Improvement SPI stability/recoverability The core part of this commit comes from careful code reading. The core is the swap of Scheduler_Request_sync_with_SPI(SPI_SYNC_METHOD_SOFT) and SPI_Start_single_TxRx_with_Master(). This code is sitting in an if-clause that is triggered on SPI comms failure. Instead of blindly trying to communicate again (which will very likely fail again), first try to reset the comms link, and then try to communicate again. That simply makes more sense in this case. This is heavily tested, on 2 simple dives, and 5 very long deco schedules from the simulator (10+ hour deco's), and a lot of small simulated dives (upto 2h runtime). Of all these tests, only one long session failed after 9 out of 11h runtime. Analyzing that one failure, suggests that the RTE is looping in some error handler, which (obviously) results in a SPI comms failure as a result. I consider this not part of this change. Additionally, some more cleanup is done in this code. Signed-off-by: Jan Mulder <jlmulder@xs4all.nl>

/* Define to prevent recursive inclusion -------------------------------------*/
#ifndef PRESSURE_H
#define PRESSURE_H

#include <stdint.h>

uint8_t init_pressure(void);
uint8_t pressure_update(void);

uint8_t is_init_pressure_done(void);

void  pressure_sensor_get_pressure_raw(void);
void  pressure_sensor_get_temperature_raw(void);
void pressure_calculation(void);

float get_temperature(void);
float get_pressure_mbar(void);
float get_surface_mbar(void);

void init_surface_ring(void);
void update_surface_pressure(uint8_t call_rhythm_seconds);

uint32_t demo_modify_temperature_and_pressure(int32_t divetime_in_seconds, uint8_t subseconds, float ceiling_mbar);

#endif /* PRESSURE_H */