Hello,
I'm debugging the following piece of code that is executed as part of the fault handler:
currentExceptionData.data.CFSR = SCB->CFSR;
currentExceptionData.data.HFSR = SCB->HFSR;
currentExceptionData.data.DFSR = SCB->DFSR; //TODO: Not copied correctly!
currentExceptionData.data.MMFAR = SCB->MMFAR; //TODO: Not copied correctly!
currentExceptionData.data.BFAR = SCB->BFAR; //TODO: Not copied correctly!
The target CPU is STM32L4P5, the code is built using IAR studio. In order to get io the fault handler, I execute an erroneous code that fires the Usage Fault.
I am able to read (via the code) correctly the CSFR and HFSR values only. The DFSR is read as 0, the MMFAR and BFAR registers are read as their addresses. I compare these values to the register values read by debugger.
The underlying assembly code is:
currentExceptionData.data.DFSR = SCB->DFSR; //TODO: Not copied correctly! was 00000002 then 00000003
0x805'839e: 0xf10d 0x0015 ADD.W R0, SP, #21 ; 0x15
0x805'83a2: 0x4910 LDR.N R1, [PC, #0x40] ; DFSR
0x805'83a4: 0x6809 LDR R1, [R1]
0x805'83a6: 0x6001 STR R1, [R0]
Is there any reason that only part of SFRs can be readable from within the fault handler?
I also tried using the inline assembly, for example:
register uint32_t generalRegister;
generalRegister = 0xE000ED30;
__ASM volatile ("LDR %0, [%1]\n" : "=r" (generalRegister) : "r" (generalRegister) );
currentExceptionData.data.DFSR = generalRegister;
It worked in single step disassembly debugging for DFSR only, nope for MMFAR/BFAR.
Any ideas?
