FDS_ERR_NOT_FOUND after two fds_record_update calls fast after eachother

[Gert]

I am using the FDS library to write a struct to flash. Lately I sometimes  get  FDS_ERR_NOT_FOUND in the event handler as a result of a fds_record_update call.

I've only seen this happen when two update happen right after eachother.  Most of the times this goes well, but when it doesn't the first one always succeeds and the second gives FDS_ERR_NOT_FOUND.

Has anyone experienced this kind of behaviour?

I am using SDK 14.2 and softdevice S332

[Hung Bui]

Hi Gert, 

I'm not sure what could be wrong here but when did you receive FDS_ERR_NOT_FOUND ? Was it the return code when you called fds_record_update() or it's the status when you receive the call back event ? 

Do you have a way to reproduce it or it happened seldom? 

Could you make an example showing the issue so we can test here ? 

[Gert]

Its the result variable in the fds_evt_t struct.

When trying to reproduce it happens about 1 in 4/5 times

[Hung Bui]

Could you do a quick check which function returning FDS_ERR_NOT_FOUND  ? We have about 5 functions that return that error code. 

Is it possible to provide us a minimum example that show the issue?

[Gert]

static void fds_evt_handler(fds_evt_t const * p_evt)
{
    NRF_LOG_INFO("Flash: event: %s received (%s)",
    fds_evt_str[p_evt->id],
    fds_err_str[p_evt->result]);

[Hung Bui]

I understand, but we need to find which exact function in the fds library send this FDS_ERR_NOT_FOUND  as the result for the event. 

[Gert]

The error is returned in function write_execute in fds.c

switch (p_op->write.step)
{
case FDS_OP_WRITE_FIND_RECORD:
{
// The first step of updating a record constists of locating the copy to be deleted.
// If the old copy couldn't be found for any reason then the update should fail.
// This prevents duplicates when queuing multiple updates of the same record.
desc.p_record = NULL;
desc.record_id = p_op->write.record_to_delete;

if (!record_find_by_desc(&desc, &page))
{
return FDS_ERR_NOT_FOUND;
}

stacktrace:
}

[Hung Bui]

Hi Gert,

Sorry for the late response. We need some more debug information, for example checking why record_find_by_desc() return not found. There must be something happened in between the 2 updates. 

How often do you see the error ? Is it easy to reproduce ? Anyway that we can reproduce the issue here ? 

If you can printout more log from record_find_by_desc() it would be very useful.