corrupted backtrace using thread aware debugging

[xhpohanka]

I'm trying to find an issue in our zephyr application and I would need to check where the threads currently are during debugging.
I have the DEBUG_THREAD_INFO and using openocd I can display them

(gdb) i threa
  Id   Target Id                                                                     Frame 
  5    Thread 604003216 "SensorTask_tid" (Name: SensorTask_tid, prio:1,useropts:0)   0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
  6    Thread 604008792 "rx_q[0]" (Name: rx_q[0], prio:-1,useropts:0)                0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
  7    Thread 603983776 "stm_eth" (Name: stm_eth, prio:-14,useropts:2)               0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
  8    Thread 604008488 "net_mgmt" (Name: net_mgmt, prio:-1,useropts:0)              0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
  9    Thread 604009264 "tcp_work" (Name: tcp_work, prio:-14,useropts:0)             0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
  10   Thread 604010368 "sysworkq" (Name: sysworkq, prio:-1,useropts:0)              0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
  11   Thread 604003768 "logging" (Name: logging, prio:14,useropts:0)                0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
* 12   Thread 604009792 "idle" (Name: idle, prio:15,useropts:1)                      z_spin_lock_valid (l=l@entry=0x24005128 <log_buffer+20>) at /zephyrproject/zephyr/kernel/thread.c:912
  13   Thread 604010064 "main" (Name: main, prio:0,useropts:3)                       0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
  16   Thread 604002264 "LongPolling" (Name: LongPolling, prio:1,useropts:0)         0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
  17   Thread 604001704 "scom_service" (Name: scom_service, prio:1,useropts:0)       0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
  19   Thread 604002944 "mqtt" (Name: mqtt, prio:0,useropts:0)                       0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107

Unfortunately it seems that backtrace of inactive thread is not reconstructed correctly

(gdb) thre 17
[Switching to thread 17 (Thread 604001704)]
#0  0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
107     __asm__ volatile(
(gdb) bt
#0  0x90019ce0 in arch_irq_unlock (key=0) at /zephyrproject/zephyr/include/zephyr/arch/arm/asm_inline_gcc.h:107
#1  arch_swap (key=key@entry=0) at /zephyrproject/zephyr/arch/arm/core/cortex_m/swap.c:43
#2  0x900427a6 in z_swap_irqlock (key=0) at /zephyrproject/zephyr/kernel/include/kswap.h:209
#3  z_swap (key=..., lock=0x24019de4 <sched_spinlock>) at /zephyrproject/zephyr/kernel/include/kswap.h:220
#4  z_tick_sleep (warning: Invalid state, unable to determine sp alias, assuming msp.
ticks=<optimized out>) at /zephyrproject/zephyr/kernel/sched.c:1465
#5  0x00000000 in ?? ()

Is there a possibility how to check where the scheduler switched my thread 17?

I'm testing this with Zephyr 3.5.0 on stm32h750, the code is running from external quadspi flash memory mapped from 0x90000000.