SPI.beginTransaction is called twice, locks up on ESP32

[robbi5]

The begin function blocks itself, when SPI_HAS_TRANSACTION is defined and Hardware SPI is used.

In begin SPI.beginTransaction is called first at

Adafruit-PN532/Adafruit_PN532.cpp

Line 197 in 941cac4

SPI.beginTransaction(PN532_SPI_SETTING);

then sendCommandCheckAck is called which itself calles writecommand which tries to start another transaction with SPI.beginTransaction.

Adafruit-PN532/Adafruit_PN532.cpp

Line 1649 in 941cac4

if (_hardwareSPI) SPI.beginTransaction(PN532_SPI_SETTING);

On an ESP32 with the arduino-esp32 framework the SPI.beginTransaction calls the spiTransaction function which uses an SPI_MUTEX_LOCK:

https://github.com/espressif/arduino-esp32/blob/1.0.0/cores/esp32/esp32-hal-spi.c#L702-L707

This combination of calling SPI.beginTransaction twice (in begin and writecommand) causes a hang on the second invocation, because the SPI_MUTEX_LOCK is never unlocked.

---

• Arduino board: ESP32 w/ arduino-esp32 (tag: 1.0.0)

• List the steps to reproduce the problem below (if possible attach a sketch or
  copy the sketch code in too):

#define SPI_HAS_TRANSACTION
#include <SPI.h>
#define PN532DEBUG
#include <Adafruit_PN532.h>

Adafruit_PN532 nfc(18);

void setup() {
  nfc.begin(); // see debug output, hangs at "Sending:" (no bytes after this message)
}
void loop() {
}

[ladyada]

oof - can you try removing it from begin() does it work now?

[robbi5]

yep, then it doesn't hang anymore. now we're still figuring out why a timeout happens, but the locking problem is gone :)

[ladyada]

ok can you submit a PR please :)

[izedgo]

Why a PR is not accepted yet? I spent a lot of time to fix this :(

[factor8]

We are also running into this problem. Can we get @izedgo 's PR merged soon?

[ladyada]

fixed in 1.2.0+