Hi,
I was hoping someone could help me shed light on an issue I'm having.
The problematic code
uint32_t eeIsFlexNvmPartitioned(void)
{
/*
"DEPART field contains 0xF for non-partitioned devices."
*/
return ((SIM->FCFG1 & SIM_FCFG1_DEPART_MASK) != SIM_FCFG1_DEPART_MASK);
}
void eePartitionFlexNvm(void)
{
while ((FTFC->FSTAT & FTFC_FSTAT_CCIF_MASK) != FTFC_FSTAT_CCIF_MASK)
{
/* Wait for command execution to finish. */
}
/*
See S32K-RM 36.5.9.1.3 Command execution and error reporting.
*/
if (FTFC->FSTAT & (FTFC_FSTAT_ACCERR_MASK | FTFC_FSTAT_FPVIOL_MASK))
{
FTFC->FSTAT |= FTFC_FSTAT_ACCERR_MASK | FTFC_FSTAT_FPVIOL_MASK;
}
/*
See AN11983 and S32K-RM 36.4.4.1.6.1
*/
FTFC->FCCOB[3] = 0x80; /* FCCOB0: Selects the PGMPART command */
FTFC->FCCOB[2] = 0x00; /* FCCOB1: No CSEc operation */
FTFC->FCCOB[1] = 0x00; /* FCCOB2: No CSEc operation */
FTFC->FCCOB[0] = 0x00; /* FCCOB3: FlexRAM loaded with valid EEPROM during reset sequence */
FTFC->FCCOB[7] = 0x02; /* FCCOB4: EEPROM data set size code: EEESIZE = 2 (4 kB) */
FTFC->FCCOB[6] = 0x03; /* FCCOB5: FlexNVM Partition code: DEPART = 3 (Data flash: 32 kB, EEPROM backup: 32 kB) */
/*
Execute command.
*/
FTFC->FSTAT |= FTFC_FSTAT_CCIF_MASK;
/*
This line causes a busfault.
*/
while ((FTFC->FSTAT & FTFC_FSTAT_CCIF_MASK) != FTFC_FSTAT_CCIF_MASK)
{
/* Wait for command execution to finish. */
}
}
Application code only calls eePartitionFlexNvm when eeIsFlexNvmPartitioned returns 0. The code is run from bank 0 P-flash. I am operating under the assumption that programming/partitioning bank 1 from bank 0 is OK.
The problem
When the application is run, the last line of code in eePartitionFlexNvm causes the following fault:
BusFault: A bus fault has occurred during instruction prefetching.
This also seems to occur when not debugging. For some reason, debugging and using step-by-step debugging seems to work well.
Other remarks
- I have tried setting OCM1 in OCMDR0 and OCMDR1 to 0x3 before the command execution to no avail.
- Executing the code from RAM seems to solve the issue, but I was under the impression this would not be necessary, and feels like a work-around rather than a proper fix.
Questions
- What is the mechanism behind this fault? I thought programming FlexNVM/D-flash from P-flash should not be a problem.
- I saw a similar problem on StackOverflow where the OT suggested "Disabling caching by writing LMEM->PCCRMR = 0;" solved their problem. Is this a valid solution? I haven't specifically enabled caching. I also do not see anything mentioned about this in AN11983 or the sample project flash_partitioning_s32k144.
- How can I solve this problem?
Kind regards,
Joey
