uart_configure crash with ambiq apollo3

[fariouche]

Describe the bug
I'm getting an crash in uart_configure with the apollo3 board.
I've enabled mcuboot, and I can see mcuboot debug traces over the uart.... then when it is executing the zephyr app, I see the zephyr boot header and then when it reaches my call to uart_configure it crashes.
I haven't tested without mcuboot yet.

The debugger shows the following stack trace:

#0  arch_system_halt (reason=25)
    at zephyr/kernel/fatal.c:30
#1  0x00047e38 in k_sys_fatal_error_handler (reason=<optimized out>, 
    esf=<optimized out>)
    at zephyr/kernel/fatal.c:44
#2  0x00038310 in z_fatal_error (reason=<optimized out>, esf=<optimized out>)
    at zephyr/kernel/fatal.c:119
#3  0x00040a0c in z_arm_fatal_error (reason=<optimized out>, 
    esf=esf@entry=0x1001d940 <z_interrupt_stacks+2048>)
    at zephyr/arch/arm/core/fatal.c:86
#4  0x00028912 in z_arm_fault (msp=<optimized out>, psp=<optimized out>, 
    exc_return=<optimized out>, callee_regs=<optimized out>)
    at zephyr/arch/arm/core/cortex_m/fault.c:1107
#5  0x000289dc in z_arm_usage_fault ()
    at zephyr/arch/arm/core/cortex_m/fault_s.S:102
#6  <signal handler called>
#7  pl011_disable (dev=0x48f80 <__device_dts_ord_58>)
    at zephyr/drivers/serial/uart_pl011_registers.h:42
#8  pl011_runtime_configure_internal (dev=0x48f80 <__device_dts_ord_58>, 
    cfg=0x1001ed48 <z_main_stack+3976>, disable=<optimized out>)
    at zephyr/drivers/serial/uart_pl011.c:217
#9  0x00022ae8 in z_impl_uart_configure (cfg=0x1001ed48 <z_main_stack+3976>, 
    dev=0x48f80 <__device_dts_ord_58>)
    at zephyr/include/zephyr/drivers/uart.h:648
#10 uart_configure (cfg=0x1001ed48 <z_main_stack+3976>, 
    dev=0x48f80 <__device_dts_ord_58>)
    at build/zephyr/include/generated/zephyr/syscalls/uart.h:157

So it is getting a fault when calling pl011_disable... I verified and dts_ord_58 is the uart0 in the dts.

I'm getting the same crash with all calls to uart_irq_xxx functions too.

The exact same code works with other zephyr boards...

Hope this will ring a bell to someone here! Meanwhile I will try to reproduce the issue with a zephyr sample app

To Reproduce
I haven't tested with hello world sample app yet.

1. mkdir build; cd build
2. cmake -DBOARD=board\apollo3
3. make
4. See error

Expected behavior
uart is working

Impact
showstopper obviously

Logs and console output

*** Using Zephyr OS build v4.0.0-2-gad0afbde8448 ***
I: Starting bootloader
I: Primary image: magic=unset, swap_type=0x1, copy_done=0x3, image_ok=0x3
I: Secondary image: magic=bad, swap_type=0x0, copy_done=0x0, image_ok=0x0
I: Boot source: none
I: Secondary image of image pair (0.) is unreachable. Treat it as empty
I: Image index: 0, Swap type: none
I: Bootloader chainload address offset: 0xc000
I: Image version: v1.3.0
I: Jumping to the first image slot

*** Booting Zephyr OS build v4.0.0-2-gad0afbde8448 ***

Environment (please complete the following information):

• Linux
• zephyr 4.0.0

Additional context
This is the resulting dts of this build (from zephyr.dts)

uart0: uart@4001c000 {
          compatible = "ambiq,uart", "arm,pl011";
          reg = < 0x4001c000 0x1000 >;
          interrupts = < 0xf 0x0 >;
          interrupt-names = "UART0";
          status = "okay";
          clocks = < &uartclk >;
          ambiq,pwrcfg = < &pwrcfg 0x8 0x80 >;
          zephyr,pm-device-runtime-auto;
          current-speed = < 0xf4240 >;
          pinctrl-0 = < &uart0 >;
          pinctrl-names = "default";
          label = "uart0";
          phandle = < 0x4 >;
  };

[fariouche]

From what I see, it crashes in pl011_disable()...

What I do not understand is that the call get_uart(dev) is returning address 0x4001C000 which is good and access is working fine before.... I see that the reason of the crash is read memory access at same address that worked before.

Reading 76 bytes @ address 0x4001C000
WARNING: Failed to read memory @ address 0x4001C000
Cannot access memory at address 0x4001c000

maybe some Power management disabled the uart?

[fariouche]

I confirm, the issue happens after the call to (void)pm_device_runtime_auto_enable(dev); in init.c

Disabling CONFIG_PM_DEVICE_RUNTIME makes it run...

[AlessandroLuo]

@fariouche I tried hello world sample with PM_DEVICE_RUNTIME enabled, and did not reproduce your issue.
Have you tried enlarge your stack size?

[fariouche]

this is strange. I remember that I had to comment line that calls am_hal_cachectrl_control in apollo3/hal/am_hal_pwrctrl.c as to was crashing when booting.
Maybe this is the reason then? I'm not sure why it crashes here (it crashed when calling set_LPMMODE, but i wanted to investigate a bit more before submitting the issue)

[fariouche]

I got some time to debug a bit more.
I uncommented the call to am_hal_cachectrl_control(AM_HAL_CACHECTRL_CONTROL_LPMMODE_RECOMMENDED, 0); to see the crash again.
It crashed in AM_HAL_CACHECTRL_CONTROL_LPMMODE_RESET when executing set_LPMMODE(AM_HAL_CACHECTRL_FLASHCFG_LPMMODE_NEVER) (I lost the debugging session while executing step by step, when entering am_hal_flash_store_ui32 when pFunc is called (value 0x10010f21 <SRAM_write_ui32+1>) with pui32Address== 0x40018004 and ui32Value==0x1751)
If I attach after the crash, I see that it is inside z_arm_usage_fault:

#0  arch_system_halt (reason=20)
    at /home/fariouche/zephyr/kernel/fatal.c:30
#1  0x00050d7a in k_sys_fatal_error_handler (reason=<optimized out>, 
    esf=<optimized out>)
    at /home/fariouche/zephyr/kernel/fatal.c:44
#2  0x0003a0c4 in z_fatal_error (reason=<optimized out>, esf=<optimized out>)
    at /home/fariouche/zephyr/kernel/fatal.c:119
Read 4 bytes @ address 0x00029A52 (Data = 0xED00E7D1)
Reading 64 bytes @ address 0x1001D780
Read 4 bytes @ address 0x00029A80 (Data = 0xBF00BD01)
Reading 64 bytes @ address 0x00029A40
#3  0x00045bc4 in z_arm_fatal_error (reason=<optimized out>, 
    esf=esf@entry=0x1001d784 <z_interrupt_stacks+2052>)
    at /home/fariouche/zephyr/arch/arm/core/fatal.c:86
#4  0x00029a52 in z_arm_fault (msp=<optimized out>, psp=<optimized out>, 
    exc_return=<optimized out>, callee_regs=<optimized out>)
    at /home/fariouche/zephyr/arch/arm/core/cortex_m/fault.c:1107
#5  0x00029a80 in z_arm_usage_fault ()
    at /home/fariouche/zephyr/arch/arm/core/cortex_m/fault_s.S:102
Read 4 bytes @ address 0x1001D764 (Data = 0x0003A0C5)
Read 4 bytes @ address 0x10013E30 (Data = 0x00000000)
#6  <signal handler called>
#7  0x10013e30 in _kernel ()
Read 4 bytes @ address 0x1001D748 (Data = 0x40018004)
Reading 64 bytes @ address 0x1001CF80
#8  0x1001d748 in z_interrupt_stacks ()
Read 4 bytes @ address 0x1001D748 (Data = 0x40018004)
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

I tried to push the main stack to 5K, but still the same issue.

I suppose I have an issue with the low power mode that is also making the uart crash if I comment this function.
I'm using a apollo3 blue dev kit (AMA3B1KK-KBR EVB)

What could be the issue?

[fariouche]

Any idea on why I get this crash?
Thanks!