Devicetree SPI node interrupt asserts

RE: Devicetree SPI node interrupt asserts

kahlenberg — Wed, 22 May 2024 11:42:48 GMT

I think I found the problem. It is not about SPI or device-tree. In ISR I am calling a function whihch calls

k_ticks_to_ms_ceil64(sys_clock_tick_get()) to get the system tick count. Normaly, sys_clock_tick_get() works in main function, but when I move it to anohter file it produces error. Anyway, it is another topic and I am opening another topic now.

RE: Devicetree SPI node interrupt asserts

kahlenberg — Tue, 21 May 2024 17:01:45 GMT

Error is coming from kernel. I am posting the code.

Device-Tree overlay:

&pinctrl {
	spi1_default_alt: spi1_default_alt {
		group1 {
			psels = <NRF_PSEL(SPIM_SCK, 1, 15)>,
					<NRF_PSEL(SPIM_MOSI, 1, 13)>,
					<NRF_PSEL(SPIM_MISO, 1, 14)>;
		};
	};

	spi1_sleep_alt: spi1_sleep_alt {
		group1 {
			psels = <NRF_PSEL(SPIM_SCK, 1, 15)>,
					<NRF_PSEL(SPIM_MOSI, 1, 13)>,
					<NRF_PSEL(SPIM_MISO, 1, 14)>;
			low-power-enable;
		};
	};
};

spi_master: &spi1 {
	compatible = "nordic,nrf-spim";
	status = "okay";
	cs-gpios = <&gpio1 12 GPIO_ACTIVE_LOW>;
	pinctrl-0 = <&spi1_default_alt>;
	pinctrl-1 = <&spi1_sleep_alt>;
	pinctrl-names = "default", "sleep";
	st25r3916b:st25r3916b@0 {
		reg = <0>;
	};
};

/{
	leds {
		led_tv: led_tv {
			gpios = <&gpio0 25 GPIO_ACTIVE_HIGH>;
			label = "Green LED TV";
		};
		
		led_ap2p: led_ap2p {
			gpios = <&gpio0 26 GPIO_ACTIVE_HIGH>;
			label = "Green LED AP2P";
		};
			irq_gpio: irq_gpio {
			gpios = <&gpio0 4 GPIO_ACTIVE_HIGH>;
			label = "Interrupt line";
		}; 
	};
		
	aliases {
		led-tv = &led_tv;
		led-ap2p = &led_ap2p;
		irq-gpio = &irq_gpio;
	};
};

prj.conf:

CONFIG_LOG=y
CONFIG_DEBUG_OPTIMIZATIONS=y
CONFIG_DEBUG_THREAD_INFO=y
CONFIG_HEAP_MEM_POOL_SIZE=16384
CONFIG_MAIN_STACK_SIZE=8192

main.c:

#define SPI_MASTER_NODE DT_NODELABEL(spi_master)

const struct device *spi_dev = DEVICE_DT_GET(SPI_MASTER_NODE);

#define INT_NODE DT_ALIAS(irq_gpio)
static const struct gpio_dt_spec irq_gpio = GPIO_DT_SPEC_GET_OR(INT_NODE, gpios, {0});

struct spi_config spi_cfg = {
    .operation = (SPI_OP_MODE_MASTER | SPI_WORD_SET(8) |
                  SPI_TRANSFER_MSB | SPI_LINES_SINGLE |
                  SPI_MODE_CPHA),
     .frequency = 400000,
    .slave = DT_REG_ADDR(DT_NODELABEL(st25r3916b)),
    .cs = {
        .gpio = SPI_CS_GPIOS_DT_SPEC_GET(DT_NODELABEL(st25r3916b)),
        .delay = T_NCS_SCLK
    }
};
static struct gpio_callback gpio_cb;

st25r3916Isr() {
 /* SPI communication for reading INT registers*/    
}


static void irq_pin_cb(const struct device *gpiob, struct gpio_callback *cb, uint32_t pins) {
    //LOG_DBG("Received interrupt %s pin %d\n", gpiob->name, pins);
    st25r3916Isr();
}

int main() {

if(!device_is_ready(irq_gpio.port) || !device_is_ready(spi_dev) || !device_is_ready(spi_cfg.cs.gpio.port)) {
    //error log
}
int err = gpio_pin_configure_dt(&irq_gpio, GPIO_INPUT);
if (err) { /*error log */}
gpio_init_callback(&gpio_cb, irq_pin_cb, BIT(irq_gpio.pin));
err = gpio_add_callback(irq_gpio.port, &gpio_cb);
if (err) { /*error log */}
gpio_pin_interrupt_configure_dt(&irq_gpio, GPIO_INT_EDGE_TO_ACTIVE);
while(true) {}
return 0;
}

RE: Devicetree SPI node interrupt asserts

Naeem Maroof — Tue, 21 May 2024 11:21:24 GMT

You need to debug where the error is coming from.

Looks like you have changed the overlay, and the error might be actually coming from the sensor (driver).

RE: Devicetree SPI node interrupt asserts

kahlenberg — Thu, 16 May 2024 19:35:13 GMT

Actually, the problem exists still. Compiling device-tree and flashing work, but as you can see in the logs, I am still getting spinlock error " during interrupt handling".
What can the problem be? I don't use any spinlock or mutex for interrupt service routine? Should I use a spinlock or mutex for interrupt? If yes, how?

RE: Devicetree SPI node interrupt asserts

Naeem Maroof — Thu, 16 May 2024 13:33:42 GMT

It should be okay as long as it is getting compiled correctly.

RE: Devicetree SPI node interrupt asserts

kahlenberg — Mon, 13 May 2024 14:59:54 GMT

Ok, in device-tree, can a node be added to spi instance only with reg = <0> (chisp-select) property? I tried and built it. But before building it, my intellisense in VSCode shows always an error, that a node must have "compatible" property. But anyway, it compiles well :)

github.com/.../nrf5340dk_nrf5340_cpuapp.overlay

RE: Devicetree SPI node interrupt asserts

Naeem Maroof — Mon, 13 May 2024 12:08:20 GMT

You can get the spi instance from the DT using zephyr Device api, e.g using DEVICE_DT_GET.

Maybe, this example could be relevant to your use-case: SPI Master Slave Sample NCS

Please look into the overlays and the source code.

Regards,
Naeem

RE: Devicetree SPI node interrupt asserts

kahlenberg — Tue, 07 May 2024 20:06:13 GMT

You are right, there was no driver for st25r3916b, I copied the driver for st25r3911b and modified (renamed files, modified Kconfig...) it for st25r3916b. But the problem is that I should not use the zephyr driver in device tree, because driver is already implemented in STM library. I want to use STM library.

So, I need to use SPI without any device driver.

Now my question is changed to: How can I use SPI without device-tree node?

RE: Devicetree SPI node interrupt asserts

Naeem Maroof — Tue, 07 May 2024 07:04:21 GMT

Is the ST driver available in zephyr?

I can not see "st,st25r3916b" in the zephyr dt-binding list.