How to properly configure GPIOTE PORT events

[sst_S.Sakamoto]

Hello,

I am using the following device and development kit:

• Device: NORA-B106-00B (nRF5340)
• nRF Connect SDK: 2.0.2

Currently, I am trying to use GPIOTE PORT events to detect interrupts on pins P0.04 and P0.28.

The mask for pins that do not use GPIOTE is defined in the Devicetree as follows:

&gpio0 {
	status = "okay";
	sense-edge-mask = < 0x10000010 >; //Mask GPIOTE events other than P0.04 and P0.28
};

Also, among the pins that mask GPIOTE events are the following pins used as SPIM4.

• P0.08/SCK
• P0.09/MOSI
• P0.10/MISO
• P0.11/CSN

I have PSRAM connected to SPIM4, which is turned off when idle and turned on when in use.

Here, if you try to write data using the nrfx spim driver after the power is turned on (P0.11/CSN:High), the write will not be executed as shown below, but if you read the data immediately after, the SPIM4 works and reads the data.

(Since no data has been written, the data read will be an undefined value.)

Also, even if you change the hardware to always have the PSRAM powered on (P0.11/CSN:Always High), the above procedure will not execute the first data write.

If you comment out "sense-edge-mask" in the Device Tree, the problem of this data not being written does not occur.

(It is possible that other pins are also behaving abnormally, but I have not been able to confirm this at present.)

Questions:

1. If I mask the GPIOTE event in Devicetree as above, could it affect other peripherals such as SPIM?
2. Is there another way to properly configure GPIOTE PORT events?

Best regard,

[Sigurd Hellesvik]

Hi

Which pins are you trying to configure with sense-edge-mask?

Is there another way to properly configure GPIOTE PORT events?

You can always use nrfx directly, as seen in this sample.

[sst_S.Sakamoto]

Hello,

Thank you for your reply.

I am trying to configure edge detection for P0.04 and P0.28 with sense-edge-mask.

I thought that I could mask the GPIOTE edge detection of other pins with the device tree settings above.

The sample you introduced does not seem to use sense-edge-mask, but is it correct to understand that if you configure GPIOTE only for the pins used for interrupts, there is no need to mask other pins?

Best regard,

[sst_S.Sakamoto]

Additional information.

Below are the results of checking the GPIOTE CONFIG register when idle with "sense-edge-mask" enabled and disabled.

sense-edge-mask	enabled	disabled
GPIOTE CONFIG[7]	0x00000000	0x00011C00
GPIOTE CONFIG[6]	0x00000000	0x00020401

*GPIOTE CONFIG[5]~[0] are all 0x00000000.

[Sigurd Hellesvik]

Im still trying to understand the issue here, so Im gonna continue asking some questions:

Can you show me how you call gpio_pin_configure(), if you do?

Can you show me the callback that triggers the spim transfer?

[sst_S.Sakamoto]

Hello,

The answers to the questions are as follows:

1.How gpio_pin_configure() is called
The configuration for the pins used for interrupts (P0.04, P0.28) is as follows:

static struct gpio_callback ms_intrtc_cb_data;

gpio_dev = device_get_binding(DT_LABEL(DT_NODELABEL(gpio0)));
gpio_pin_configure(gpio_dev, 4, GPIO_INPUT | GPIO_PULL_UP | GPIO_ACTIVE_LOW);
gpio_pin_interrupt_configure(gpio_dev, 4, GPIO_INT_EDGE_TO_ACTIVE);
gpio_init_callback(&ms_intrtc_cb_data, RTC_interrupt, BIT(4));
gpio_add_callback(gpio_dev, &ms_intrtc_cb_data);


static struct gpio_callback intmems1_cb_data;

gpio_dev_mems = device_get_binding(DT_LABEL(DT_NODELABEL(gpio0)));
gpio_pin_configure(gpio_dev_mems, 28, GPIO_INPUT | GPIO_ACTIVE_LOW);
gpio_pin_interrupt_configure(gpio_dev_mems, 28, GPIO_INT_EDGE_TO_INACTIVE);
gpio_init_callback(&intmems1_cb_data, AccelMEMS_mems1_int, BIT(28));
gpio_add_callback(gpio_dev_mems, &intmems1_cb_data);

The configuration for the pins used with SPIM4 (P0.08/SCK, P0.09/MOSI, P0.10/MISO, P0.11/CSN) is as follows:

static const nrfx_spim_t m_spim4 = NRFX_SPIM_INSTANCE(4);
#define SPIM4_NODE  DT_NODELABEL(spi4)
#define D_PSRAM_SEND_CLOCK  NRF_SPIM_FREQ_32M

nrfx_spim_uninit(&m_spim4);
nrfx_spim_config_t config = NRFX_SPIM_DEFAULT_CONFIG(
	NRFX_SPIM_PIN_NOT_USED,
	NRFX_SPIM_PIN_NOT_USED,
	NRFX_SPIM_PIN_NOT_USED,
	NRF_DT_GPIOS_TO_PSEL(SPIM4_NODE, cs_gpios)
);
config.mode = NRF_SPIM_MODE_0;
config.skip_psel_cfg = true;
config.use_hw_ss = true;
config.frequency = D_PSRAM_SEND_CLOCK;

nrfx_spim_init(&m_spim4, &config, NULL, NULL); 

NRF_SPIM4->PSEL.CSN = 0x0000000B;

2.Callback that triggers the SPIM transfer
The data transfer using SPIM4 is triggered by the following callback:

nrfx_spim_xfer_desc_t xfer_desc = NRFX_SPIM_XFER_TX(
    	access_write, 
	D_APS6404L_COMMAND_OFFSET_DATA_BYTE0 + D_PSRAM_NRFX_LENGTH_MAX
);

nrfx_spim_xfer(&m_spim4, &xfer_desc, 0);

If there's any other information you need, please let me know.

Best regards,

[Jared]

Hi there,

I'll be taking over this case from Sigurd. 

I've read the post and the comments, and here some comments and questions I have:

My understanding is that you are trying to write to some external memory using SPIM peripheral, while at the same time enabling the PORT event for two other GPIOs.

It's not entirely clear for me what the issue is, but it seems like using the sense-edge-mask is effecting the SPIM transfer somehow, is this correct?

Here, if you try to write data using the nrfx spim driver after the power is turned on (P0.11/CSN:High), the write will not be executed as shown below, but if you read the data immediately after, the SPIM4 works and reads the data.

(Since no data has been written, the data read will be an undefined value.)

Here I can see data is being written on MISO and data being returned on MOSI. So this is a successful write and read? Is this with sense-edge-mask disabled?

If you comment out "sense-edge-mask" in the Device Tree, the problem of this data not being written does not occur.

(It is possible that other pins are also behaving abnormally, but I have not been able to confirm this at present.)

Here I see that data is being written on MOSI but no data is being returned on MISO. Is this with sense-edge-mask enabled?

The relation between the SPIM transfer and the PORT events interrupts is not clear, can you explain it?

Are you triggering the SPIM transfer from one of the interrupts on either P0.04 and P0.28?

Looking at your code it looks somewhat ok, but I would prefer if you checked the return value from the gpio functions and confirm that none of them returns an error. Can you do that?

sst_S.Sakamoto said:

sense-edge-mask enabled disabled GPIOTE CONFIG[7] 0x00000000 0x00011C00 GPIOTE CONFIG[6] 0x00000000 0x00020401

The value for GPIOTE CONFIG[7] is correct in that it disables the IN event and sets up PORT event for P0.28 for Lo-to-High as you do in your code.

However for GPIOTE CONFIG [6] it's not correct for some reason. For some reason MODE field is set to EVENT for P0.04 but it should be disabled.

Can you try to comment out the last part for P0.28 and then read out the CONFIG registers and se if they have changed?

static struct gpio_callback ms_intrtc_cb_data;

gpio_dev = device_get_binding(DT_LABEL(DT_NODELABEL(gpio0)));
gpio_pin_configure(gpio_dev, 4, GPIO_INPUT | GPIO_PULL_UP | GPIO_ACTIVE_LOW);
gpio_pin_interrupt_configure(gpio_dev, 4, GPIO_INT_EDGE_TO_ACTIVE);
gpio_init_callback(&ms_intrtc_cb_data, RTC_interrupt, BIT(4));
gpio_add_callback(gpio_dev, &ms_intrtc_cb_data);


/*static struct gpio_callback intmems1_cb_data;

gpio_dev_mems = device_get_binding(DT_LABEL(DT_NODELABEL(gpio0)));
gpio_pin_configure(gpio_dev_mems, 28, GPIO_INPUT | GPIO_ACTIVE_LOW);
gpio_pin_interrupt_configure(gpio_dev_mems, 28, GPIO_INT_EDGE_TO_INACTIVE);
gpio_init_callback(&intmems1_cb_data, AccelMEMS_mems1_int, BIT(28));
gpio_add_callback(gpio_dev_mems, &intmems1_cb_data);*/

best regards
Jared