_CLI.exe which is included in the STM32Cubeprogrammer package. 

You an access this tool using one of two ways: 

1- Under the STM32CubeIDE installation path: (Windows example) C:\ST\STM32CubeIDE_1.17.0\STM32CubeIDE\plugins\com.st.stm32cube.ide.mcu.externaltools.cubeprogrammer.win32_2.2.0.202409170845\tools\bin\ 

2- Under the STM32CubeProgrammer path: (Windows example) C:\Program Files\STMicroelectronics\STM32Cube\STM32CubeProgrammer\bin 

To sign the binary we need to launch this command: STM32_SigningTool_CLI.exe -bin Template_XIP_AppS.bin -nk -of 0x80000000 -t fsbl -o Project-trusted.bin -hv 2.3 -dump Project-trusted.bin 

This step could be done also from within the STM32CuebIDE tool using the post build command option as described below. 

1-Right Click on Template_XIP_AppS(in Apps), [Properties]>[[C/C++ Build]> [Settings] [Build Steps]  

2- Tap on the Post-build steps this command: C:\ST\STM32CubeIDE_1.17.0\STM32CubeIDE\plugins\com.st.stm32cube.ide.mcu.externaltools.cubeprogrammer.win32_2.2.0.202409170845\tools\bin\STM32_SigningTool_CLI.exe -bin Template_XIP_AppS.bin -nk -of 0x80000000 -t fsbl -o Project-trusted.bin -hv 2.3 -dump Project-trusted.bin 

 

3-Click on [Apply], then build the Template_XIP_AppS to generate Project-trusted.bin  

after generating the trusted bin go to 

 

select add in order to add Project_trusted.bin and make sure to choose the download address for the Themplate_FSBL_XIP :0x70100000. 

 

Please check Readme under C:\Users\XXXXX\STM32Cube\Repository\STM32Cube_FW_N6_V1.1.1\Projects\STM32N6570-DK\Templates\Template_FSBL_XIP

 

 

[Kamurasi-Jordan-Arthur]

Thanks alot @Ghofrane GSOURI ,
I managed to get it to work with your approach,

Your explanation however requires a little more explanation.
I had to read into boot loaders, STM IDE debugging, and also include the symbols for the binary,
Especially given the fact that binaries don't have .text section that contains information about the binary symbols

 

[sai2]

Hi @Ghofrane GSOURI ,

 

Thank you for detail steps but unfortunately, I am still unable to debug FSBL and Application together after I followed the steps that you provided.

Here is the output from OpenOCD.

Open On-Chip Debugger 0.12.0+dev-00600-g090b431b1 (2024-09-13-19:14) [https://github.com/STMicroelectronics/OpenOCD]
Licensed under GNU GPL v2
For bug reports, read
	http://openocd.org/doc/doxygen/bugs.html
Info : Listening on port 6666 for tcl connections
Info : Listening on port 4444 for telnet connections
Info : STLINK V3J15M7 (API v3) VID:PID 0483:3754
Info : Target voltage: 3.280106
Info : clock speed 8000 kHz
Info : stlink_dap_op_connect(connect)
Info : SWD DPIDR 0x6ba02477
Info : [STM32N657X0HxQ.ap0] Examination succeed
Info : [STM32N657X0HxQ.cpu] Cortex-M55 r1p1 processor detected
Info : [STM32N657X0HxQ.cpu] target has 8 breakpoints, 8 watchpoints
STM32N657X0HxQ.cpu in Secure state
Info : [STM32N657X0HxQ.cpu] Examination succeed
Info : gdb port disabled
Info : starting gdb server for STM32N657X0HxQ.cpu on 3333
Info : Listening on port 3333 for gdb connections
Info : [STM32N657X0HxQ.cpu] external reset detected
Info : accepting 'gdb' connection on tcp/3333
[STM32N657X0HxQ.cpu] halted due to breakpoint, current mode: Thread 
xPSR: 0x81000000 pc: 0x18003a1a msp: 0x341037a0
Info : New GDB Connection: 1, Target STM32N657X0HxQ.cpu, state: halted
Info : accepting 'gdb' connection on tcp/3333
Info : New GDB Connection: 2, Target STM32N657X0HxQ.cpu, state: halted
Error: Failed to read memory at 0x70101100
Error: Failed to read memory at 0x70101100
Error: Failed to read memory at 0x70101100
Polling target STM32N657X0HxQ.cpu failed, trying to reexamine
Error: Failed to read memory at 0x70101100
Error: [STM32N657X0HxQ.cpu] Examination failed
Examination failed, GDB will be halted. Polling again in 100ms
Error: Failed to read memory at 0x70101100
Error: Failed to read memory at 0x70101100
Error: Failed to read memory at 0x70101100
Error: Failed to read memory at 0x70101100
Error: Size 2 not supported
Error: Size 2 not supported
Error: Failed to read memory at 0x70101100
Error: Size 2 not supported
Error: Size 2 not supported
Error: Failed to read memory at 0x70101100
Error: Size 2 not supported
Error: Size 2 not supported
Error: Failed to read memory at 0x70101100
Error: Failed to read memory at 0x70101100
Info : dropped 'gdb' connection
shutdown command invoked
Info : dropped 'gdb' connection

 

[sai2]

Hi @Ghofrane GSOURI,

 

UPDATE 1: I think I found the cause.

If we load the application symbols during the debugger's start-up sequence, GDB won't be able to resolve those addresses because the external flash isn't memory-mapped at that point (i.e., after a reset). I confirmed this by disabling symbol loading during start-up. Instead, I set a breakpoint just before the FSBL was about to jump to the application. Once the program counter hit that breakpoint, I manually loaded the application symbols using the add-symbol-file command in the Debugger Console. After doing this, I was able to see the source and symbol information when I hit pause.

Now, I need to figure out how to set breakpoints within the application itself. Simply placing breakpoints in the application code doesn't work at this stage.

UPDATE 2: I just need to add source dir and memory region at this point. Once we hit the breakpoint at JumpToApplication() enter following commands in Debugger console.

add-symbol-file "your application.elf location"
set directories "your application source code dir"
// check mem region in gdb
info mem

// if info mem output doesnt contains external flash regions, add those regions
// following address is for STM32N6 XSPI2
mem 0x70000000 0x7fffffff

// if above command gives error delete existing memory regions and add memory regions manually. That should follow linker script
// following command delete region 0
delete mem 0

 

[romanx]

Hi @sai2 
Did you manage to make any progress with breakpoints ?
I'm able to observe my application in free running mode, e.g. whenever I press "pause"
I'm able to see where PC lands up, but can't set a breakpoint still
I get these :

Error: Failed to set breakpoint at address: 0x7004107e.

Error: Failed to set more breakpoints

Kindly appreciate any help there
Thank you 

[JaoIndio]

Hey @romanx ,

I've been debugging N6-DK with zephyr, however I hope my approach can work for you as well.
According to this MCU tutorial  before to set any breakpoints at the application code that lands in the external memory you need to make sure FSBL layer had already successfully configured and mapped the external flash. So consider first set a breakpoint at function `BOOT_Application()` and just then try to jump to your application.

Here is the whole gdb script I've been using

# Load MCUboot (FSBL) symbols
arm-none-eabi-gdb mcuboot/zephyr/zephyr.elf

target remote :61234
monitor reset halt
# Load MCUboot (FSBL) into device
load
# Zephyr BOOT_Application equivalent
hbreak boot_go
continue
# When breakpoint hits, step a few times to ensure 
# external flash initialization completes
next
next

# ============================================================
# PHASE 2: EXTERNAL FLASH IS NOW READY - LOAD APP SYMBOLS
# ============================================================

# Add application symbols
add-symbol-file hello_world/zephyr/zephyr.elf
# Set source directories
set directories <zephyr-project-paht>/zephyr/samples/hello_world/src
# Set hardware breakpoint in YOUR code
hbreak main.c:11
# Continue into application
continue

# Debugging normally after this point...