[hal] SPI: fixes threading safty issue

[XuGuohui]

Problem

hal_spi_acquire() may fail to acquire mutex and when the caller doesn't check the result, it may corrupt stack.

Solution

Use os_mutex_recursive_lock() in hal_spi_acquire() when timeout is not specified or is non-zero.

Steps to Test

1. Install MSoM on Muon
2. Connect to the Particle cloud over Etehrnet only while keeping accessing the Lora module (or the Muon's on-board IO expander).

Example

#include "Particle.h"
#include "mcp23s17.h"

SYSTEM_MODE(SEMI_AUTOMATIC);

Serial1LogHandler logHandler(115200, LOG_LEVEL_ALL);

STARTUP (
    System.setPowerConfiguration(SystemPowerConfiguration()
        .powerSourceMinVoltage(3880)
        .powerSourceMaxCurrent(3000)
        .batteryChargeVoltage(4200)
        .batteryChargeCurrent(900)
        .auxiliaryPowerControlPin(D7)
        .interruptPin(A7)
        .feature(SystemPowerFeature::PMIC_DETECTION));

    System.enableFeature(FEATURE_ETHERNET_DETECTION);

    if_wiznet_pin_remap remap = {};
    remap.base.type = IF_WIZNET_DRIVER_SPECIFIC_PIN_REMAP;
    remap.cs_pin = A3;
    remap.reset_pin = PIN_INVALID;
    remap.int_pin = A4;
    if_request(nullptr, IF_REQ_DRIVER_SPECIFIC, &remap, sizeof(remap), nullptr);
);

void setup() {
    Ethernet.on();
    Ethernet.connect();
    waitUntil(Ethernet.ready);
    Log.info("Local IP: %s", Ethernet.localIP().toString().c_str());
    Particle.connect();
    waitUntil(Particle.connected);
    Log.info("Cloud connected");
}

void loop() {
    delay(1000);
    Log.info("Local IP: %s", Ethernet.localIP().toString().c_str());
    uint8_t temp;
    Mcp23s17::getInstance().readRegister(0x00, &temp);
}

References

N/A

---

Completeness

• User is totes amazing for contributing!
• Contributor has signed CLA (Info here)
• Problem and Solution clearly stated
• Run unit/integration/application tests on device
• Added documentation
• Added to CHANGELOG.md after merging (add links to docs and issues)

[avtolstoy]

This still needs to be supported. If conf && conf->timeout == 0 we need to trylock due to wiring SPIClass::trylock().

[XuGuohui]

Same for Gen4?

[avtolstoy]

Yup.

[avtolstoy]

But again, only if config is provided and timeout is 0.