How to start RTX application built by gcc successfully?

[tk]

Hi. I’m developing software on nRF51822 with SoftDevice s130-v2.0.1 and SDK v12.3.

I want to apply RTX to the software. In the case of building it using arm-gcc, the software stops before reaching main(). I set macro "__START=pre_main" to asm/C compiler in order to call software_init_hook() in the pre_main() function. By using gdb remote debugger, I found that the software stops at the line of atexit after __init_fini_array() in software_init_handle() of RTX_CM_lib.h.

My environments is macOS (Sierra) and arm-none-eabi-gcc version is "gcc version 6.3.1 20170215 (release) [ARM/embedded-6-branch revision 245512] (GNU Tools for ARM Embedded Processors 6-2017-q1-update)".

I examined some other arm-none-eabi-gcc versions. I also tested the case of linking library (libRTX_CM0.a) and that of compiling the RTX related source codes. But, in any cases, the results are the same.

Here is the output of gdb when I press Ctrl-C.

0x00000000 in ?? ()
Breakpoint 1 at 0x1b1a8: file ../src/app/main.c, line 39.
Resetting target
(gdb) c
Continuing.
^C
Program received signal SIGTRAP, Trace/breakpoint trap.
0x000261ea in HardFault_Handler ()
(gdb) backtrace 
#0  0x000261ea in HardFault_Handler ()
#1  <signal handler called>
#2  0x00029dca in __register_exitproc ()
#3  0x00029c6a in atexit ()
#4  0x000266ce in software_init_hook () at ../SDK/external/rtx/include/RTX_CM_lib.h:345
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

How can I make the software start normally?

[RK]

well you're in the hardfault handler so you have a hardfault. You'll need to figure out what caused it. Looks like register_exitproc() executed the code which caused the problem.

[tk]

Thanks for your comment. I don't really understand the startup mechanism. I don't have to call software_init_hook() manually, right? I remove pre_main() and -D__START macro.

Then, I go forward a bit... This time, I reach the final step of osKernelStart() in rt_CMSIS.c. It seems that error occurs when calling __svcKernelStart(). I found that the program is in SVC_Handler().

Resetting target
(gdb) c
Continuing.
^C
Program received signal SIGTRAP, Trace/breakpoint trap.
0x0002625c in SVC_Handler ()
(gdb) bt
#0  0x0002625c in SVC_Handler ()
#1  <signal handler called>
#2  0x00027498 in __svcKernelStart () at ../SDK/external/rtx/source/rt_CMSIS.c:485
#3  osKernelStart () at ../SDK/external/rtx/source/rt_CMSIS.c:612
#4  0xd7f001d0 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

By using Keil, the program starts normally. What should I do in the case of using gcc?

[Jørgen Holmefjord]

Which example are you basing this on? Could you upload the Makefile you use?

[tk]

Sorry, it's a private project, so I cannot upload it. I cannot find an example of RTX with armgcc. Is there any example?

[Jørgen Holmefjord]

I'm not aware of any recent examples of RTX with GCC that include a softdevice, this is why I asked you to upload the Makefile. In the blinky_rtx example in SDK 12.3.0, there is a GCC project. There is also a GCC project for ble_app_hrs_rtx in SDK 10. Could you upload the Makefile in MyPage? Then the information within will be keept confidential.