lexus2k / lcdgfx

55%

master: 55%

LAST BUILD ON BRANCH fix/esp32-hal-error-checking

branch: SELECT

CHANGE BRANCH

x

• No branch selected
• circleci-project-setup
• esp8266_spi_compilation_fix
• feature/coalesce-line-runs
• feature/esp32-spi-dma
• feature/per-group-font-width
• fix/esp32-hal-error-checking
• fix/esp32-spi-host-mapping
• master
• v1.0.0_dev

Build # 33541442816

Build Type

Pull #146

github

Committed by lexus2k

Commit Message

fix(hal): report ESP-IDF failures in the ESP SPI/I2C HAL

Every esp_err_t returned by the IDF SPI and i2c calls was discarded, so a
failure to bring up the bus produced a blank display and no diagnostic at
all. That is how the SPI1_HOST regression on ESP32-S3 and ESP32-C3 stayed
hidden: spi_bus_initialize() rejected the flash bus, spi_bus_add_device()
then failed too, and the code carried on transmitting on an uninitialised
device handle.

Check and log the results of spi_bus_initialize, spi_bus_add_device,
spi_bus_free, spi_device_transmit, i2c_param_config, i2c_driver_install
and i2c_master_cmd_begin, including the host, pins and esp_err_to_name()
text needed to tell a wrong host from a pin conflict.

m_spi is now initialised to nullptr and cleared when device creation
fails; forceSpiTransfer() drops pending bytes while it is null instead of
transmitting on an invalid handle.

Verified under QEMU on esp32c3: with the bus id forced back to SPI1_HOST
the new logs report ESP_ERR_INVALID_STATE from spi_bus_initialize and
ESP_ERR_INVALID_ARG from spi_bus_add_device, naming the host; with the
correct host both succeed silently.

Pull Request Pull Request #146: fix(hal): report ESP-IDF failures in the ESP SPI/I2C HAL

Coverage Stats

1060 of 1925 relevant lines covered (55.06%)

15925.7 hits per line