Re: [PATCH] gdb/arm: Terminate frame unwinding in M-profile lockup state

[Torbjorn SVENSSON <torbjorn.svensson@foss.st.com>]

On 2022-10-16 15:59, Tomas Vanek wrote:
> Torbjorn,
> 
> thanks for the fast reply.
> 
> On 2022-10-16 10:13, Torbjorn SVENSSON wrote:
> 
>> >/  #0  0xeffffffe in ?? () />/  #1  0x0c000a9c in HardFault_Handler 
>> () />/      at 
>> />/C:/dvl/stm32l5trustzone/GPIO_IOToggle_TrustZone/Secure/Src/stm32l5xx_it.c:99 />/  #2  0x2002ffd8 in ?? () />/  Backtrace stopped: previous frame identical to this frame (corrupt />/stack?) />/  (gdb) />/---------------- />/The frame #1 is at correct PC taken from LR, #2 is a total nonsense. /
>> I think your assumption here is wrong as 0xeffffffe is a valid value for
>> Cortex-M33 with TrustZone (FNC_RETURN with S=0). See more on the comment
>> below.
> 'B3.17 Function returns from Non-secure state' states:
> The FNC_RETURN value is:
> ... the original is a binary value in a table, hard to convert to plain 
> text ...
> so if I convert correctly
> 0xfefffffe or 0xfeffffff depending on S bit.
> 
> Starts with 0xfe not 0xef as the lockup magic!

Ohh, right, my bad!

> 
>> Reading B3.33 (revision B.m) in the ARMv8-M documentation about lockup
>> state, it's written that DHCSR.S_LOCKUP=1 AND PC=0xeffffffe. My
>> interpretations that the PC value alone is not enough.
> 
> Hmm, how would we read DHCSR then?
> IMO the PC value is enough.

Reading DHCSR should be similar to FPCCR and FPCAR.
According to the ARMv8-M manual, DHSCR should be located at 0xE000EDF0 
as a 32-bit read/write register and bit 19 is S_LOCKUP.

> 
> Tomas
> 

Kind regards,
Torbjörn