/
Accessing the hardware (Kirkstone)

Accessing the hardware (Kirkstone)

Owned by Bernd Mierzowski
Last updated: 29. Aug 2024 by Hai Nam Tran
11 min read

This chapter gives a short overview of how to access the different hardware parts and interfaces from within the Linux operating system. It is written universally in order to fit all SECO Northern Europe platforms in general.

Machine type

The SECO Northern Europe i.MX6 based platforms are generally very similar and run the same OS versions. If it is needed to find out the actual hardware that the software is run on, it can be read out using the command line.

Reading detailed information in /proc/cpuinfo file provided by the kernel:

root@santaro:/# cat /proc/cpuinfo processor : 0 model name : ARMv7 Processor rev 10 (v7l) BogoMIPS : 3.00 Features : half thumb fastmult vfp edsp neon vfpv3 tls vfpd32 CPU implementer : 0x41 CPU architecture: 7 CPU variant : 0x2 CPU part : 0xc09 CPU revision : 10 processor : 1 model name : ARMv7 Processor rev 10 (v7l) BogoMIPS : 3.00 Features : half thumb fastmult vfp edsp neon vfpv3 tls vfpd32 CPU implementer : 0x41 CPU architecture: 7 CPU variant : 0x2 CPU part : 0xc09 CPU revision : 10 Hardware : Garz & Fricke SANTARO (i.MX6) Revision : 63015 Board revision : 1.2 Serial : 03515530 FNGBoot : 13.0r3797

These files can also be accessed from inside other programs using the standard open() and read() functionality.

Ethernet

If all network devices are properly configured as described in the manual, they can be used by the POSIX socket API.

There is a huge amount of documentation about socket programming available. Therefore it is not documented here.

The POSIX specification is available here.

If you have some problem in connecting to network through ethernet. Please look into the following trouble shoot link.

Serial interfaces (RS-232 / RS-485 / MDB)

Most of the serial interfaces are exported as TTY devices and thus accessible via their device nodes under /dev/ttymxc<number>. Depending on your hardware platform (and maybe its assembly option), different transceivers (RS232, RS485, MDB) can be connected to these TTY devices. See the following table for the mapping between device nodes and hardware connectors on i.MX6:

TTY device

Hardware function

TTY device

Hardware function

/dev/ttymxc0

RS-232 #1

/dev/ttymxc1

RS-232 #2

/dev/ttymxc2

MDB (Master)

/dev/ttymxc3

MDB (Slave)

RS485 can be used in half duplex or full duplex mode. This mode has to be set on the hardware (see the according hardware manual) as well as in the software. Per default, the interface is working in full duplex mode.

#include <linux/serial.h> // serial_rs485 #include <unistd.h> // close #include <fcntl.h> // O_RDWR, O_SYNC #include <sys/ioctl.h> // ioctl, TIOCSRS485 void set_rs485_half_duplex_mode() { struct serial_rs485 rs485; int fd; /* Open port */ fd = open ("/dev/ttymxc2", O_RDWR | O_SYNC); /* Enable RS485 half duplex mode */ rs485.flags = SER_RS485_ENABLED | SER_RS485_RTS_ON_SEND; ioctl(fd, TIOCSRS485, &rs485); close(fd); } int main(void) { set_rs485_half_duplex_mode(); return 0; }

For a full source code example, see the BSP folder sources/meta-guf/meta/recipes-test/ltp-guf-tests/ltp-guftests/testcases/rs485echo.

Interfaces with an MDB transceiver should not be accessed directly via their device nodes. Instead, there is a library for MDB communication in the BSP. Please see the folder sources/meta-guf/meta/recipes-guf/mdbtest for a full source code example.

Real Time Clock (RTC)

All SECO Northern Europe systems are equipped with a hardware real time clock. The system time is automatically set during the boot sequence by reading the RTC.

root@santaro:~# hwclock --show Thu Apr 14 14:51:12 UTC 2022

The RTC time cannot be adjusted directly in one command because only the current system time can be transferred to the RTC. Thus, the system time has to be set first, using the timedatectl command, and can then be written to the RTC:

root@santaro:~# timedatectl set-time "2022-11-02 16:26:00" root@santaro:~# hwclock --systohc

Keypad connector

The so called keypad connector is a general purpose connector. It can be used in different modes. The mode is selected using the kernel command line. See [Kernel command line] on how it can be modified.

The command line configuration of the keypad functionality is only needed and possible on the platforms SANTARO and SANTOKA. On the other platforms the available functionality is fix.

The actual pin mapping is described in the hardware guide for your device.

There are three available functions.

All pins are available as GPIOs. You can access them in usermode using the sysfs entries:

/sys/class/gpio/gpio<pin-number>

Before using these sysfs entries, the "export" should be done.
May it's better to use gpiod interface (gpioget/gpioset tools) to access GPIOs.

keypad=i2c

The i2c-1 interface is mapped to the keypad connector pins 11 and 12. It is available in usermode as:

/dev/i2c-1

This interface supports I2C environments where multiple masters are driving the bus (MultiMaster).

 

See I2C doc.

keypad=spi

The SPI interface is mapped to the pins 12 to 18. The interface is available in usermode as:

/dev/spidev0.0 /dev/spidev0.1 /dev/spidev0.2

See SPI doc.

The options can be combined to use multiple functions at the same time, for example:

keypad=gpio,i2c,spi

will enable all three functions. This leads to the following mapping:

  • pins 3-10 are used as GPIOs

  • 11 and 12 are used for I2C

  • 13-18 are used for SPI

The default setting is:

keypad=gpio,i2c

SPI

There are two ways of controlling SPI bus devices from a Linux system:

  • Writing a kernel SPI device driver and accessing this driver from user space by using its interface.

  • Accessing the SPI bus via the Linux kernel’s spidev API directly.

Describing the process of writing a Linux SPI device driver is out of the scope of this manual. The AT25 SPI eeprom can serve as a good and simple example for such a driver. It is located in the kernel directory under:

  • drivers/misc/eeprom/at25.c

The interface provided to the user space by such a kernel driver depends on the sort of this driver (e.g. character misc driver, input subsystem device driver, etc.). A very common usecase for an SPI driver is a touchscreen driver that uses the input event subsystem.

Accessing the SPI bus from userspace directly via spidev is described in the Linux kernel documentation and available in the kernel directory under:

  • Documentation/spi/spidev.rst

Additionally, there is an example C program available in the same location:

  • tools/spi/spidev_test.c

The header for spidev is available under:

  • include /uapi/linux/spi/spidev.h

If spidev is used to access the SPI bus directly, the user is responsible for keeping the interoperability consistent with all other SPI devices that are controlled by the Linux kernel.

I2C

There are two ways of controlling I2C bus devices from a Linux system:

  • Writing a kernel I2C device driver and accessing this driver from user space by using its interface.

  • Accessing the I2C bus via the Linux kernel’s i2c-dev API directly.

Describing the process of writing a Linux I2C device driver is out of this scope of this manual. The AT24 I2C EEPROM can serve as a good and simple example for such a driver. It is located in the kernel directory under:

  • drivers/misc/eeprom/at24.c

The interface provided to the user space by such a kernel driver depends on the sort of this driver (e.g. character misc driver, input subsystem device driver, etc.). A very common usecase for an I2C driver is a touchscreen driver that uses the input event subsystem.

Accessing the I2C bus from userspace directly via i2c-dev is described in the Linux kernel documentation and available inside the kernel directory under:

  • Documentation/i2c/dev-interface.rst

The header for i2c-dev is available under:

  • include/linux/i2c-dev.h

If i2c-dev is used to access the I2C bus directly, the user is responsible for keeping the interoperability consistent with all other I2C devices that are controlled by the Linux kernel.

CAN

CAN bus devices are controlled through the SocketCAN framework in the Linux kernel. As a consequence, CAN interfaces are network interfaces. Applications receive and transmit CAN messages via the BSD Socket API. CAN interfaces are configured via the netlink protocol. Refer to the hardware manual corresponding to your device to find the physical port location..

root@santaro:~# ip link set can1 type can bitrate 125000 root@santaro:~# ifconfig can1 up

The structure for a CAN message is defined in the kernel header include/uapi/linux/can.h :

struct can_frame { u32 can_id; /* 29 bit CAN_ID + flags */ u8 can_dlc; /* data length code: 0 .. 8 */ u8 data[8]; };

Initializing SocketCAN:

int iSock; struct sockaddr_can addr; iSock = socket(PF_CAN, SOCK_RAW, CAN_RAW); addr.can_family = AF_CAN; addr.can_ifindex = if_nametoindex("can0"); bind(iSock, (struct sockaddr *)&addr, sizeof(addr);

Sending CAN messages:

struct can_frame frame; frame.can_id = 0x123; frame.can_dlc = 1; frame.data[0] = 0xAB; nbytes = write(iSock, &frame, sizeof(frame));

Receiving CAN messages:

struct can_frame frame; nbytes = read(iSock, &frame, sizeof(frame)); if (nbytes > 0) { printf("ID=0x%X DLC=%d data[0]=0x%X\n", frame.can_id, frame.can_dlc, frame.data[0]); }

Sending a CAN message with four bytes with the standard ID 0x20 on can1 from the Linux shell, using the cansend tool. The CAN bus has to be physically prepared properly and there has to be at least one other node that is configured to read on this message ID for this task. Furthermore, all nodes must have the same bit-timing.

root@santaro:~# cansend can1 -i 0x20 0xca 0xbe 0xca 0xbe

Reading all CAN messages on can1 using candump tool:

root@santaro:~# candump can1

A more detailed documentation of the CAN bus handling in the Linux kernel can be found in the documentation directory of the Linux kernel source tree.

 

"can_dlc" is deprecated in 5.15 kernel.
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/Documentation/networking/can.rst?h=linux-5.15.y#n236

USB

There are two general types of USB devices:

  • USB Host: the Linux platform device is the host and controls several devices supported by corresponding Linux drivers.

  • USB Device: the Linux platform device acts as a USB device itself by emulating a specific device type.

USB Host

For USB Host functionality, SECO Northern Europe platforms per default support the following devices:

  • USB Mass Storage

  • USB Keyboard

There are many more device drivers available in the Linux kernel. They are not activated by default, because SECO Northern Europe cannot maintain and test the huge amount of existing drivers. Instead, the customer may do this themselves or engage SECO Northern Europe to implement their special use case. Existing drivers can easily be activated by reconfiguring and rebuilding the Linux kernel inside the BSP.

The USB Host bus can also be directly accessed by using libusb. This library is installed on SECO Northern Europe platforms per default. Further information can be found here.

If libusb is used to access the USB bus directly, the user is responsible to keep the interoperability consistent with all other USB devices that are controlled by the Linux kernel.

USB Device

For USB Device functionality, the following device emulations are supported per default:

  • USB Serial Gadget

Again, further drivers can be activated by reconfiguring the Linux kernel. The USB Device drivers are not compiled into the kernel by default, but are located as modules in the file system. In order to use the Serial Gadget for example, the according module has to be loaded:

root@santaro:~# modprobe g_serial

The Serial Gadget creates a virtual serial port over USB, accessible via the device node /dev/ttyGS0.

SD cards and USB mass storage

SD cards and USB mass storage devices can be accessed directly by using their devices nodes. The SD card can be found under /dev/mmcblk1, its single partitions are located under /dev/mmcblk1pX with X being a positive number. The USB mass storage devices can be found under /dev/sdX with X=a..z, its single partitions under /dev/sdXY with Y being a positive number

Create a FAT32 file system on the first partition of an SD card:

root@santaro:~# mkfs.vfat /dev/mmcblk1p1

Mount the first partition on an SD Card into a newly created directory:

root@santaro:~# mkdir /my_sdcard root@santaro:~# mount /dev/mmcblk1p1 /my_sdcard

Mount the first partition on a USB mass storage device into a newly created directory:

root@santaro:~# mkdir /my_usb_drive root@santaro:~# mount /dev/sda1 /my_usb_drive

For security reasons, the automount of SD cards and USB storage are disabled by default. To enable this feature, delete/rename the file automount_disable_all in /etc/udev/mount.ignorelist.d
and reboot the system afterwards.

The mounting point for the automount of SD cards and USB storage can be found at /run/media/

Temperature Sensor

Most SECO Northern Europe systems are equipped with an on-board hardware temperature sensor. The sensor is a Texas Instruments LM73 / LM75 connected via I2C. To measure the current temperature you can read the corresponding sysfs file in /sys/class/hwmon/hwmon0/.

Read current temperature:

root@santaro:~# cat /sys/class/hwmon/hwmon0/temp1_input 23500

Note that the value will be displayed in millidegree Celsius.

The sensor generates an alert when the measured temperature exceeds the maximum temperature defined in temp1_max. The alert flag will be set to 1 (active low) and an interrupt is generated. The interrupt will trigger an event on the temp1_max_alarm sysfs entry, that can be caught using the poll() function. Additionally there is a temp1_min that works similar.

Read the temperature alert flag:

root@santaro:~# cat /sys/class/hwmon/hwmon0/temp1_max_alarm 1

Touchscreen

The touchscreen device is used by the window manager (a Wayland compositor) using libinput which itself accesses the touchscreen via the Linux input subsystem kernel API, i.e. its device node /dev/input/event0. Applications ( i.e. based on Qt) get the input events from the compositor.

Calibration for touchscreens, which is mandatory for resistive touchscreens but also useful for capacitive ones, is done by libinput internally.

The calibration data is still based on tslib’s /etc/shared/pointercal, which is written during the production tests. This data is converted to the libinput format during the first boot and is found in /etc/udev/rules.d/libinput-touchcalibration.rules. The latter file should not be changed manually.

SECO Northern Europe devices are pre-calibrated and should be shipped out in a ready-to-use fashion. However, If the touchscreen does not behave correctly, the user can perform a recalibration of the touchscreen with:

ts_calibrate

After following the on-screen instruction, the new calibration data is written to /etc/shared/pointercal. Now execute the following command to update the libinput calibration:

touchcal-conv to-udevrule sync

After a reboot the touchscreen should be correctly calibrated and work normally.

Audio

SECO Northern Europe systems with an integrated audio codec make use of ALSA (Advanced Linux Sound Architecture) as a software interface. This project includes a user-space library (alsa-lib) and a set of tools (aplay, arecord, amixer, alsactl) for controlling and accessing the audio hardware from user applications or the console. Please refer to the official ALSA webpage for a full documentation.

Play the audio file my_audio_file.wav from the console using aplay:

root@santaro:~# aplay my_audio_file.wav

Record the audio file my_recorded_audio_file.wav from the console using arecord:

root@santaro:~# arecord -f cd -t wav > my_recorded_audio_file.wav

Set the playback volume to half of the maximum:

root@santaro:~# amixer sset 'PCM' 50%

The alsamixer command can also be used for several settings regarding the audio hardware, it includes a GUI that help with the configuration of the sound device.

On devices with HDMI output, the HDMI sound device might be chosen as default. To play sound through the internal speaker, the internal sound device must be chosen with the command alsamixer (and subsequently F6).

HDMI

The HDMI port can be used as primary or secondary display of the device. Furthermore, an audio-enabled HDMI device can be used as secondary sound card.

There are a few different use case for the HDMI display:

  • HDMI as primary display

  • HDMI as secondary display

  • HDMI as mirror of the internal display

Depends on configurations, different settings for the HDMI output can be configured by an xml file. See the chapter [Initial Device Configuration] for further instructions.

Display power

The display can be powered on or off by software. The value is exported as a virtual file in the sysfs under /sys/class/backlight/backlight/bl_power. It can be accessed using the standard file operations open(), read(), write() and close().

Value of 0 means on, 1 means off

root@santaro:~# echo 1 > /sys/class/backlight/backlight/bl_power root@santaro:~# echo 0 > /sys/class/backlight/backlight/bl_power

Display backlight

The brightness of the display backlight can be adjusted in a range from 0 to a maximum level. The value is exported as a virtual file in the sysfs under /sys/class/backlight/backlight/brightness. It can be accessed using the standard file operations open(), read(), write() and close().

root@santaro:~# cat /sys/class/backlight/backlight/brightness 1 root@santaro:~# cat /sys/class/backlight/backlight/max_brightness > /sys/class/backlight/backlight/brightness