i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide5 / 103
NXP Semiconductors
Chapter 1
Overview
This document describes how to build and install the i.MX Linux® OS BSP, where BSP stands for Board Support Package, on the
i.MX platform. It also covers special i.MX features and how to use them.
The document also provides the steps to run the i.MX platform, including board DIP switch settings, and instructions on configuring
and using the U-Boot bootloader.
The later chapters describe how to use some i.MX special features when running the Linux OS kernel.
Features covered in this guide may be specific to particular boards or SoCs. For the capabilities of a particular board or SoC, see
the
i.MX Linux® Release Notes
1.1 Audience
This document is intended for software, hardware, and system engineers who are planning to use the product, and for anyone
who wants to know more about the product.
1.2 Conventions
(IMXLXRN).
This document uses the following conventions:
• Courier New font: This font is used to identify commands, explicit command parameters, code examples, expressions,
data types, and directives.
1.3 Supported hardware SoCs and boards
These are the systems covered in this guide:
• i.MX 6Quad SABRE-SD board and platform
• i.MX 6DualLite SABRE-SD platform
• i.MX 6Quad SABRE-AI platform
• i.MX 6SoloX SABRE-SD platform
• i.MX 7Dual SABRE-SD platform
• i.MX 6QuadPlus SABRE-AI platform
• i.MX 6QuadPlus SABRE-SD platform
• i.MX 6UltraLite EVK platform
• i.MX 6ULL EVK platform
• i.MX 6ULZ EVK platform
• i.MX 7ULP EVK platform
• i.MX 8QuadMax MEK board
• i.MX 8QuadXPlus MEK platform
• i.MX 8DXL EVK Platform
• i.MX 8M Quad EVK platform
• i.MX 8M Mini EVK Board
• i.MX 8M Nano EVK Board
• i.MX 8M Plus EVK board
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide6 / 103
NXP Semiconductors
• i.MX 8DualX MEK Board
Some abbreviations are used in places in this document.
• SABRE-SD refers to the i.MX 6Quad SABRE-SD, i.MX 6DualLite SABRE-SD, i.MX 6QuadPlus SABRE-SD, and i.MX
7Dual SABRE-SD boards.
• SABRE-AI refers to the i.MX 6Quad SABRE-AI, i.MX 6DualLite SABRE-AI, and i.MX 6QuadPlus SABRE-AI boards.
• SoloX or SX refers to the i.MX 6SoloX SABRE-SD and SABRE-AI boards.
• 6UL refers to the i.MX 6UltraLite board.
• 6ULL refers to the i.MX 6ULL board.
• 6ULZ refers to the i.MX 6ULZ board.
• 7ULP refers to the i.MX 7Ultra Low Power platform.
• 8QXP refers to the 8QuadXPlus platform.
• 8QM refers to the 8QuadMax platform.
• 8MQ refers to the 8M Quad platform.
• 8MM refers to the 8M Mini platform.
• 8MN refers to the 8M Nano platform.
Overview
• 8MP refers to the 8M Plus platform.
• 8DXL refers to the 8DualXLite platform.
• 8DX refers to the 8DualX platform.
1.4 References
i.MX has multiple families supported in software. The following are the listed families and SoCs per family. The i.MX Linux
Release Notes describes which SoC is supported in the current release. Some previously released SoCs might be buildable in
the current release but not validated if they are at the previous validated level.
• i.MX 8M Plus Evaluation Kit Quick Start Guide (IMX8MPLUSQSG)
Documentation is available online at nxp.com.
• i.MX 6 information is at nxp.com/iMX6series
Overview
• i.MX SABRE information is at nxp.com/imxSABRE
• i.MX 6UltraLite information is at nxp.com/iMX6UL
• i.MX 6ULL information is at nxp.com/iMX6ULL
• i.MX 7Dual information is at nxp.com/iMX7D
• i.MX 7ULP information is at nxp.com/imx7ulp
• i.MX 8 information is at nxp.com/imx8
• i.MX 6ULZ information is at nxp.com/imx6ulz
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide8 / 103
NXP Semiconductors
Chapter 2
Introduction
The i.MX Linux BSP is a collection of binary files, source code, and support files that can be used to create a U-Boot bootloader,
a Linux kernel image, and a root file system for i.MX development systems. The Yocto Project is the framework of choice to build
the images described in this document, although other methods can be used.
All the information on how to set up the Linux OS host, how to run and configure a Yocto Project, generate an image, and generate
a rootfs, are covered in the
When Linux OS is running, this guide provides information on how to use some special features that i.MX SoCs provide. The
release notes provide the features that are supported on a particular board.
i.MX Yocto Project User's Guide
(IMXLXYOCTOUG).
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide9 / 103
NXP Semiconductors
Chapter 3
Basic Terminal Setup
The i.MX boards can communicate with a host server (Windows® OS or Linux OS) using a serial cable. Common serial
communication programs such as HyperTerminal, Tera Term, or PuTTY can be used. The example below describes the serial
terminal setup using HyperTerminal on a host running Windows OS.
The i.MX 6Quad/QuadPlus/DualLite SABRE-AI boards connect to the host server using a serial cable.
The other i.MX boards connect the host driver using the micro-B USB connector.
1. Connect the target and the PC running Windows OS using a cable mentioned above.
2. Open HyperTerminal on the PC running Windows OS and select the settings as shown in the following figure.
Figure 1. Teraterm settings for terminal setup
The i.MX 8 board connects the host driver using the micro USB connector. The USB to serial driver can be found under
www.ftdichip.com/Drivers/VCP.htm. The FT4232 USB to serial convertor provides four serial ports. The i.MX 8 board uses the
first port for the Arm® Cortex®-A cores console and the second port for SCU's console. Users need to select the first port (COM)
in the terminal setup. The i.MX 8DXL board uses the third and fourth ports respectively for Arm Cortex-A cores console and
SCU console.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide10 / 103
NXP Semiconductors
Chapter 4
Booting Linux OS
Before booting the Linux OS kernel on an i.MX board, copy the images (U-Boot, Linux kernel, device tree, and rootfs) to a boot
device and set the boot switches to boot that device. There are various ways to boot the Linux OS for different boards, boot
devices, and results desired. This section describes how to prepare a boot device, where files need to be in the memory map, how
to set switches for booting, and how to boot Linux OS from U-Boot.
4.1 Software overview
This section describes the software needed for the board to be able to boot and run Linux OS.
To boot a Linux image on i.MX 6 and i.MX 7, the following elements are needed:
• Bootloader (U-Boot)
• Linux kernel image (zImage)
• A device tree file (.dtb) for the board being used
• A root file system (rootfs) for the particular Linux image
• Arm Cortex-M4 image for i.MX 7ULP
To boot a Linux image on i.MX 8QuadMax, i.MX 8QuadXPlus, i.MX 8DXL, and i.MX 8DualX, four elements are needed:
• Bootloader (imx-boot built by imx-mkimage, which is a tool that combines firmware and U-Boot to create a bootloader for i.MX
8), which includes U-Boot, Arm Trusted Firmware, DCD file, System controller firmware, and the SECO firmware since B0.
• Arm Cortex-M4 image
• Linux kernel image (Image built by linux-imx)
• A device tree file (.dtb) for the board being used
• A root file system (rootfs) for the particular Linux image
On i.MX 8M Quad, i.MX 8M Mini, i.MX 8M Nano, and i.MX 8M Plus, four elements are needed:
• imx-boot (built by imx-mkimage), which includes SPL, U-Boot, Arm Trusted Firmware, DDR firmware
• HDMI firmware (only supported by i.MX 8M Quad)
• Linux kernel image
• A device tree file (.dtb) for the board being used.
• A root file system (rootfs) for the particular Linux image
The system can be configured for a specific graphical backend. For i.MX 8, the graphical backend is XWayland. For i.MX 7ULP,
the default backend is XWayland.
4.1.1 Bootloader
U-Boot is the tool recommended as the bootloader for i.MX 6 and i.MX 7. i.MX 8 requires a bootloader that includes U-boot as
well as other components described below. U-Boot must be loaded onto a device to be able to boot from it. U-Boot images are
board-specific and can be configured to support booting from different sources.
The pre-built or Yocto project default bootloader names start with the name of the bootloader followed by the name of the
platform and board and followed by the name of the device that this image is configured to boot from: u-boot-[platform]
[board]_[machine_configuration].bin. If no boot device is specified, it boots from SD/MMC.
The manufacturing tool can be used to load U-Boot onto all devices with i.MX 6 and i.MX 7. U-Boot can be loaded directly onto
an SD card using the Linux dd command. U-Boot can be used to load a U-Boot image onto some other devices.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide11 / 103
NXP Semiconductors
Booting Linux OS
On i.MX 8, the U-Boot cannot boot the device by itself. The i.MX 8 pre-built images or Yocto Project default bootloader is imx-boot
for the SD card, which is created by the imx-mkimage. The imx-boot binary includes the U-boot, Arm trusted firmware, DCD
file (8QuadMax/8QuadXPlus/8DXL), system controller firmware (8QuadMax/8QuadXPlus/8DXL), SPL (8M SoC), DDR firmware
(8M), HDMI firmware (8M Quad), and SECO firmware for B0 (8QuadMax/8QuadXPlus/8DXL).
On i.MX 8M SoC, the second program loader (SPL) is enabled in U-Boot. SPL is implemented as the first-level bootloader running
on TCML (For i.MX 8M Nano and i.MX 8M Plus, the first-level bootloader runs in OCRAM). It is used to initialize DDR and load
U-Boot, U-Boot DTB, Arm trusted firmware, and TEE OS (optional) from the boot device into the memory. After SPL completes
loading the images, it jumps to the Arm trusted firmware BL31 directly. The BL31 starts the optional BL32 (TEE OS) and BL33
(U-Boot) for continue booting kernel.
In imx-boot, the SPL is packed with DDR Firmware together, so that ROM can load them into Arm Cortex-M4 TCML or OCRAM
(only for i.MX 8M Nano and i.MX 8M Plus). The U-Boot, U-Boot DTB, Arm Trusted firmware, and TEE OS (optional) are packed
into a FIT image, which is finally built into imx-boot.
4.1.2 Linux kernel image and device tree
This i.MX BSP contains a pre-built kernel image based on the 5.10.9 version of the Linux kernel and the device tree files associated
with each platform.
The same kernel image is used for all the i.MX 6 and i.MX 7 with name zImage. Device trees are tree data structures, which
describe the hardware configuration allowing a common kernel to be booted with different pin settings for different boards or
configurations. Device tree files use the .dtb extension. The configuration for a device tree can be found in the Linux source code
under arch/arm/boot/dts in the *.dts files.
The i.MX Linux delivery package contains pre-built device tree files for the i.MX boards in various configurations. File names for
the prebuilt images are named Image-[platform]-[board]-[configuration].dtb. For example, the device tree file of the i.MX
8QuadMax MEK board is Image-imx8qm-mek.dtb.
For i.MX 6 and i.MX 7, the *ldo.dtb device trees are used for LDO-enabled feature support. By default, the LDO bypass is
enabled. If your board has the CPU set to 1.2 GHz, you should use the *ldo.dtb device tree instead of the default, because LDO
bypass mode is not supported on the CPU at 1.2 GHz. The device tree *hdcp.dtb is used to enable the HDCP feature because
of a pin conflict, which requires this to be configured at build time.
On i.MX 8 and i.MX 8M, the kernel is 64 bit and device trees are located in the arch/arm64/boot/dts/freescale folder and
use the dts extension. The kernel is built using linux-imx software provided in the release package and the file name starting
with Image.
4.1.3 Root file system
The root file system package (or rootfs) provides busybox, common libraries, and other fundamental elements.
The i.MX BSP package contains several root file systems. They are named with the following convention: [image name]-
[backend]-[platform][board].[ext4|sdcard]. The ext4 extension indicates a standard file system. It can be mounted as
NFS, or its contents can be stored on a boot media such as an SD/MMC card.
The graphical backend to be used is also defined by the rootfs.
4.2 Universal update utility
The Universal Update Utility (UUU) runs on a Windows or Linux OS host and is used to download images to different devices on
an i.MX board.
4.2.1 Downloading UUU
Download UUU version 1.4.72 or later from https://github.com/NXPmicro/mfgtools/releases.
4.2.2 Using UUU
Follow these instructions to use the UUU for i.MX 6, i.MX 7, i.MX 8:
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide12 / 103
NXP Semiconductors
Booting Linux OS
1. Connect a USB cable from a computer to the USB OTG/TYPE C port on the board.
2. Connect a USB cable from the OTG-to-UART port to the computer for console output.
3. Open a Terminal emulator program. See Section "Basic Terminal Setup" in this document.
4. Set the boot pin to serial download mode mode. See Section "Serial download mode for the Manufacturing Tool" in this
document.
For detailed usage of UUU, see github.com/NXPmicro/mfgtools/wiki.
For example, the following command writes rootfs.wic into eMMC.
uuu -b emmc_all <bootloader> <rootfs.wic>
The following command decompresses bz2 file and writes into eMMC:
uuu -b emmc_all <bootloader> <rootfs.wic.bz2/*>
The following command executes downloading and bootloader (SPL and U-Boot) by USB:
uuu -b spl <bootloader>
The follow command burns into eMMC (If only one board is supported in such a release package and the board supports
eMMC chip):
uuu <release package>.zip
NOTE
For i.MX 8QuadXPlus B0, UUU flashes the eMMC image to boot partition with 32 KB offset. It may not be
compatible with all eMMC devices. It is recommended to enable eMMC fastboot mode and use the UUU kernel
version script to flash the eMMC image to boot partition with 0 offset.
4.3 Preparing an SD/MMC card to boot
This section describes the steps to prepare an SD/MMC card to boot up an i.MX board using a Linux host machine. These
instructions apply to SD and MMC cards although for brevity, and usually only the SD card is listed.
For a Linux image to be able to run, four separate pieces are needed:
• Linux OS kernel image (zImage/Image)
• Device tree file (*.dtb)
• Bootloader image
• Root file system (i.e., EXT4)
The Yocto Project build creates an SD card image that can be flashed directly. This is the simplest way to load everything needed
onto the card with one command.
A .wic image contains all four images properly configured for an SD card. The release contains a pre-built .wic image that is built
specifically for the one board configuration. It runs the Wayland graphical backend. It does not run on other boards unless U-Boot,
the device tree, and rootfs are changed.
When more flexibility is desired, the individual components can be loaded separately, and those instructions are included here as
well. An SD card can be loaded with the individual components one-by-one or the .wic image can be loaded and the individual
parts can be overwritten with the specific components.
The rootfs on the default .wic image is limited to a bit less than 4 GB, but re-partitioning and re-loading the rootfs can increase
that to the size of the card. The rootfs can also be changed to specify the graphical backend that is used.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide13 / 103
NXP Semiconductors
Booting Linux OS
The device tree file (.dtb) contains board and configuration-specific changes to the kernel. Change the device tree file to change
the kernel for a different i.MX board or configuration.
By default, the release uses the following layout for the images on the SD card. The kernel image and DTB move to use the FAT
partition without a fixed raw address on the SD card. The users have to change the U-Boot boot environment if the fixed raw
address is required.
An SD/MMC card reader, such as a USB card reader, is required. It is used to transfer the bootloader and kernel images to
initialize the partition table and copy the root file system. To simplify the instructions, it is assumed that a 4 GB SD/MMC card
is used.
Any Linux distribution can be used for the following procedure.
The Linux kernel running on the Linux host assigns a device node to the SD/MMC card reader. The kernel might decide the device
node name or udev rules might be used. In the following instructions, it is assumed that udev is not used.
To identify the device node assigned to the SD/MMC card, carry out the following command:
$ cat /proc/partitions
major minor #blocks name
8 0 78125000 sda
8 1 75095811 sda1
8 2 1 sda2
8 5 3028221 sda5
8 32 488386584 sdc
8 33 488386552 sdc1
8 16 3921920 sdb
8 18 3905535 sdb1
In this example, the device node assigned is /dev/sdb (a block is 1024 Bytes).
NOTE
Make sure that the device node is correct for the SD/MMC card. Otherwise, it may damage your operating system
or data on your computer.
4.3.2 Copying the full SD card image
The SD card image (with the extension .wic) contains U-Boot, the Linux image and device trees, and the rootfs for a 4 GB SD
card. The image can be installed on the SD card with one command if flexibility is not required.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide14 / 103
NXP Semiconductors
Booting Linux OS
Carry out the following command to copy the SD card image to the SD/MMC card. Change sdx below to match the one used by
the SD card.
The entire contents of the SD card are replaced. If the SD card is larger than 4 GB, the additional space is not accessible.
4.3.3 Partitioning the SD/MMC card
The full SD card image already contains partitions. This section describes how to set up the partitions manually. This needs to be
done to individually load the bootloader, kernel, and rootfs.
There are various ways to partition an SD card. Essentially, the bootloader image needs to be at the beginning of the card, followed
by the Linux image and the device tree file. These can either be in separate partitions or not. The root file system needs to be
in a partition that starts after the Linux section. Make sure that each section has enough space. The example below creates
two partitions.
On most Linux host operating systems, the SD card is mounted automatically upon insertion. Therefore, before running fdisk,
make sure that the SD card is unmounted if it was previously mounted (through sudo umount /dev/sdx).
Start by running fdisk with root permissions. Use the instructions above to determine the card ID. We are using sdx here as
an example.
$ sudo fdisk /dev/sdx
Type the following parameters (each followed by <ENTER>):
p [lists the current partitions]
d [to delete existing partitions. Repeat this until no unnecessary partitions
are reported by the 'p' command to start fresh.]
n [create a new partition]
p [create a primary partition - use for both partitions]
1 [the first partition]
20480 [starting at offset sector]
1024000 [ending position of the first partition to be used for the boot images]
p [to check the partitions]
n
p
2
1228800 [starting at offset sector, which leaves enough space for the kernel,
the bootloader and its configuration data]
<enter> [using the default value will create a partition that extends to
the last sector of the media]
p [to check the partitions]
w [this writes the partition table to the media and fdisk exits]
4.3.4 Copying a bootloader image
This section describes how to load only the bootloader image when the full SD card image is not used. Execute the following
command to copy the U-Boot image to the SD/MMC card.
• 33 - for i.MX 8QuadMax A0, i.MX 8QuadXPlus A0, and i.MX 8M Quad
• 32 - for i.MX 8QuadXPlus B0, i.MX 8QuadMax B0, i.MX 8DualX, i.MX 8DXL, i.MX 8M Nano, i.MX 8M Mini, and i.MX 8M Plus
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide15 / 103
NXP Semiconductors
Booting Linux OS
The first 16 KB of the SD/MMC card, which includes the partition table, is reserved.
4.3.5 Copying the kernel image and DTB file
This section describes how to load the kernel image and DTB when the full SD card image is not used. The pre-built SD card image
uses the VFAT partition for storing kernel image and DTB, which requires a VFAT partition that is mounted as a Linux drive and
the files are simply copied into it. This is the preferred method.
Another method that can be used is for users to put the kernel image and DTB to the fixed raw address of the SD card by using the
dd command. The later method needs to modify the U-Boot default environment variables for loading the kernel image and DTB.
Default: VFAT partition
1. Format partition 1 on the card as VFAT with this command:
$ sudo mkfs.vfat /dev/sdx1
2. Mount the formatted partition with this command:
$ mkdir mountpoint
$ sudo mount /dev/sdx1 mountpoint
3. Copy the zImage and *.dtb files to the mountpoint by using cp. The device tree names should match the one used by the
variable specified by U-Boot. Unmount the partition with this command:
$ sudo umount mountpoint
Alternative: Pre-defined raw address
The following command can be used to copy the kernel image to the SD/MMC card:
Each of them copies the kernel to the media at offset 1 MB (bs x seek = 512 x 2048 = 1 MB). The file zImage_imx_v7_defconfig
refers to the zImage file created when using the imx_v7_defconfig configuration file, which supports both i.MX 6 and i.MX 7 SoCs.
The i.MX DTB image can be copied by using the copy command and copying the file to the 2nd partition or the following commands
copy an i.MX DTB image to the SD/MMC card by using dd command.
For i.MX 6 and i.MX 7, the following command can be used to copy the kernel image to the boards, such as the i.MX 6UltraLite
EVK board and i.MX 6ULL EVK board:
For i.MX 6 and i.MX 7, this copies the board-specific .dtb file to the media at offset 10 MB (bs x seek = 512 x 20480 = 10 MB).
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide16 / 103
NXP Semiconductors
Booting Linux OS
4.3.6 Copying the root file system (rootfs)
This section describes how to load the rootfs image when the full SD card image is not used.
Copy the target file system to a partition that only contains the rootfs. This example uses partition 2 for the rootfs. First format the
partition. The file system format ext3 or ext4 is a good option for the removable media due to the built-in journaling. Replace sdx
with the partition in use in your configuration.
$ sudo mkfs.ext3 /dev/sdx2
Or
$ sudo mkfs.ext4 /dev/sdx2
Copy the target file system to the partition:
$ mkdir /home/user/mountpoint
$ sudo mount /dev/sdx2 /home/user/mountpoint
Extract a rootfs package to a directory: for example, extract imx-image-multimedia-imx7ulpevk.tar.bz2 to /home/
user/rootfs:
$ cd /home/user/rootfs
$ tar -jxvf imx-image-multimedia-imx7ulpevk.tar.bz2
The rootfs directory needs to be created manually.
Assume that the root file system files are located in /home/user/rootfs as in the previous step:
$ cd /home/user/rootfs
$ sudo cp -a * /home/user/mountpoint
$ sudo umount /home/user/mountpoint
$ sync
The file system content is now on the media.
NOTE
Copying the file system takes several minutes depending on the size of your rootfs.
4.4 Downloading images
Images can be downloaded to a device using a U-Boot image that is already loaded on the boot device or by using the
Manufacturing Tool UUU. Use a terminal program to communicate with the i.MX boards.
4.4.1 Downloading images using U-Boot
The following sections describe how to download images using the U-Boot bootloader.
The commands described below are generally useful when using U-Boot. Additional commands and information can be found by
typing help at the U-Boot prompt.
The U-Boot print command can be used to check environment variable values.
The setenv command can be used to set environment variable values.
4.4.1.1 Flashing an Arm Cortex-M4 image on QuadSPI
i.MX 6SoloX SABRE-SD/SABRE-AI, i.MX 7ULP EVK, and i.MX 7Dual SABRE-SD boards have the Arm Cortex-M4 processor and
QuadSPI memory that can be used to flash an image to it.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide17 / 103
NXP Semiconductors
Booting Linux OS
NOTE
To enable the full features for i.MX 7ULP, burn the Arm Cortex-M4 image to QuadSPI. It is recommended to use the
MfgTool script uuu LF5.10.9_1.0.0_images_MX7ULPEVK.zip\uuu_sd_m4.auto to burn both BSP and
Arm Cortex-M4 images.
i.MX U-Boot provides a reference script on i.MX 7Dual SABRESD and i.MX 6SoloX SABRE-AI/SABRE-SD to flash the Arm
Cortex-M4 image from the SD card. To execute the script, perform the following steps:
1. Copy the Arm Cortex-M4 image to the first VFAT partition of the boot SD card. Name the file to “m4_qspi.bin”.
2. Boot from the SD card.
3. Flash the Arm Cortex-M4 image from the SD card to the NOR flash on QuadSPI2 PortB CS0 on the i.MX 6SoloX
SABRE-SD board, QuadSPI1 PortB CS0 on the i.MX 6SoloX SABRE-AI board, or QuadSPI1 PortA CS0 offset 1 MB on
the i.MX 7Dual SABRE-SD board.
U-Boot > run update_m4_from_sd
Alternatively, users can flash the Arm Cortex-M4 image from TFTP by performing the following steps:
1. Boot from the SD card.
2. TFTP the Arm Cortex-M4 image.
U-Boot > tftp ${loadaddr} m4_qspi.bin
3. Select the NOR flash on QuadSPI2 PortB CS0 on the i.MX 6SoloX SABRE-SD board or QuadSPI1 PortB CS0 on the
i.MX 6SoloX SABRE-AI board.
U-Boot > sf probe 1:0
Select the NOR flash on QuadSPI1 PortA CS0 on the i.MX 7Dual SABRE-SD board and i.MX 7ULP EVK board.
U-Boot > sf probe 0:0
4. Flash the Arm Cortex-M4 image to the selected NOR flash. The erase size is ${filesize}, around 64 Kbytes. This
example assumes that it is 128 Kbytes.
U-Boot > sf erase 0x0 0x20000
U-Boot > sf write ${loadaddr} 0x0 ${filesize}
i.MX 7Dual SABRE-SD needs to program the Arm Cortex-M4 images to 1 MB offset, because the first 1 MB is used by the
U-Boot image in QuadSPI.
U-Boot > sf erase 0x100000 0x20000
U-Boot > sf write ${loadaddr} 0x100000 ${filesize}
NOTE
On i.MX 7Dual SABRE-SD, the Arm Cortex-M4 image on QuadSPI is supported only when the U-Boot image is
built by the target mx7dsabresd_qspi1_defconfig booted by U-Boot from QuadSPI.
The default U-Boot for the i.MX 7Dual SABRESD board uses the Cortex-M4 image from the SD card and runs it
on OCRAM.
On i.MX 7ULP EVK, the Arm Cortex-M4 image needs to be programmed. Otherwise, it will not boot.
4.4.1.2 Downloading an image to MMC/SD
This section describes how to download U-Boot to an MMC/SD card that is not the one used to boot from.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide18 / 103
NXP Semiconductors
Booting Linux OS
Insert an MMC/SD card into the SD card slot. This is slot SD3 on i.MX 6 SABRE, SD2 on i.MX 6UltraLite EVK and i.MX 6ULL EVK,
SD1 on i.MX 7Dual SABRE-SD and i.MX 7ULP EVK (MicroSD), and SD1 on i.MX 8QuadMax MEK , 8QuadXPlus MEK, and i.MX
8M Quad EVK.
NOTE
To enable the full features for i.MX 7ULP, burn the Arm Cortex-M4 image to QuadSPI. It is recommended to use the
MfgTool script uuu LF5.10.9_1.0.0_images_MX7ULPEVK.zip\uuu_sd_m4.auto to burn both BSP and
Arm Cortex-M4 images.
For i.MX 7ULP, to burn the Arm Cortext-M4 image to QuadSPI, perform the following steps:
1. Copy the Arm Cortext-M4 image to the SD card vfat partition, insert the SD card, and then boot to the U-Boot console.
2. Probe the Quad SPI in U-Boot, and erase an enough big size QuardSPI flash space for this Arm Cortext-M4 image.
U-Boot > sf probe
U-Boot > sf erase 0x0 0x30000;
3. Read the Arm Cortext-M4 image (in the first vfat partition on the SD card) to memory address, the Arm Cortext-M4 image
name is sdk20-app.img here.
4. Write the Arm Cortext-M4 image to the QuardSPI.
U-Boot > sf write 0x62000000 0x0 0x30000
To flash the original U-Boot, see Section Preparing an SD/MMC card to boot.
The U-Boot bootloader is able to download images from a TFTP server into RAM and to write from RAM to an SD card. For this
operation, the Ethernet interface is used and U-Boot environment variables are initialized for network communications.
The boot media contains U-Boot, which is executed upon power-on. Press any key before the value of the U-Boot environment
variable, "bootdelay", decreases and before it times out. The default setting is 3 seconds to display the U-Boot prompt.
1. To clean up the environment variables stored on MMC/SD to their defaults, execute the following command in the
U-Boot console:
2. Configure the U-Boot environment for network communications. The folllowing is an example. The lines preceded by the
"#" character are comments and have no effect.
U-Boot > setenv serverip <your TFTPserver ip>
U-Boot > setenv bootfile <your kernel zImage/Image name on the TFTP server>
U-Boot > setenv fdt_file <your dtb image name on the TFTP server>
The user can set a fake MAC address through ethaddr enviroment if the MAC address is not fused.
U-Boot > setenv ethaddr 00:01:02:03:04:05
U-Boot > save
3. Copy zImage/Image to the TFTP server. Then download it to RAM:
U-Boot > dhcp
4. Query the information about the MMC/SD card.
U-Boot > mmc devU-Boot > mmcinfo
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide19 / 103
NXP Semiconductors
Booting Linux OS
5. Check the usage of the "mmc" command. The "blk#" is equal to "<the offset of read/write>/<block length of the card>".
The "cnt" is equal to "<the size of read/write>/<block length of the card>".
U-Boot > help mmc
mmc - MMC sub system
Usage:
mmc read addr blk# cnt
mmc write addr blk# cnt
mmc erase blk# cnt
mmc rescan
mmc part - lists available partition on current mmc device
mmc dev [dev] [part] - show or set current mmc device [partition]
mmc list - lists available devices
6. Program the kernel zImage/Image located in RAM at ${loadaddr} into the SD card. For example, the command to
write the image with the size 0x800000 from ${loadaddr} to the offset of 0x100000 of the microSD card. See the
following examples for the definition of the MMC parameters.
This example assumes that the kernel image is equal to 0x800000. If the kernel image exceeds 0x800000, increase
the image length. After issuing the TFTP command, filesize of the U-Boot environment variable is set with the number
of bytes transferred. This can be checked to determine the correct size needed for the calculation. Use the U-Boot
command printenv to see the value.
U-Boot > mmc dev 2 0
U-Boot > tftpboot ${loadaddr} ${bootfile}
### Suppose the kernel zImage is less than 8M.
U-Boot > mmc write ${loadaddr} 0x800 0x4000
7. Program the dtb file located in RAM at ${fdt_addr} into the microSD.
U-Boot > tftpboot ${fdt_addr} ${fdt_file}
U-Boot > mmc write ${fdt_addr} 0x5000 0x800
8. On i.MX 6 SABRE boards, you can boot the system from rootfs on SD card, using the HannStar LVDS as display. The
kernel MMC module now uses a fixed mmcblk index for the uSDHC slot. The SD3 slot uses mmcblk2 on i.MX 6 SABRE
boards, the SD1 slot uses mmcblk0 on the i.MX 7Dual SABRE-SD board, and the SD2 slot uses mmcblk1 on the i.MX
6UltraLite board and i.MX 6ULL EVK board. The SD1 slot uses mmcblk1 on i.MX 8 MEK boards and i.MX 8M boards.
There is an eMMC chip on i.MX SABRE boards, i.MX 8 MEK and EVK boards, and i.MX 8M EVK boards. It is accessed through
SDHC4 on i.MX 6 SABRE boards or SDHC3 on i.MX 7Dual SABRE-SD board, SDHC1 on i.MX 8 MEK and EVK boards, and i.MX
8M EVK boards. The i.MX 7ULP EVK board also supports to rework eMMC on the MicroSD port. The following steps describe
how to use this memory device.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide20 / 103
NXP Semiconductors
NOTE
To enable the full features for i.MX 7ULP, burn the Arm Cortex-M4 image to QuadSPI. It is recommended to use the
MfgTool script uuu LF5.10.9_1.0.0_images_MX7ULPEVK.zip\uuu_sd_m4.auto to burn both BSP and
Arm Cortex-M4 images.
1. Execute the following command on the U-Boot console to clean up the environments stored on eMMC:
U-Boot > env default -f -a
U-Boot > save
U-Boot > reset
2. Configure the boot pin. Power on the board and set the U-Boot environment variables as required. For example,
U-Boot > setenv serverip <your tftpserver ip>
U-Boot > setenv bootfile <your kernel zImage/Image name on the tftp server>
U-Boot > setenv fdt_file <your dtb image name on the tftp server>
### The user can set fake MAC address via ethaddr enviroment if the MAC address is not fused
U-Boot > setenv ethaddr 00:01:02:03:04:05
U-Boot > save
3. Copy zImage to the TFTP server. Then download it to RAM:
U-Boot > dhcp
Booting Linux OS
4. Query the information about the eMMC chip.
U-Boot > mmc dev
U-Boot > mmcinfo
5. Check the usage of the "mmc" command. "blk#" is equal to "<the offset of read/write>/<block length of the card>". "cnt"
is equal to "<the size of read/write>/<block length of the card>".
mmc read addr blk# cnt
mmc write addr blk# cnt
mmc erase blk# cnt
mmc rescan
mmc part - lists available partition on current mmc device
mmc dev [dev] [part] - show or set current mmc device [partition]
mmc list - lists available devices
6. Program the kernel zImage/Image into eMMC. For example, the command below writes the image with the size
0x800000 from ${loadaddr} to the offset 0x100000 of the eMMC chip. Here, the following equations are used: 0x800
=0x100000/0x200, 0x4000=0x800000/0x200. The block size of this card is 0x200. This example assumes that the
kernel image is less than 0x800000 bytes. If the kernel image exceeds 0x800000, enlarge the image length.
### Select mmc dev 2 (USDHC4) on the i.MX 6 SABRESD board:
U-Boot > mmc dev 2 0
### Select mmc dev 1 (USDHC3) on the i.MX 7Dual SABRESD board:
U-Boot > mmc dev 1 0
### Select mmc dev 1 (USDHC2) on the i.MX 6UltraLite EVK board:
U-Boot > mmc dev 1 0
### Select mmc dev 0 (USDHC1) on the i.MX 7ULP EVK board:
U-Boot > mmc dev 0 0
### Select mmc dev 0 (eMMC0) on the i.MX 8QuadMax MEK, i.MX 8QuadXPlus MEK, i.MX 8M Quad,
8DualX, and 8DXL boards:
U-Boot > mmc dev 0 0
### select mmc dev 2 (USDHC3) on the i.MX 8M Mini EVK, i.MX 8M Nano EVK, and i.MX 8M Plus EVK:
U-Boot > mmc dev 2 0
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide21 / 103
NXP Semiconductors
Booting Linux OS
### Suppose kernel zImage is less than 8 MB:
U-Boot > tftpboot ${loadaddr} ${bootfile}
U-Boot > mmc write ${loadaddr} 0x800 0x4000
7. Program the dtb file located in RAM at ${fdt_addr} into the eMMC chip.
U-Boot > tftpboot ${fdt_addr} ${fdt_file}
U-Boot > mmc write ${fdt_addr} 0x5000 0x800
8. Boot up the system through the rootfs in eMMC, using the HannStar LVDS as display. The kernel MMC module now
uses the fixed mmcblk indices for the USDHC slots. The eMMC/SD4 slot on the i.MX 6 SABRE boards is mmcblk3.
The eMMC5.0 on the i.MX 8QuadMax MEK board, i.MX 8QuadXPlus MEK board, and i.MX 8M Quad EVK board are
mmcblk0. The eMMC5.0/SD3 slot on the i.MX 7Dual SABRE board is mmcblk2. eMMC is not populated on the i.MX
7Dual SABRE board.
U-Boot > setenv mmcboot 'run bootargs_base mmcargs; mmc dev 2;
• For i.MX 8QuadMax/8QuadXPlus/8M Quad/8M Plus, the following display kernel parameters are supported:
a. Pick a particular video mode for legacy FB emulation since system startup.
video=HDMI-A-{n}: {video_mode}
n can be 1 to the maximum number of HDMI connectors in the system. video_mode should be the one that
the monitor on the connector supports. For example, video=HDMI-A-1:1920x1080@60. By default, if there is no
parameter in the command line, the system uses the video mode that the monitor recommends.
b. Enable or disable legacy FB emulation.
drm_kms_helper.fbdev_emulation=0 or 1
0 to disable, 1 to enable. By default, if there is no parameter in the command line, the emulation is enabled.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide22 / 103
NXP Semiconductors
Booting Linux OS
c. Set legacy FB emulation framebuffer’s bits per pixel (bpp) parameter.
imxdrm.legacyfb_depth=16 or 24 or 32
By default, if there is no parameter in the command line, bpp is 16.
To program the rootfs to MMC/SD, see Using an i.MX board as the host server to create a rootfs or Preparing an SD/MMC card
to boot.
4.4.1.4 Flashing U-Boot on SPI-NOR from U-Boot
Flashing directly to SPI-NOR with TFTPBoot is limited to i.MX 6 SABRE-AI boards. To flash U-Boot on SPI-NOR, perform the
following steps:
1. Boot from an SD card.
2. Set Jumper J3 to position: 2-3.
3. Fetch the U-Boot image with built-in SPI-NOR support. This example uses u-boot.imx.
U-Boot > tftpboot ${loadaddr} u-boot.imx
4. Flash the U-Boot image in SPI-NOR.
U-Boot > sf probe
U-Boot > sf erase 0 0x80000
U-Boot > sf write ${loadaddr} 0x400 0x7FC00
5. Set boot switches to boot from SPI-NOR on SABRE-AI.
• S2-1 1
• S2-2 1
• S2-3 0
• S2-4 0
• S1-[1:10] X
6. Reboot the target board.
4.4.1.4.1 Flashing an Arm Cortex-M4 image on QuadSPI
i.MX 6SoloX SABRE-SD/SABRE-AI, i.MX 7ULP EVK, and i.MX 7Dual SABRE-SD boards have the Arm Cortex-M4 processor and
QuadSPI memory that can be used to flash an image to it.
NOTE
To enable the full features for i.MX 7ULP, burn the Arm Cortex-M4 image to QuadSPI. It is recommended to use the
MfgTool script uuu LF5.10.9_1.0.0_images_MX7ULPEVK.zip\uuu_sd_m4.auto to burn both BSP and
Arm Cortex-M4 images.
i.MX U-Boot provides a reference script on i.MX 7Dual SABRESD and i.MX 6SoloX SABRE-AI/SABRE-SD to flash the Arm
Cortex-M4 image from the SD card. To execute the script, perform the following steps:
1. Copy the Arm Cortex-M4 image to the first VFAT partition of the boot SD card. Name the file to “m4_qspi.bin”.
2. Boot from the SD card.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide23 / 103
NXP Semiconductors
Booting Linux OS
3. Flash the Arm Cortex-M4 image from the SD card to the NOR flash on QuadSPI2 PortB CS0 on the i.MX 6SoloX
SABRE-SD board, QuadSPI1 PortB CS0 on the i.MX 6SoloX SABRE-AI board, or QuadSPI1 PortA CS0 offset 1 MB on
the i.MX 7Dual SABRE-SD board.
U-Boot > run update_m4_from_sd
Alternatively, users can flash the Arm Cortex-M4 image from TFTP by performing the following steps:
1. Boot from the SD card.
2. TFTP the Arm Cortex-M4 image.
U-Boot > tftp ${loadaddr} m4_qspi.bin
3. Select the NOR flash on QuadSPI2 PortB CS0 on the i.MX 6SoloX SABRE-SD board or QuadSPI1 PortB CS0 on the
i.MX 6SoloX SABRE-AI board.
U-Boot > sf probe 1:0
Select the NOR flash on QuadSPI1 PortA CS0 on the i.MX 7Dual SABRE-SD board and i.MX 7ULP EVK board.
U-Boot > sf probe 0:0
4. Flash the Arm Cortex-M4 image to the selected NOR flash. The erase size is ${filesize}, around 64 Kbytes. This
example assumes that it is 128 Kbytes.
U-Boot > sf erase 0x0 0x20000
U-Boot > sf write ${loadaddr} 0x0 ${filesize}
i.MX 7Dual SABRE-SD needs to program the Arm Cortex-M4 images to 1 MB offset, because the first 1 MB is used by the
U-Boot image in QuadSPI.
U-Boot > sf erase 0x100000 0x20000
U-Boot > sf write ${loadaddr} 0x100000 ${filesize}
NOTE
On i.MX 7Dual SABRE-SD, the Arm Cortex-M4 image on QuadSPI is supported only when the U-Boot image is
built by the target mx7dsabresd_qspi1_defconfig booted by U-Boot from QuadSPI.
The default U-Boot for the i.MX 7Dual SABRESD board uses the Cortex-M4 image from the SD card and runs it
on OCRAM.
On i.MX 7ULP EVK, the Arm Cortex-M4 image needs to be programmed. Otherwise, it will not boot.
4.4.1.5 Flashing U-Boot on Parallel NOR from U-Boot
Flashing directly to Parallel NOR with TFTPBoot is limited to i.MX 6 SABRE-AI boards. To flash U-Boot on Parallel NOR, perform
the following steps:
1. Check the jumper J3, should not between pins 2 and 3.
2. Update the SD U-Boot with EIM NOR version. For details on commands, see Copying a bootloader image. Then boot
from the SD card.
3. TFTP the U-Boot image.
tftpboot ${loadaddr} u-boot.imx
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide24 / 103
NXP Semiconductors
Booting Linux OS
4. Flash the U-Boot image.
cp.b ${loadaddr} 0x08001000 ${filesize}
5. Change boot switches and reboot.
S2 all 0 S1-6 1 others 0
6. By default, rootfs is mounted on NFS.
4.4.2 Using an i.MX board as the host server to create a rootfs
Linux OS provides multiple methods to program images to the storage device. This section describes how to use the i.MX platform
as a Linux host server to create the rootfs on an MMC/SD card or the SATA device. The following example is for an SD card. The
device file node name needs to be changed for a SATA device.
1. Boot from NFS or other storage. Determine your SD card device ID. It could be mmcblk* or sd*. (The index is
determined by the USDHC controller index.) Check the partition information with the command:
$ cat /proc/partitions
2. To create a partition on the MMC/SD card, use the fdisk command (requires root privileges) in the Linux console:
root@ ~$ sudo fdisk /dev/$SD
Replace $SD above with the name of your device.
3. If this is a new SD card, you may get the following message:
The device contains neither a valid DOS partition table, nor Sun, SGI or OSF disk label
Building a new DOS disklabel. Changes will remain in memory only,
until you decide to write them. After that the previous content
won't be recoverable.
The number of cylinders for this disk is set to 124368.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
(e.g., DOS FDISK, OS/2 FDISK)
The usual prompt and commands to partition the card are as follows. Text in boldface indicates what the user types.
Command (m for help): p
Disk /dev/sdd: 3965 MB, 3965190144 bytes
4 heads, 32 sectors/track, 60504 cylinders, total 7744512 sectors
4. As described in Flash memory maps, the rootfs partition should be located after the kernel image. The first 0x800000
bytes can be reserved for MBR, bootloader, and kernel sections. From the log shown above, the Units of the current
MMC/SD card is 32768 bytes. The beginning cylinder of the first partition can be set to "0x300000/32768 = 96." The last
cylinder can be set according to the rootfs size. Create a new partition by typing the letters in bold:
Command (m for help): n
e extended
p primary partition (1-4)
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide25 / 103
NXP Semiconductors
Select (default p): p
Partition number (1-4): 1
First cylinder (1-124368, default 1): 96
Last cylinder or +size or +sizeM or +sizeK (96-124368, default 124368): Using default value
124368
Command (m for help): w
The partition table has been altered!
Calling ioctl() to re-read $SD partition table
5. Check the partitions (see above) to determine the name of the partition. $PARTITION is used here to indicate the
partition to be formatted. Format the MMC/SD partitions as ext3 or ext4 type. For example, to use ext3:
Writing superblocks and filesystem accounting information: done
This filesystem will be automatically checked every 20 mounts or
180 days, whichever comes first. Use tune2fs -c or -i to override.
Booting Linux OS
6. Copy the rootfs contents to the MMC/SD card. The name may vary from the one used below. Check the directory for the
rootfs desired. (Copy the *.ext2 to NFS rootfs).
mkdir /mnt/tmpmnt
mount -t ext3 -o loop /imx-image-multimedia.ext3 /mnt/tmpmnt
cd /mnt
mkdir mmcblk0p1
mount -t ext3 /dev/$PARTITION /mnt/mmcblk0p1
cp -af /mnt/tmpmnt/* /mnt/mmcblk0p1/
umount /mnt/mmcblk0p1
umount /mnt/tmpmnt
7. Type sync to write the contents to MMC/SD.
8. Type poweroff to power down the system. Follow the instructions in Running Linux OS on the target to boot the image
from the MMC/SD card.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide26 / 103
NXP Semiconductors
Booting Linux OS
NOTE
By default, v2013.04 and later versions of U-Boot support loading the kernel image and DTB file from the SD/MMC
vfat partition by using the fatload command. To use this feature, perform the following steps:
1. Format the first partition (for example 50 MB) of the SD/MMC card with vfat filesystem.
2. Copy zImage and the DTB file into the VFAT partition after you mount the VFAT partition into your
host computer.
3. Make sure that the zImage and DTB file name are synchronized with the file name pointed to by the
U-Boot environment variables: fdt_file and image. Use the print command under U-Boot to display these
two environment variables. For example:
print fdt_file image
4. U-Boot loads the kernel image and the DTB file from your VFAT partition automatically when you boot from
the SD/MMC card.
The following is an example to format the first partition to a 50 MB vfat filesystem and format the second partition to an
ext4 filesystem:
~$ fdisk /dev/sdb
Command (m for help): n
Partition type:
p primary (0 primary, 0 extended, 4 free)
e extended
Select (default p): p
Partition number (1-4, default 1): 1
First sector (2048-30318591, default 2048): 4096
Last sector, +sectors or +size{K,M,G} (4096-30318591, default 30318591): +50M
Command (m for help): p
Disk /dev/sdb: 15.5 GB, 15523119104 bytes
64 heads, 32 sectors/track, 14804 cylinders, total 30318592 sectors
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide27 / 103
NXP Semiconductors
Booting Linux OS
Calling ioctl() to re-read partition table.
Syncing disks.
~$ mkfs.vfat /dev/mmcblk0p1
~$ mkfs.ext4 /dev/mmcblk0p2
4.5 How to boot the i.MX boards
When U-Boot is loaded onto one of the devices that support booting, the DIP switches can be used to boot from that device. The
boot modes of the i.MX boards are controlled by the boot configuration DIP switches on the board. For help locating the boot
configuration switches, see the quick start guide for the specific board as listed under References above.
The following sections list basic boot setup configurations. The tables below represent the DIP switch settings for the switch blocks
on the specified boards. An X means that particular switch setting does not affect this action.
4.5.1 Booting from an SD card in slot SD1
The following table shows the DIP switch settings for booting from the SD card slot labeled SD1 on the i.MX 7Dual SABRESD boards.
Table 2. Booting from SD1 on i.MX 7Dual SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW2OFFOFFONOFFOFFOFFOFFOFF
SW3ONOFF------
The following table shows the DIP switch settings for booting from the SD card slot labeled SD1 on the i.MX 7ULP EVK boards.
Table 3. Booting from SD1 on i.MX 7ULP EVK
SwitchD1D2D3D4
SW1ONOFFOFFON
The following table shows the bootcfg pin settings for booting from the SD card slot labeled SD1 on the i.MX 8QuadMax
MEK boards.
Table 4. Booting from SD1 on i.MX 8QuadMax MEK
SwitchD1D2D3D4D5D6D7D8
SW2OFFOFFONONOFFOFF--
The following table shows the bootcfg pin settings for booting from the SD card slot labeled SD1 on the i.MX 8QuadXPlus
MEK boards.
NOTE
This is the same setting for the i.MX 8DualX MEK and i.MX 8DXL EVK boards
Table 5. Booting from SD1 on i.MX 8QuadXPlus MEK
SwitchD1D2D3D4
SW2ONONOFFOFF
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide28 / 103
NXP Semiconductors
Booting Linux OS
4.5.2 Booting from an SD card in slot SD2
The SD card slot that is labeled SD2 indicates that this slot is connected to the uSDHC pin SD2 on the processor. Most boards
label this slot as SD2. This slot is referred to as SD2 in this document.
The following table shows the DIP switch settings for booting from the SD card slot labeled SD2 and J500 on the i.MX 6 SABRE-SD
boards. The SD2 card slot is located beside the LVDS1 connection on the back of the board.
Table 6. Booting from SD2 (J500) on i.MX 6 SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW6ONOFFOFFOFFOFFOFFONOFF
The i.MX 6UltraLite EVK board or i.MX 6ULL EVK board has one TF card slot on the CPU board. This slot uses the USDHC2
controller. The following table shows the DIP switch settings for booting from the TF slot.
Table 7. Booting from TF on i.MX 6UltraLite EVK and i.MX 6ULL EVK
SwitchD1D2D3D4
SW601OFFOFFONOFF
SW602ONOFF--
The i.MX 8M Quad EVK board has one TF card slot. This slot uses the USDHC2 controller. The following table shows the DIP
switch settings for booting from the TF slot.
Table 8. Booting from TF on i.MX 8M Quad EVK
SwitchD1D2D3D4
SW801ONONOFFOFF
SW802ONOFF--
The i.MX 8M Mini EVK board has one TF card slot. This slot uses the USDHC2 controller. The following table shows the DIP switch
settings for booting from the TF slot.
Table 9. Booting from TF on i.MX 8M Mini EVK
SwitchD1D2D3D4D5D6D7D8
SW1101OFFONOFFOFFOFFONONOFF
SW1102OFFOFFONONOFFONOFFOFF
The i.MX 8M Nano EVK board has one TF card slot. This slot uses the USDHC2 controller. The following table shows the DIP
switch settings for booting from the TF slot.
Table 10. Booting from TF on i.MX 8M Nano EVK
SwitchD1D2D3D4
SW1101ONONOFFOFF
The i.MX 8M Plus EVK board has one TF card slot. This slot uses the USDHC2 controller. The following table shows the DIP switch
settings for booting from the TF slot.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide29 / 103
NXP Semiconductors
Booting Linux OS
Table 11. Booting from TF on i.MX 8M Plus EVK
SwitchD1D2D3D4
SW4OFFOFFONON
4.5.3 Booting from an SD card in slot SD3
The SD card slot that is labeled SD3 indicates that this slot is connected to the uSDHC pin SD3 on the processor. Most boards
label this slot as SD3. This slot is referred to as SD3 in this document.
The following table shows the DIP switch settings to boot from an SD card in slot SD3 on i.MX 6 SABRE-AI boards.
Table 12. Booting from an SD card in slot SD3 on i.MX 6 SABRE-AI
SwitchD1D2D3D4D5D6D7D8D9D10
S1XXXOFFONXXXXX
S2XOFFONOFF------
S3OFFOFFONOFF------
The following table shows the DIP switch settings to boot from an SD card in slot SD3 on i.MX 6SoloX SABRE-AI boards.
Table 13. Booting from an MMC card in Slot SD3 on i.MX 6SoloX SABRE-AI
SwitchD1D2D3D4D5D6D7D8
S4OFFONOFFXOFFOFFONOFF
S3XOFFOFFOFFONONOFFOFF
S1OFFOFFONOFF----
The following table shows the DIP switch settings for booting from SD3, also labeled as J507. The SD3 slot is located between
the HDMI and UART ports.
Table 14. Booting from an SD card in slot SD3 on i.MX 6 SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW6OFFONOFFOFFOFFOFFONOFF
4.5.4 Booting from an SD card in slot SD4
The following table describes the dip switch settings for booting from an SD card in slot SD4.
The SD4 slot is on the center of the edge of the SoloX board.
Table 15. Booting from an SD card in slot SD4 on i.MX 6SoloX SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW10OFFOFFOFFOFFOFFOFFOFFOFF
SW11OFFOFFONONONOFFOFFOFF
SW12OFFONOFFOFFOFFOFFOFFOFF
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide30 / 103
NXP Semiconductors
Booting Linux OS
Table 16. Booting from an MMC card in slot SD4 on i.MX 6SoloX SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW10OFFOFFOFFOFFOFFOFFOFFOFF
SW11OFFOFFONONONOFFOFFOFF
SW12OFFONONOFFOFFOFFOFFOFF
4.5.5 Booting from eMMC
eMMC 4.4 is a chip permanently attached to the board that uses the SD4 pin connections from the i.MX 6 processor. For more
information on switch settings, see table "MMC/eMMC Boot Fusemap" in the IC reference manual.
The following table shows the boot switch settings to boot from eMMC4.4 (SDIN5C2-8G) on i.MX 6 SABRE-SD boards.
Table 17. Booting from eMMC on i.MX 6 SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW6ONONOFFONOFFONONOFF
i.MX 7Dual is different from i.MX 6. The eMMC uses the SD3 pin connections from the i.MX 7Dual processor.
Table 18. Booting from eMMC on i.MX 7Dual SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW2OFFONOFFONOFFOFFOFFOFF
SW3ONOFF------
The following table shows the boot switch settings to boot from eMMC4.4 on the i.MX 7ULP EVK boards.
Table 19. Booting from eMMC on i.MX 7ULP EVK
SwitchD1D2D3D4
SW1ONOFFOFFOFF
The following table shows the boot switch settings to boot from eMMC5.0 on the i.MX 8QuadMax MEK boards.
Table 20. Booting from eMMC on i.MX 8QuadMax MEK
SwitchD1D2D3D4D5D6D7D8
SW2OFFOFFOFFONOFFOFF--
The following table shows the boot switch settings to boot from eMMC5.0 on the i.MX 8QuadXPlus MEK boards.
NOTE
This is the same setting for the i.MX 8DualX MEK and i.MX 8DXL EVK boards, except that 8DXL EVK uses SW1.
Table 21. Booting from eMMC on i.MX 8QuadXPlus MEK
SwitchD1D2D3D4
SW2OFFONOFFOFF
The following table shows the boot switch settings to boot from eMMC5.0 on the i.MX 8M Quad EVK boards.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide31 / 103
NXP Semiconductors
Booting Linux OS
Table 22. Booting from eMMC on i.MX 8M Quad EVK
SwitchD1D2D3D4
SW801OFFOFFONOFF
The following table shows the boot switch settings to boot from eMMC5.1 on the i.MX 8M Mini EVK boards.
Table 23. Booting from eMMC on i.MX 8M Mini EVK
SwitchD1D2D3D4D5D6D7D8
SW1101OFFONONONOFFOFFONOFF
SW1102OFFOFFOFFOFFONOFFONOFF
The following table shows the boot switch settings to boot from eMMC5.1 on the i.MX 8M Nano EVK boards.
Table 24. Booting from eMMC on i.MX 8M Nano EVK
SwitchD1D2D3D4
SW1101OFFONOFFOFF
The following table shows the boot switch settings to boot from eMMC5.1 on the i.MX 8M Plus EVK boards.
Table 25. Booting from eMMC on i.MX 8M Plus EVK
SwitchD1D2D3D4
SW4OFFOFFOFFON
4.5.6 Booting from SATA
The following switch settings enable booting from SATA.
SATA booting is supported only by the i.MX 6Quad/6QuadPlus SABRE boards.
Table 26. Booting from SATA on i.MX 6 SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW6OFFOFFOFFOFFOFFONOFFOFF
4.5.7 Booting from NAND
The following table shows the DIP switch settings needed to boot from NAND on i.MX 6 SABRE-AI boards.
Table 27. Booting from NAND on i.MX 6 SABRE-AI
SwitchD1D2D3D4D5D6D7D8D9D10
S1OFFOFFOFFONOFFOFFOFFOFFOFFOFF
S2OFFOFFOFFON------
S3OFFOFFONOFF------
The following table shows the DIP switch settings needed to boot from NAND for i.MX 7Dual SABRE-SD boards.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide32 / 103
NXP Semiconductors
Booting Linux OS
Table 28. Booting from NAND on i.MX 7Dual SABRE-SD
SwitchD1D2D3D4D5D6D7D8D9D10
S2OFFONONXXXXOFF--
S3ONOFFXXXXXX--
The following table shows the DIP switch settings needed to boot from NAND for i.MX 8M Mini DDR4 EVK boards.
Table 29. Booting from NAND on i.MX 8M Mini DDR4 EVK
SwitchD1D2D3D4D5D6D7D8D9D10
SW1101OFFONOFFOFFOFFOFFOFFON--
SW1102OFFOFFOFFONONONONOFF--
4.5.8 Booting from SPI-NOR
Enable booting from SPI NOR on i.MX 6 SABRE-AI boards by placing a jumper on J3 between pins 2 and 3.
Table 30. Booting from SPI-NOR on i.MX 6 SABRE-AI
SwitchD1D2D3D4D5D6D7D8D9D10
S1XXXXXXXXXX
S2ONONOFFOFFOFFOFFOFFOFFOFFOFF
S3OFFOFFONOFF------
4.5.9 Booting from EIM (Parallel) NOR
The following table shows the DIP switch settings to boot from NOR.
Table 31. Booting from EIM NOR on i.MX 6 SABRE-AI
SwitchD1D2D3D4D5D6D7D8D9D10
S1XXXOFFOFFONXXXX
S2XOFFOFFOFF------
S3OFFOFFONOFF------
NOTE
SPI and EIM NOR have pin conflicts on i.MX 6 SABRE-AI boards. Neither can be used for the same configuration.
The default U-Boot configuration is set to SPI NOR.
4.5.10 Booting from QuadSPI or FlexSPI
The following tables list the DIP switch settings for booting from QuadSPI.
Table 32. Booting from QuadSPI on i.MX 6SoloX SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW10OFFOFFOFFOFFOFFOFFOFFOFF
Table continues on the next page...
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide33 / 103
NXP Semiconductors
Booting Linux OS
Table 32. Booting from QuadSPI on i.MX 6SoloX SABRE-SD (continued)
SwitchD1D2D3D4D5D6D7D8
SW11OFFOFFOFFOFFOFFOFFOFFOFF
SW12OFFOFFOFFONONOFFOFFOFF
Table 33. Booting from QuadSPI on i.MX 6SoloX SABRE-AI
SwitchD1D2D3D4D5D6D7D8
SW4OFFOFFOFFOFFONOFFOFFOFF
SW3OFFOFFOFFOFFOFFOFFOFFOFF
SW1OFFOFFONOFF----
Table 34. Booting from QuadSPI on i.MX 7Dual SABRE-SD
SwitchD1D2D3D4D5D6D7D8
SW2ONOFFOFFOFFOFFOFFOFFOFF
SW3ONOFF------
Table 35. Booting from QuadSPI on i.MX 6UltraLite EVK and i.MX 6ULL EVK
SwitchD1D2D3D4
SW601OFFOFFOFFOFF
SW602ONOFF--
Table 36. Booting from FlexSPI on i.MX 8QuadMax MEK
SwitchD1D2D3D4D5D6D7D8
SW2OFFOFFOFFONONOFF--
Table 37. Booting from FlexSPI on i.MX 8QuadXPlus MEK
SwitchD1D2D3D4
SW2OFFONONOFF
Table 38. Booting from FlexSPI on i.MX 8M Mini LPDDR4 EVK
SwitchD1D2D3D4D5D6D7D8
SW1101OFFONOFFOFFOFFOFFOFFOFF
SW1102OFFONOFFOFFOFFOFFOFFON
Table 39. Booting from QuadSPI on i.MX Nano EVK
SwitchD1D2D3D4
SW601OFFOFFOFFOFF
SW602ONOFF--
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide34 / 103
NXP Semiconductors
Booting Linux OS
Table 40. Booting from QuadSPI on i.MX 8M Plus EVK
SwitchD1D2D3D4
SW4OFFONONOFF
4.5.11 Serial download mode for the Manufacturing Tool
No dedicated boot DIP switches are reserved for serial download mode on i.MX 6 SABRE-SD. There are various ways to enter
serial download mode. One way is to set the boot mode to boot from SD slot SD3 (set SW6 DIP switches 2 and 7 to on, and the rest
are off). Do not insert the SD card into slot SD3, and power on the board. After the message "HID Compliant device" is displayed,
the board enters serial download mode. Then insert the SD card into SD slot SD3. Another way to do this is to configure an invalid
boot switch setting, such as setting all the DIP switches of SW6 to off.
The following table shows the boot switch settings for i.MX 6 SABRE-AI boards, which are used to enter serial download mode
for the Manufacturing Tool. If the boot image in the boot media is not validated, the system also enters the serial download mode.
Table 41. Setup for the Manufacturing Tool on i.MX 6 SABRE-AI
SwitchD1D2D3D4
S3OFFONOFFOFF
Table 42. Setup for the Manufacturing Tool on i.MX 7Dual SABRE-SD
SwitchD1D2D3D4
S3OFFON--
Table 43. Setup for Manufacturing Tool on i.MX 6UltraLite EVK and i.MX 6ULL EVK
SwitchD1D2
SW602OFFON
Table 44. Setup for Manufacturing Tool on i.MX 7ULP EVK
SwitchD1D2D3D4
SW1OFFON--
Table 45. Setup for Manufacturing Tool on i.MX 8M Quad EVK
SwitchD1D2
SW802OFFON
Table 46. Setup for Manufacturing Tool on i.MX 8M Plus EVK
SwitchD1D2D3D4
SW4OFFOFFOFFON
Table 47. Setup for Manufacturing Tool on i.MX 8M Mini EVK
SwitchD1D2D3D4D5D6D7D8
SW1101ONOFFXXXXXX
SW1102XXXXXXXX
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide35 / 103
NXP Semiconductors
Table 48. Setup for Manufacturing Tool on i.MX 8M Nano EVK
SwitchD1D2D3D4
SW1101ONOFFOFFOFF
Table 49. Setup for Manufacturing Tool on i.MX 8QuadMax MEK
SwitchD1D2D3D4D5D6
SW2OFFOFFONOFFOFFOFF
NOTE
The following settings are same for the i.MX 8DualX MEK and i.MX 8DXL EVK boards (8DXL EVK uses SW1).
Table 50. Setup for Manufacturing Tool on i.MX 8QuadXPlus MEK
SwitchD1D2D3D4
SW2ONOFFOFFOFF
NOTE
If the SD card with bootable image is plugged in SD2 (baseboard), ROM will not fall back into the serial
download mode.
Booting Linux OS
4.5.12 How to build U-Boot and Kernel in standalone environment
To build U-Boot and Kernel in a standalone environment, perform the following steps.
First, generate an SDK, which includes the tools, toolchain, and small rootfs to compile against to put on the host machine.
• Generate an SDK from the Yocto Project build environment with the following command. To set up the Yocto Project build
environment, follow the steps in the
Target-Machine to the machine you are building for. The populate_sdk generates a script file that sets up a standalone
environment without Yocto Project. This SDK should be updated for each release to pick up the latest headers, toolchain,
and tools from the current release.
If the building process is interrupted, modify conf/local.conf to comment out the line: PACKAGE_CLASSES =
"package_deb", and then execute the populate_sdk command again.
• From the build directory, the bitbake was run in, copy the sh file in tmp/deploy/sdk to the host machine to build on and
execute the script to install the SDK. The default location is in /opt but can be placed anywhere on the host machine.
On the host machine, these are the steps to build U-Boot and Kernel:
• For i.MX 8 builds on the host machine, set the environment with the following command before building.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide36 / 103
NXP Semiconductors
Booting Linux OS
• To build an i.MX 8 U-Boot in the standalone environment, find the configuration for the target boot. In the following
example, i.MX 8QuadMax MEK board is the target and it runs on the Arm Cortex-A53 core by default. SPL image
(u-boot-spl.bin) is also generated with the default defconfig. It is needed when booting with OP-TEE image.
make distclean
make imx8qm_mek_defconfig
make
For i.MX 8QuadXPlus MEK and i.MX 8DualX board:
make distclean
make imx8qxp_mek_defconfig
make
For i.MX 8DXL EVK board:
make distclean
make imx8dxl_evk_defconfig
make
• To build U-Boot for i.MX 8M Quad EVK:
make distclean
make imx8mq_evk_defconfig
make
• For i.MX 8M LPDDR4 EVK:
make distclean
make imx8mm_evk_defconfig
make
• For i.MX 8M DDR4 EVK:
make distclean
make imx8mm_ddr4_evk_defconfig
make
• For i.MX 8M Plus LPDDR4 EVK board:
make distclean
make imx8mp_evk_defconfig
make
• To build an i.MX 6 or i.MX 7 U-Boot in the standalone environment, find the configuration for the target boot. In the
following example, i.MX 6ULL is the target.
Download source by cloning with git clone https://source.codeaurora.org/external/imx/uboot-imx -b
lf_v2020.04
cd uboot-imx
make clean
make mx6ull_14x14_evk_defconfig
make
• To build the kernel in the standalone environment for i.MX 8, execute the following commands:
make imx_v8_defconfig
make
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide37 / 103
NXP Semiconductors
• To build the kernel in the standalone environment for i.MX 6 and i.MX 7, execute the following commands:
Download source by cloning with git clone https://source.codeaurora.org/external/imx/linux-imx -b
lf-5.10.y
cd linux-imx
make imx_v7_defconfig
make
NOTE
Users need to modify configurations for fused parts. For example, the i.MX 6UltraLite has four parts, G0, G1, G2,
and G3.
4.5.13 How to build imx-boot image by using imx-mkimage
For i.MX 8QuadMax, to build imx-boot image by using imx-mkimage, perform the following steps:
1. Copy u-boot.bin from u-boot/u-boot.bin to imx-mkimage/iMX8QM/.
2. Copy scfw_tcm.bin from SCFW porting kit to imx-mkimage/iMX8QM/.
3. Copy bl31.bin from Arm Trusted Firmware (imx-atf) to imx-mkimage/iMX8QM/.
4. Copy the SECO firmware container image (ahab-container.img) to imx-mkimage/iMX8QM/.
5. Run make SOC=iMX8QM flash to generate flash.bin.
6. If using OP-TEE, copy tee.bin to imx-mkimage/iMX8QM/ and copy u-boot/spl/u-boot-spl.bin to imx-mkimage/iMX8QM/.
Run make SOC=iMX8QM flash_spl to generate flash.bin.
For i.MX 8QuadXPlus, to build imx-boot image by using imx-mkimage, perform the following steps:
1. Copy u-boot.bin from u-boot/u-boot.bin to imx-mkimage/iMX8QX/.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide38 / 103
NXP Semiconductors
Booting Linux OS
2. Copy scfw_tcm.bin from SCFW porting kit to imx-mkimage/iMX8QX/.
3. Copy bl31.bin from Arm Trusted Firmware (imx-atf) to imx-mkimage/iMX8QX/.
4. Copy the SECO firmware container image (ahab-container.img) to imx-mkimage/iMX8QX/.
5. Run make SOC=iMX8QX flash to generate flash.bin.
6. If using OP-TEE, copy tee.bin to imx-mkimage/iMX8QX/ and copy u-boot/spl/u-boot-spl.bin to imx-mkimage/iMX8QX/.
Run make SOC=iMX8QX flash_spl to generate flash.bin.
For i.MX 8DXL, to build imx-boot image by using imx-mkimage, perform the following steps:
1. Copy u-boot.bin from u-boot/u-boot.bin to imx-mkimage/iMX8DXL/.
2. Copy scfw_tcm.bin from SCFW porting kit to imx-mkimage/iMX8DXL/.
3. Copy bl31.bin from Arm Trusted Firmware (imx-atf) to imx-mkimage/iMX8DXL/.
4. Copy the image of SECO firmware container (mx8dxla1-ahab-container.img) to imx-mkimage/iMX8DXL/.
5. Run make SOC=iMX8DXL flash to generate flash.bin.
6. If using OP-TEE, copy tee.bin to imx-mkimage/iMX8DXL/ and copy u-boot/spl/u-boot-spl.bin to imx-mkimage/ iMX8DXL/.
Run make SOC=iMX8DXL flash_spl to generate flash.bin.
7. If skipping loading V2X firmware, add V2X=NO to make command, like make SOC=iMX8DXL V2X=NO flash.
The following is a matrix table for targets of i.MX 8QuadMax and i.MX 8QuadXPlus.
Table 51. Matrix table for targets of i.MX 8QuadMax, i.MX 8QuadXPlus, and i.MX 8DXL
-OPTEEU-BootSPLCortex-M4
flash_splYesYesYesNo
flashNoYesNoNo
flash_linux_m4YesYesYesYes
flash_regression_linux_m4NoYesNoYes
For i.MX 8M EVK boards to build imx-boot image by using imx-mkimage, perform the following steps:
1. Copy and rename mkimage from u-boot/tools/mkimage to imx-mkimage/iMX8M/mkimage_uboot.
2. Copy u-boot-spl.bin from u-boot/spl/u-boot-spl.bin to imx-mkimage/iMX8M/.
3. Copy u-boot-nodtb.bin from u-boot/u-boot-nodtb.bin to imx-mkimage/iMX8M/.
4. Copy imx8mq-evk.dtb (for i.MX 8M Quad EVK) or imx8mm-evk.dtb (for i.MX 8M Mini LPDDR4 EVK) or imx8mm-ddr4evk.dtb (for i.MX 8M Mini DDR4 EVK) or imx8mp-evk.dtb (for i.MX 8M Plus LPDDR4 EVK) from u-boot/arch/arm/dts/ to
imx-mkimage/iMX8M/.
5. Copy bl31.bin from ARM Trusted Firmware (imx-atf) to imx-mkimage/iMX8M/.
6. Copy firmware/hdmi/cadence/signed_hdmi_imx8m.bin from firmware-imx package to imx-mkimage/iMX8M/.
7. For i.MX 8M Quad and i.MX 8M Mini LPDDR4 EVK, copy lpddr4_pmu_train_1d_dmem.bin,
lpddr4_pmu_train_1d_imem.bin, lpddr4_pmu_train_2d_dmem.bin, and lpddr4_pmu_train_2d_imem.bin from
firmware/ddr/synopsys of firmware-imx package to imx-mkimage/iMX8M/.
For i.MX 8M Mini DDR4 EVK, copy ddr4_imem_1d.bin, ddr4_dmem_1d.bin, ddr4_imem_2d.bin, and ddr4_dmem_2d.bin
from firmware/ddr/synopsys of firmware-imx package to imx-mkimage/iMX8M.
For i.MX 8M Plus LPDDR4 EVK, copy lpddr4_pmu_train_1d_dmem_201904.bin, lpddr4_pmu_train_1d_imem_201904.bin,
lpddr4_pmu_train_2d_dmem_201904.bin, and lpddr4_pmu_train_2d_imem_201904.bin from firmware/ddr/synopsys of
firmware-imx package to imx-mkimage/iMX8M/.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide39 / 103
NXP Semiconductors
Booting Linux OS
8. For i.MX 8M Quad EVK, run make SOC=iMX8M flash_evk to generate flash.bin (imx-boot image) with HDMI FW
included. For i.MX 8M Mini LPDDR4 EVK, run make SOC=iMX8MM flash_evk to generate flash.bin (imx-boot image).
For i.MX 8M Mini DDR4 EVK, run make SOC=iMX8MM flash_ddr4_evk to generate flash.bin (imx-boot image).
For i.MX 8M Plus LPDDR4 EVK, run make SOC=iMX8MP flash_evk to generate flash.bin (imx-boot-image).
4.6 Flash memory maps
This section describes the software layout in memory on memory devices used on the i.MX boards.
This information is useful for understanding subsequent sections about image downloading and how the images are placed
in memory.
The mtdparts directive can be used in the Linux boot command to specify memory mapping. The following example briefly
describes how to use memory maps. Memory is allocated in the order of how it is listed. The dash (-) indicates the the rest of
the memory.
mtdparts=[memory type designator]:[size]([name of partition]),[size]([name of partition]),-([name of
final partition])
4.6.1 MMC/SD/SATA memory map
The MMC/SD/SATA memory scheme is different from the NAND and NOR flash, which are deployed in the BSP software. The
MMC/SD/SATA must keep the first sector (512 bytes) as the Master Boot Record (MBR) to use MMC/SD as the rootfs.
Upon boot-up, the MBR is executed to look up the partition table to determine which partition to use for booting. The bootloader
should be after the MBR. The kernel image and rootfs may be stored at any address after the bootloader. By default, the the U-Boot
boot arguments uses the first FAT partition for kernel and DTB, and the following ext3 partition for the root file system. Alternatively,
users can store the kernel and the DTB in any raw memory area after the bootloader. The boot arguments must be updated to
match any changed memory addresses.
The MBR can be generated through the fdisk command when creating partitions in MMC/SD cards on a Linux host server.
4.6.2 NAND flash memory map
The NAND flash memory map is configured from the Linux kernel command line.
The default configuration contains only one parallel NOR partition. The parallel NOR device is generally 4 MB. U-Boot is loaded
at the beginning of parallel NOR so that the device can boot from it. The default configuration is that on boot up, U-Boot loads the
kernel, DTB, and root file system from the SD/MMC card into DDRAM. The end user can change the default settings according to
their needs. More partitions can be added through the kernel command line. The memory type designator for the command below
consists of the NOR address and the designator. This information can be found in the imx .dtsi device tree file in arch/arm/boot/dts.
The following is an example of what might be added to the Linux boot command line:
mtdparts=8000000.nor:1m(uboot),-(rootfs)
The address for parallel NOR is 0x8000000 for i.MX 6 SABRE-AI.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide40 / 103
NXP Semiconductors
Booting Linux OS
4.6.4 SPI-NOR flash memory map
The SPI-NOR flash memory can be configured using the Linux kernel command line.
U-Boot should be loaded at the 1 KB offset of the SPI-NOR memory, so that the device can boot from it. The default configuration is
that on boot up, U-Boot loads the kernel, DTB, and root file system from the SD/MMC card into DDRAM. The end user can change
the default settings according to their needs. More partitions can be added through the kernel command line. The following is an
example of what might be added to the Linux boot command line:
The QuadSPI flash memory can be configured using the Linux kernel command line.
U-Boot is loaded at the beginning of the QuadSPI memory so that the device can boot from it. The default configuration is that on
boot up, U-Boot loads the kernel, DTB, and root file system from the SD/MMC card into DDRAM. The end user can change the
default settings according to their requirements. More partitions can be added through the kernel command line. The following is
an example of what might be added to the Linux boot command line:
QuadSPI1 Port B CS0sf probe 1:0 on i.MX 6 SoloX SABRE-
AI board
QuadSPI2 Port A CS0sf probe 0:0 on i.MX 6SoloX SABRE-
SD board
QuadSPI2 Port B CS0sf probe 1:0 on i.MX 6SoloX SABRE-
SD board
0x60000000
0x08000000
0x68000000
0x70000000
0x78000000
-
-
-
-
4.7 Running Linux OS on the target
This section describes how to run a Linux image on the target using U-Boot.
These instructions assume that you have downloaded the kernel image using the instructions in Downloading images or Preparing
an SD/MMC card to boot. If you have not set up your Serial Terminal, see Basic Terminal Setup.
The basic procedure for running Linux OS on an i.MX board is as follows. This document uses a specific set of environment
variable names to make it easier to describe the settings. Each type of setting is described in its own section as follows.
1. Power on the board.
2. When U-Boot comes up, set the environment variables specific to your machine and configuration. Common settings
are described below and settings specific to a device are described in separate sections.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide41 / 103
NXP Semiconductors
Booting Linux OS
3. Save the environment setup:
U-Boot > saveenv
4. Run the boot command:
U-Boot > run bootcmd
The commands env default -f -a and saveenv can be used to return to the default environment.
Specifying the console
The console for debug and command-line control can be specified on the Linux boot command line. The i.MX 6Quad SABRE-AI
board uses ttymxc2, so it is not same for all boards. It is usually specified as follows, but the baudrate and the port can be modified.
Therefore, for NFS, it might be ttymxc3.
For the i.MX 7ULP EVK, i.MX 8QuadMax MEK boards, and i.MX 8QuadXPlus MEK board, change to " console=ttyLP0,115200".
Specifying displays
The display information can be specified on the Linux boot command line. It is not dependent on the source of the Linux image.
If nothing is specified for the display, the settings in the device tree are used. Add ${displayinfo} to the environment macro
containing bootargs. The specific parameters can be found in the
i.MX Linux® Release Notes
(IMXLXRN). The following are some
examples of these parameters.
• U-Boot > setenv displayinfo 'video=mxcfb0:dev=hdmi,1920x1080M@60,if=RGB24' for an HDMI display
• U-Boot > setenv displayinfo 'video=mxcfb1:dev=ldb video=mxcfb0:dev=hdmi,1920x1080M@60,if=RGB24' for
LVDS and HDMI dual displays
• U-Boot > setenv displayinfo 'video=mxcfb0:dev=lcd,if=RGB565' for an LCD
• U-Boot > setenv displayinfo 'video=mxcepdcfb:E060SCM,bpp=16 max17135:pass=2,vcom=-2030000' for an EPDC
connection
video=mxcfb1:dev=hdmi,1920x1080M@60,if=RGB24' for LCD and HDMI dual displays
Specifying memory addresses
The addresses in the memory where the kernel and device tree are loaded to do not change based on the device that runs Linux
OS. The instructions in this chapter use the environment variables loadaddr and ftd_addr to indicate these values. The following
table shows the addresses used on different i.MX boards.
Table 53. Board-specific default values
Variable6Quad,
6QuadPlus,
and
6DualLite
6SoloX SD7Dual
SABRE-SD
6UltraLite,
6ULL and
6ULZ EVK
7ULP EVK8QuadMax,
8QuadXPlu
s, 8DualX,
and 8DXL
8M Quad/
8M Mini
EVK
Description
SABRE (AI
and SD)
loadaddr0x120000000x808000000x808000000x808000000x608000000x802800000x40480000Address in
the memory
the kernel
Table continues on the next page...
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
fdt_addr0x180000000x830000000x830000000x830000000x630000000x830000000x43000000Address in
the memory
the device
tree code
are copied
to
In addition, fdt_file is used to specify the filename of the device tree file. The commands used to set the U-Boot environment
variables are as follows:
U-Boot > setenv loadaddr 0x80080000
U-Boot > setenv fdtaddr 0x80f00000
U-Boot > setenv fdt_file fsl-imx7ulp-evk.dtb
Specifying the location of the root file system
The rootfs can be located on a device on the board or on NFS. The settings below show some options for specifying these.
6Solo, and 6UltraLite can specify uart_from_osc on the command line to specify that the OSC clock rather than PLL3 should be
used. This allows the system to enter low power mode.
U-Boot > setenv special 'uart_from_osc'
Creating the boot command line
For clarification, this document groups the bootargs into one macro as follows:
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide43 / 103
NXP Semiconductors
Booting Linux OS
Power on the board, and then enter the commands provided. The following settings may be used to boot the Linux system
from NAND.
Assume that the kernel image starts from the address 0x1400000 byte (the block starting address is 0x800). The kernel image
size is less than 0x400000 byte. The rootfs is located in /dev/mtd2.
QuadSPI is available on i.MX 6SoloX SABRE-SD boards, i.MX 7Dual SABRE-SD boards, i.MX 6UltraLite EVK boards, i.MX 6ULL
EVK boards, and i.MX 8QuadMax MEK, and i.MX 8QuadXPlus MEK. The following procedure may be used to boot the Linux
system from QuadSPI NOR.
1. Assume that the kernel image starts from the address 0xA00000 byte and the DTB file starts from address 0x800000.
2. At the U-Boot prompt, set the following environment variables:
U-Boot > setenv bootcmd 'run bootargsset; sf probe; sf read ${loadaddr} 0xA00000 0x2000; sf read
On the i.MX 6SoloX boards, there are two ways to boot Arm Cortex-M4 images in U-Boot:
• Arm Cortex-M4 processor Normal Up (supported on i.MX 6SoloX SABRE-AI and SABRE-SD boards). Performed by
running the U-Boot command. Requires:
1. U-Boot normal SD image if Arm Cortex-A9 processor boots from the SD card. U-Boot normal QSPI image if Arm
Cortex-A9 processor boots from the QSPI NOR flash.
2. Kernel DTB: imx6sx-sdb-m4.dtb for i.MX 6SoloX SABRE-SD board. imx6sx-sabreauto-m4.dtb for i.MX 6SoloX
SABRE-AI board.
3. Have the Arm Cortex-M4 image burned. (NOR flash of QuadSPI2 PortB CS0 for i.MX 6SoloX SABRE-SD board.
NOR flash of QuadSPI1 PortB CS0 for i.MX 6SoloX SABRE-AI board.)
• Arm Cortex-M4 processor Fast Up (only supported on i.MX 6SoloX SABRE-SD boards). Initiated by U-Boot at a very early
boot phase to meet the requirement of Arm Cortex-M4 processor booting in 50 ms. No U-Boot command is involved.
Requires:
1. U-Boot Arm Cortex-M4 fast up image and Arm Cortex-A9 processor must boot from the QSPI2 NOR flash.
2. Kernel DTB: imx6sx-sdb-m4.dtb.
3. Have the Arm Cortex-M4 image burned (NOR flash of QuadSPI2 PortB CS0).
To facilitate the Arm Cortex-M4 processor Normal Up, a script has been added to the default U-Boot. The following steps may help
users who need to run the Cortex-M4 processor Normal Up script.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide44 / 103
NXP Semiconductors
Booting Linux OS
1. Power on the board.
2. On the i.MX 6SoloX SABRE-SD board, assumed that the Arm Cortex-M4 image is at address 0x78000000 (NOR flash
of QuadSPI2 PortB CS0). On the i.MX 6SoloX SABRE-AI board, assumed that the Arm Cortex-M4 image is at address
0x68000000 (NOR flash of QuadSPI1 PortB CS0).
At the U-Boot prompt:
U-Boot > run m4boot
Or users can perform the commands without depending on the script:
U-Boot > sf probe 1:0
For the i.MX 6SoloX SABRE-SD board:
U-Boot > bootaux 0x78000000
For the i.MX 6SoloX SABRE-AI board:
U-Boot > bootaux 0x68000000
NOTE
For how to add the MCC demo to the kernel and limit RAM available to kernel to use it, see Chapter 53 "i.MX 6
SoloX MultiCore Communication (MCC)" of the
i.MX Linux® Reference Manual
(IMXLXRM).
As well as supporting running the Arm Cortex-M4 image from QuadSPI, the default i.MX 7Dual SABRE-SD board supports loading
the Arm Cortex-M4 image from the SD card and running it on OCRAM.
Prepare the Arm Cortex-M4 image to the FAT partition of the SD card. Name the file to "m4_qspi.bin" when using "m4boot" script.
After the board is powered on, the following information is displayed at the U-Boot prompt:
U-Boot > run m4boot
Or perform the commands without depending on the script:
u-boot=> fatload mmc 0:0 ${loadaddr} m4_qspi.bin
u-boot=> cp.b ${loadaddr} 0x7e0000 ${filesize}
u-boot=> bootaux 0x7e0000
On the i.MX 8M boards, perform the commands to boot the Arm Cortex-M Core core:
u-boot=> fatload mmc 1:1 ${loadaddr} m4.bin
u-boot=> cp.b ${loadaddr} 0x7e0000 ${filesize}
u-boot=> bootaux 0x7e0000
On the i.MX 8QuadMax and i.MX 8QuadXPlus boards, there are two ways to boot the Arm Cortex-M4 cores:
• Booting from ROM
Users need to use imx-mkimage to pack the Arm Cortex-M4 images into imx-boot image. It is necessary to specify the core
ID and its TCML address in the build command. The following is an example:
NOTE
When booting with the packed Cortex-M4 image (flash_linux_m4), use kernel DTB with RPMSG enabled, like
fsl-imx8qm-mek-rpmsg.dtb for i.MX 8QuadMax MEK or fsl-imx8qxp-mek-rpmsg.dtb for i.MX 8QuadXPlus MEK.
• Booting from U-Boot (not support multiple partitions enabled)
U-Boot supports loading the Arm Cortex-M4 image from the FAT partitions of the SD card with default name "m4_0.bin" and
"m4_1.bin". After the board is booted into the U-Boot console, use the following command to boot Arm Cortex-M4 core 0:
U-Boot > run m4boot_0
Or the command to boot M4 core 1:
U-Boot > run m4boot_1
Or perform the commands for core 0 without depending on the script:
U-Boot > fatload mmc 1:1 0x80280000 m4_0.bin
U-Boot > dcache flush; bootaux 0x80280000 0
4.7.5 Linux OS login
The default login username for the i.MX Linux OS is root with no password.
4.7.6 Running Linux OS from MMC/SD
This scenario assumes that the board is configured to boot U-Boot, that the Linux kernel image is named zImage and is stored
on the SD card in an MSDOS FAT partition, and one or more device tree files are also stored in this partition. The rootfs is also
stored on the SD/MMC card in another partition.
When U-Boot boots up, it detects the slot where it is booting from and automatically sets mmcdev and mmcroot to use the rootfs on
that SD card. In this scenario, the same SD card can be used to boot from any SD card slot on an i.MX 6/7 board, without changing
any U-Boot settings. From the U-Boot command line, type boot to run Linux OS.
The following instructions can be used if the default settings are not desired.
Set mmcautodetect to "no" to turn off the automatic setting of the SD card slot in mmcdev and mmcroot. The U-Boot mmcdev is based
on the soldered SD/MMC connections, so it varies depending on the board. The U-Boot mmc dev 0 is the lowest numbered SD
slot present, 1 is the next, and so on. The Linux kernel, though, indexes all the uSDHC controllers whether they are present or
not. The following table shows this mapping.
Table 54. Linux uSDHC relationships
uSDHCmmcroot
uSDHC 1mmcblk0*
uSDHC 2mmcblk1*
uSDHC 3mmcblk2*
uSDHC 4mmcblk3*
In the default configuration of the SD card and the example here, U-Boot is at the 1024 byte offset, 32 KB offset for the i.MX
8QuadXPlus B0 and i.MX 8QuadMax B0, or 33 KB offset for the i.MX 8QuadXPlus A0, i.MX 8QuadMax A0, i.MX 8M EVK boards
before the first partition, partition 1 is the partition with the Linux kernel and device trees, and partition 2 is the rootfs.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide46 / 103
NXP Semiconductors
Booting Linux OS
Setting up the environment variables
For convenience, this document uses a standard set of variables to describe the information in the Linux command line. The values
used here may be different for different machines or configurations. By default, U-Boot supports setting mmcdev and mmcroot
automatically based on the uSDHC slot that we boot from. This assumes zImage, the device tree file (DTB), and the rootfs are on
the same SD/MMC card. To set these environment variables manually, set mmcautodetect to no to disable the feature.
The following is one way to set up the items needed to boot Linux OS.
U-Boot > setenv bootcmd 'mmc dev ${mmcdev}; run loadkernel; run mmcargs; run loadfdt; bootz $
{loadaddr} - ${fdt_addr};'
The descriptions of the variables used above are as follows:
• mmcpart - This is the partition on the MMC/SD card containing the kernel image.
• mmcroot - The location of the root file system on the MMC SD card along with directives for the boot command for the
rootfs.
NOTE
The U-Boot environment on the pre-built SD card does not match this. It is more complex so that it can automatically
deal with more variations. The example above is designed to be easier to understand and use manually.
Reading the kernel image from eMMC
eMMC has user area, boot partition 1, and boot partition 2. To switch between the eMMC partitions, the user needs to use the
command mmc dev [dev id] [partition id]. For example,
mmc dev 2 0 ---> user area
mmc dev 2 1 ---> boot partition 1
mmc dev 2 2 ---> boot partition 2
4.7.7 Running the Linux image from NFS
To boot from NFS, set the following environment variables at the U-Boot prompt:
U-Boot > setenv serverip <your server IP>
U-Boot > setenv image <your kernel zImage name on the TFTP server>
U-Boot > setenv fdt_file <your dtb image name on the TFTP server>
If the MAC address has not been burned into the fuses, set the MAC address to use the network in U-Boot.
setenv ethaddr xx:xx:xx:xx:xx:xx
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide47 / 103
NXP Semiconductors
Chapter 5
Enabling Solo Emulation
Solo emulation can be enabled on the i.MX 6DualLite SABRE-SD and i.MX 6DualLite SABRE-AI boards. This is achieved by using
a specific U-Boot configuration in the bootloader build process.
When this Solo emulation is enabled on the i.MX 6DualLite SABRE platforms, the capabilities of the i.MX 6DualLite change to
the following:
• For solo emulation, use 6DualLite DTB and add maxcpus=1 to bootcmd of U-Boot.
• 32-bit data bus on DDR RAM.
• 1 GB of RAM for i.MX 6DualLite SABRE-AI.
• 512 MB of RAM for i.MX 6DualLite SABRE-SD.
To build U-Boot for an i.MX 6Solo on an i.MX 6DualLite SABRE-SD card, use the following command:
MACHINE=imx6solosabresd bitbake u-boot-imx
To build U-Boot for an i.MX 6Solo on an i.MX 6DualLite SABRE-AI card, use the following command:
MACHINE=imx6solosabreauto bitbake u-boot-imx
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide48 / 103
NXP Semiconductors
Chapter 6
Power Management
The i.MX power management uses the standard Linux interface. Check the standard Linux power documentation for information
on the standard commands. The
available and other i.MX-specific information in the power management section.
There are three main power management techniques on i.MX boards: suspend and resume commands, CPU frequency scaling,
and bus frequency scaling. They are described in the following sections.
6.1 Suspend and resume
The power state can be changed by setting the standard Linux state, /sys/power/state. The command used to set the power
state into suspend mode, available from the command line, is echo mem > /sys/power/state. The value mem can be replaced
by any of the valid power states, as described by the
Use one of the following methods to wake up the system from suspend mode.
• The debug UART can be set as a wakeup source with:
(IMXLXRM) contains information on the power modes that are
(IMXLXRM).
NOTE
It is ttylp0 for i.MX 8QuadXPlus and i.MX 8QuadMax, and ttyLP0 for i.MX 8DXL.
• RTC can be used to enter and exit from suspend mode by using the command:
/unit_test/SRTC/rtcwakeup.out -d rtc0 -m mem -s 10
This command indicates to sleep for 10 secs. This command automatically sets the power state to mem mode.
6.2 CPU frequency scaling
Scaling governors are used in the Linux kernel to set the CPU frequency. CPU frequencies can be scaled automatically depending
on the system load either in response to ACPI events or manually by userspace programs. For more information about governors,
read governors.txt from www.kernel.org/doc/Documentation/cpu-freq/governors.txt.
The following are some of the more frequently used commands:
These commands return information about the system and the current settings.
• The kernel is pre-configured to support only certain frequencies. The list of frequencies currently supported can be
obtained from:
• The following two commands set the scaling governor to a specified frequency, if that frequency is supported. If the
frequency is not supported, the closest supported frequency is used:
This release does not support the bus frequency scaling feature on i.MX 7ULP EVK.
This release does not support the bus frequency scaling feature on i.MX 8QuadXPlus and i.MX 8QuadMax.
The system automatically adjusts the bus frequency (DDR, AHB, etc.) for optimal performance based on the devices that
are active.
The bus frequency driver is enabled by default. The following DDR frequencies are supported:
• Normal DDR frequency – Default frequency in U-Boot
• Audio DDR frequency – 50 MHz on i.MX 6Quad, i.MX 6DualLite, and i.MX 6SoloX, and 100 MHz on i.MX 7Dual.
• Low power idle DDR frequency – 24 MHz
On the i.MX 8M board:
• For LPDDR4, the Audio DDR frequency is 25 MHz, the low power idle DDR frequency is 25 MHz.
• For DDR4, the audio DDR frequency is 166 MHz, the low power idle DDR frequency is 166 MHz.
To enter a low power idle DDR frequency, ensure that all devices that require high DDR frequency are disabled. Most drivers do
active clock management, but certain commands can be used to avoid waiting for timeouts to occur:
echo 1 > /sys/class/graphics/fb0/blank -> to blank the display (may need to blank fb1, fb2, and so on, if more than one
display is active).
ifconfig eth0 down -> disables the Ethernet module. On i.MX 6SoloX, i.MX 7Dual, i.MX 6UltraLite, and i.MX 6UltraLiteLite
should also disable Ethernet 1 (eth1).
i.MX 8M Plus needs some additional steps to enable USB runtime PM:
echo auto > /sys/bus/platform/devices/32f10100.usb/38100000.dwc3/power/control
echo auto > /sys/bus/platform/devices/32f10108.usb/38200000.dwc3/power/control
echo auto > /sys/bus/platform/devices/32f10108.usb/38200000.dwc3/xhci-hcd.1.auto/power/control
On most systems, the chip enters low power IDLE mode after the above two commands are executed.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide50 / 103
NXP Semiconductors
Power Management
To manipulate bus frequency, use the following commands to achieve the results desired:
cat /sys/bus/platform/drivers/imx_busfreq/soc\:busfreq/enable -> displays the status of bus frequency.
echo 0 > /sys/bus/platform/drivers/imx_busfreq/soc\:busfreq/enable -> disables bus frequency.
echo 1 > /sys/bus/platform/drivers/imx_busfreq/soc\:busfreq/enable -> enables bus frequency.
The
i.MX Linux® Reference Manual
(IMXLXRM) has more information on the bus frequency in the chapter about DVFS.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide51 / 103
NXP Semiconductors
Chapter 7
Multimedia
i.MX provides audio optimized software codecs, parsers, hardware acceleration units, and associated plugins. The i.MX provides
GStreamer plugins to access the i.MX multimedia libraries and hardware acceleration units. This chapter provides various
multimedia use cases with GStreamer command line examples.
7.1 i.MX multimedia packages
Due to license limitations, i.MX multimedia packages can be found in two locations:
• Standard packages: provided on the NXP mirror.
• Limited access packages: provided on nxp.com with controlled access.
For details, see the
i.MX Release Notes
(IMXLXRN).
7.2 Building limited access packages
Place the limited access package in the downloads directory and read the readme file in each package.
For example, README-microsoft in the package imxcodec-microsoft-$version.tar.gz.
7.3 Multimedia use cases
GStreamer is the default multimedia framework on Linux OS. The following sections provide examples of GStreamer commands
to perform the specific functions indicated. The following table shows how this document refers to common functions and what
the actual command is.
One option is to set these as environment variables as shown in the following examples. Use the full path to the command on
your system.
export GSTL=gst-launch-1.0
export PLAYBIN=playbin
export GPLAY=gplay-1.0
export GSTINSPECT=gst-inspect-1.0
In this document, variables are often used to describe the command parameters that have multiple options. These variables
are of the format $description where the type of values that can be used are described. The possible options can
be found in the Section about Multimedia in the
"gstreamer.freedesktop.org/ for the community options.
The GStreamer command line pipes the input through various plugins. Each plugin section of the command line is marked by an
exclamation mark (!). Each plugin can have arguments of its own that appear on the command line after the plugin name and
before the next exclamation mark (!). Use $GSTINSPECT $plugin to get information on a plugin and what arguments it can use.
Square brackets ([ ]) indicate optional parts of the command line.
i.MX Linux® Release Notes
(IMXLXRN) for i.MX-specific options, or at
7.3.1 Playback use cases
Playback use cases include the following:
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
imx8mp-evk-dsp-lpa.dtb is used for LPA audio playback.
• If tinycompresssink is with property enable-lpa=true, the system suspends or resumes during audio playback
to save power.
• If tinycompresssink is with property provide-clock=true, the pipeline gets media clock from DSP to avoid clock
error when the system enters suspending mode during audio playback.
For the platforms without VPU hardware, $video_decoder_plugin could be a software decoder plugin like avdec_h264.
7.3.1.4 Multichannel audio playback
For the multichannel audio playback settings to be used when PulseAudio is enabled, see PulseAudio input/output settings.
7.3.1.5 Other methods for playback
Use the $PLAYBIN plugin or the i.MX $GPLAY command line player for media file playback.
$GSTL $PLAYBIN uri=file:///mnt/sdcard/test.avi
$GPLAY /mnt/sdcard/test.avi
7.3.1.6 Video playback to multiple displays
Video playback to multiple displays can be supported by a video sink plugin. The video sink for multidisplay mode does not work
on i.MX 8 family SoCs.
This use case requires that the system boots in multiple-display mode (dual/triple/four, the number of displays supported
is determined by the SOC and the BSP). For how to configure the system to boot in this mode, see the
Guide
(IMXBSPPG).
i.MX Porting
7.3.1.6.1 Playing different videos on different displays
The command line to play two videos on different displays might look like this:
The overlaysink plugin provides support for compositing multiple videos together and rendering them to the same display. The
result might look like the following image.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
The commands below presents some suggestions on how to use the plugins accelerated by VPU hardware to encode some media
files (though they only work on a SoC with a VPU).
VPU video encoding only works on SoC with VPU encoder support.
capsfilter is the container's mime type. CAPS1 is the target video resolution, and the vpuenc_xxx can be vpuenc_mpeg4,
vpuenc_h263, vpuenc_h264, vpuenc_jpeg, vpuenc_hevc, and vpuenc_vp8.
• For some mux, such as matroskamux, add h264parse/h265parse before mux.
• For i.MX 6, h264parse is not required, because the VPU can output AVC and byte-stream formats. For
i.MX 8, h264parse/h265parse should be added before some mux, because the VPU supports only the
byte-stream output.
Video recording is done using the camera input, so this activity only applies to platforms with a camera. Different cameras need
to be set with different capture modes for special resolutions. See Chapter 14 "supporting cameras with CSI" and Chapter 15
"supporting cameras with MIPI-CSI" in the
VPU video encoding only works on SoC with VPU encoder support. imxv4l2src is only supported on i.MX 6 and i.MX 7. i. MX 8
supports opensource plugin v4l2src as camera source.
Use the $GSTINSPECT command to obtain more information about the codec property.
• Get the camera support format and resolution using gst-inspect-1.0 v4l2src.
• Set caps filter according to the camera's supported capabilities if the user needs other format or resolution.
• Ensure that the right caps filter has been set, which also needs to be supported by v4l2sink.
7.3.9 Recording the TV-in source
The TV-in source plugin gets video frames from the TV decoder. It is based on the V4l2 capture interface. A command line example
is as follows:
gst-launch-1.0 v4l2src ! autovideosink
NOTE
The TV decoder is ADV7180. It supports NTSC and PAL TV mode. The output video frame is interlaced, so the
sink plugin needs to enable deinterlace. The default value of v4l2sink deinterface is True.
7.3.10 Web camera
The following command line is an example of how to record and transfer web camera input.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide59 / 103
NXP Semiconductors
Multimedia
• PLAYBIN
$GSTL $PLAYBIN uri=$RTSP_URI
Two properties of rtspsrc that are useful for RTSP streaming are:
• Latency: This is the extra added latency of the pipeline, with the default value of 200 ms. If you need low-latency RTSP
streaming playback, set this property to a smaller value.
• Buffer-mode: This property is used to control the buffering algorithm in use. It includes four modes:
— None: Outgoing timestamps are calculated directly from the RTP timestamps, not good for real-time applications.
— Slave: Calculates the skew between the sender and receiver and produces smoothed adjusted outgoing timestamps,
good for low latency communications.
— Buffer: Buffer packets between low and high watermarks, good for streaming communication.
— Auto: Chooses the three modes above depending on the stream. This is the default setting.
To pause or resume the RTSP streaming playback, use a buffer-mode of slave or none for rtspsrc, as in buffer-mode=buffer.
After resuming, the timestamp is forced to start from 0, and this causes buffers to be dropped after resuming.
Playback does not exit automatically in GStreamer 1.x, if buffer-mode is set to buffer in the rtpsrc plugin.
7.3.14 RTP/UDP MPEGTS streaming
There are some points to keep in mind when doing RTP/UDP MPEGTS Streaming:
• The source file that the UDP/RTP server sends must be in TS format.
• Start the server one second earlier than the time client starts.
• Two properties of aiurdemux that are useful for UDP/RTP TS streaming are:
streaming-latency: This is the extra added latency of the pipeline, and the default value is 400 ms. This value is designed for
the situation when the client starts first. If the value is too small, the whole pipeline may not run due to lack of audio or video
buffers. In that situation, you should cancel the current command and restart the pipeline. If the value is too large, wait for a
long time to see the video after starting the server.
low_latency_tolerance: This value is a range that total latency can jitter around streaming-latency. This property is disabled
by default. When the user sets this value, the maximum latency is (streaming-latency + low_latency_tolerance).
The UDP MPEGTS streaming command line format looks like this:
The RTSP streaming server use case is based on the open source gst-rtsp-server package. It uses the i.MX aiurdemux plugin to
demultiplex the file to audio or video elementary streams and to send them out through RTP. Start the RTSP streaming server on
one board, and play it on another board with the RTSP streaming playback commands.
The gst-rtsp-server package is not installed by default in the Yocto Project release. Follow these steps to build and install it.
1. Enable the layer meta-openembedded/meta-multimedia:
Add the line BBLAYERS += "${BSPDIR}/sources/meta-openembedded/meta-multimedia" to the configuration file
$yocto_root/build/conf/bblayers.conf.
2. Include gst-rtsp-server into the image build:
Add the line IMAGE_INSTALL_append += "gstreamer1.0-rtsp-server" to the configuration file $yocto_root/build/
conf/local.conf.
3. Run bitbake for your image to build with gst-rstp-server.
4. You can find the test-uri binary in the folder:
• Client operations supported are Play, Stop, Pause, Resume, and Seek.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide61 / 103
NXP Semiconductors
Multimedia
7.3.16 Video conversion
There are three video conversion plugins, imxvideoconvert_ipu, imxvideoconvert_g2d, and imxvideoconvert_pxp. All of them can
be used to perform video color space conversion, resize, and rotate. imxvideoconvert_ipu can also be used to perform video
deinterlacing. They can be used to connect before ximagesink to enable the video rendering on X Windows or used in transcoding
to change video size, rotation, or deinterlacing.
Use gst-inspect-1.0 to get each convertor's capability and supported input and output formats. Note that imxvideoconvert_g2d can
only perform color space converting to RGB space.
It is possible to combine CSC, resize, rotate, and deinterlace at one time. Both of imxvideoconvert_ipu and imxvideoconvert_g2d
can be used at the same time in a pipeline. The following is an example:
imxcompositor_g2d uses corresponding hardware to accelerate video composition. It can be used to composite multiple videos
into one. The video position, size, and rotation can be specified while composition. Video color space conversion is also performed
automatically if input and output video are not same. Each video can be set to an alpha and z-order value to get alpha blending
and video blending sequence.
Note that imxcompositor_g2d can only output RGB color space format. Use gst-inspect-1.0 to get more detailed information,
including the supported input and output video format.
Use the pacmd command to set the default audio source according to the source number in the list shown above:
Multimedia
$ pacmd set-default-source $sink-number
$sink-number could be 0 or 1 in the example above. If record and playback at the same time is not needed, there is no need to
set the monitor mode.
The PulseAudio I/O path setting status can be checked with:
$ pactl stat
Multichannel output support settings
For those boards that need to output multiple channels, these are the steps needed to enable the multichannel output profile:
1. Use the pacmd command to list the available cards:
$ pacmd list-cards
The available sound cards and the profiles supported are listed.
2 card(s) available.
index: 0
name: <alsa_card.platform-sound-cs42888.34>
driver: <module-alsa-card.c>
owner module: 6
properties:
alsa.card = "0"
alsa.card_name = "cs42888-audio"
...
...
profiles:
input:analog-mono: Analog Mono Input (priority 1, available: unknown)
input:analog-stereo: Analog Stereo Input (priority 60, available: unknown)
...
...
active profile: <output:analog-stereo+input:analog-stereo>
...
...
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide64 / 103
NXP Semiconductors
2. Use the pacmd command to set the profile for particular features.
$ pacmd set-card-profile $CARD $PROFILE
$CARD is the card name listed by pacmd list-cards (for example, alsa_card.platform-sound-cs42888.34 in the
example above), and $PROFILE is the profile name. These are also listed by pamcd list-cards. (for example,
output:analog-surround-51 in the example above).
3. After setting the card profile, use $ pactl list sinks and $pacmd set-default-sink $sink-number to set the
default sink.
7.5 Installing gstreamer1.0-libav into rootfs
The following steps show how to install gstreamer1.0-libav into a rootfs image.
1. Add the following lines into the configuration file conf/local.conf.
IMAGE_INSTALL_append = " gstreamer1.0-libav"
LICENSE_FLAGS_WHITELIST = "commercial"
2. Build gstreamer1.0-libav.
$ bitbake gstreamer1.0-libav
Multimedia
3. Build the rootfs image.
$ bitbake
$ <image_name>
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide65 / 103
NXP Semiconductors
Chapter 8
Audio
8.1 DSP support
DSP support is provided on specific i.MX 8QuadXPlus, i.MX 8QuadMax, and i.MX 8M Plus SoC.
8.1.1 HiFi 4 DSP framework
Supporting HiFi 4 on a custom board is documented in the i.MX DSP Porting Guide delivered with the i.MX DSP Redistribution
package, which is available to customers who have a HiFi 4 license with Cadence.
8.1.2 Sound Open Firmware
Sound Open Firmware is an open source alternative to HiFi 4 DSP framework. For supporting the HiFi 4 on a custom board, see
the SOF project documentation https://thesofproject.github.io available in the public domain.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide66 / 103
NXP Semiconductors
Chapter 9
Graphics
There are a number of graphics tools, tests, and example programs that are built and installed in the Linux rootfs. There are
some variation on what is included based on the build and packages selected, the board, and the backend specified. This section
describes some of them.
The kernel module version of graphics used on the system can be found by running the following command on the board:
dmesg | grep Galcore
The user-side GPU drivers version of graphics can be displayed using the following command on the board:
grep VERSION /usr/lib/libGAL*
When reporting problems with graphics, this version number is needed.
9.1 imx-gpu-sdk
This graphics package contains source for several graphics examples for OpenGLES 2.0 and OpenGLES 3.0 APIs for X11,
Framebuffer, and XWayland graphical backends. These applications show that the graphics acceleration is working for different
APIs. The package includes samples, demo code, and documentation for working with the i.MX family of graphic cores. More
details about this SDK are in the
hardware acceleration.
i.MX Graphics User's Guide
. This SDK is only supported for hardware that has OpenGLES
9.2 G2D-imx-samples
The G2D Application Programming Interface (API) is designed to make it easy to use and understand the 2D BLT functions. It
allows the user to implement customized applications with simple interfaces. It is hardware and platform independent when using
2D graphics.
The G2D API supports the following features and more:
• Simple BLT operation from source to destination
• Alpha blend for source and destination with Porter-Duff rules
• High-performance memory copy from source to destination
• Up-scaling and down-scaling from source to destination
• 90/180/270 degree rotation from source to destination
• Horizontal and vertical flip from source to destination
• Enhanced visual quality with dither for pixel precision-loss
• High performance memory clear for destination
• Pixel-level cropping for source surface
• Global alpha blend for source only
• Asynchronous mode and synchronization
• Contiguous memory allocator
• VG engine
The G2D API document includes the detailed interface description and sample code for reference. The API is designed with
C-Style code and can be used in both C and C++ applications.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide67 / 103
NXP Semiconductors
Graphics
The G2D is supported on all i.MX. The hardware that supports G2D is described below. For more details, see the Frame Buffer
information in the
i.MX Release Notes
(IMXLXRN) to check which hardware is used for G2D.
• For i.MX 6 with GPU, the G2D uses the 2D GPU.
• For i.MX with PXP, the G2D uses the PXP with limited G2D features.
The following is the directory structure for the G2D test applications located under /opt.
• g2d_samples
— g2d_test
◦ g2d_overlay_test
◦ g2d_multiblit_test
9.3 viv_samples
The directory viv_samples is found under /opt. It contains binary samples for OpenGL ES 1.1/2.0 and OpenVG 1.1.
The following are the basic sanity tests, which help to make sure that the system is configured correctly.
• cl11: This contains unit tests and FFT samples for OpenCL 1.1 Embedded Profile. OpenCL is implemented on the i.MX
6Quad, i.MX 6QuadPlus, and i.MX 8 boards.
• es20: This contains tests for Open GLES 2.0.
— vv_launcher
◦ coverflow.sh
◦ vv_launcher
• tiger: A simple OpenVG application with a rotating tiger head. This is to demonstrate OpenVG.
• vdk: Contains sanity tests for OpenGL ES 1.1 and OpenGL ES 2.0.
The tiger and VDK tests show that hardware acceleration is being used. They will not run without it.
• cl11
— UnitTest
◦ clinfo
◦ loadstore
◦ math
◦ threadwalker
◦ test_vivante
▪ functions_and_kernels
▪ illegal_vector_sizes
▪ initializers
▪ multi_dimensional_arrays
▪ reserved_data_types
▪ structs_and_enums
▪ unions
▪ unsupported_extensions
— fft
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide68 / 103
NXP Semiconductors
Graphics
9.4 Qt 5
Qt 5 is built into the Linux image in the Yocto Project environment with the command bitbake imx-image-full. For more details
on Qt enablement, check out the README in the meta-imx repo and the
i.MX Yocto Project User's Guide
(IMXLXYOCTOUG).
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide69 / 103
NXP Semiconductors
Chapter 10
Security
The i.MX platforms define a series of security acceleration subsystems.
10.1 CAAM kernel driver
10.1.1 Introduction
The Linux kernel contains a Scatterlist Crypto API driver for the NXP CAAM security hardware block. It integrates seamlessly with
in-kernel crypto users, such as DM-Crypt, Keyctl, in a way that any disk encryption and key management suites will automatically
use the hardware to do the crypto acceleration. CAAM hardware is known in Linux kernel as 'caam', after its internal block name:
Cryptographic Accelerator and Assurance Module.
There are several HW interfaces ("backends") that can be used to communicate (for example, submit requests) with the engine,
their availability depends on the SoC:
• Register Interface (RI) - available on all SoCs (though access from kernel is restricted on DPAA2 SoCs).
Its main purpose is debugging (such as single-stepping through descriptor commands), though it is used also for
RNG initialization.
• Job Ring Interface (JRI) - legacy interface, available on all SoCs; on most SoCs there are 4 rings.
NOTE
There are cases when fewer rings are accessible or visible in the kernel, for example, when firmware like Trusted
Firmware-A (TF-A) reserves one of the rings.
On top of these backends, there are the "frontends" - drivers that sit between the Linux Crypto API and backend drivers. Their
main tasks aim to:
• Register supported crypto algorithms.
• Process crypto requests coming from users (through the Linux Crypto API) and translate them into the proper format
understood by the backend being used.
• Forward the CAAM engine responses from the backend being used to the users.
To use a specific implementation, it is possible to ask for it explicitly by using the specific (unique) "driver name" instead of
the generic "algorithm name". See official Linux Kernel Crypto API documentation (section Crypto API Cipher References And
Priority). Currently, the default priority is 3000 for JRI frontend.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide70 / 103
NXP Semiconductors
Security
Figure 3. Kernel CAAM Driver Architecture
10.1.2 Source files
The drivers source code is maintained in the Linux kernel source tree, under drivers/crypto/caam.The following is a
non-exhaustive list of files, mapping to CAAM (some files have been omitted because their existence is justified only by driver logic
or design).
Table 56. Source files
Source File(s)DescriptionModule name
ctrl.[c,h]Init (global settings, RNG, power management, etc.)caam
desc.hHW description (CCSR registers, etc.)N/A
desc_constr.hInline append - descriptor construction libraryN/A
caamkeyblob.[c,h]JRI frontend (black keys and blobs)N/A
caamkeygen.cIOCTL calls for key and blob generation/importN/A
10.1.3 Module loading
CAAM backend drivers can be compiled either built-in or as modules. Frontend drivers are linked to the backend driver. See
Section Source files for the list of module names and Section Kernel configuration for how kernel configuration looks like and a
mapping between menu entries and modules and/or functionalities enabled.
10.1.4 Kernel configuration
The designated driver should be configured in the kernel by default for the target platform. If unsure, check
CONFIG_CRYPTO_DEV_FSL_CAAM, which is located in the "Cryptographic API" -> "Hardware crypto devices" sub-menu
in the kernel configuration.
• Backends/interfaces: Freescale CAAM Job
Ring driver backend - JRI
• Frontends/crypto algorithms: symmetric
encryption, AEAD, "stitched" AEAD;
Register algorithm implementations with the
Crypto API - via JRI (caamalg driver)
• Register hash algorithm implementations
with Crypto API - hashing (only via JRI caamhash driver)
• Register public key cryptography
implementations with Crypto API asymmetric/public key (only via JRI caampkc driver)
• Register CAAM device for hwrng API - HW
RNG (only via JRI - caamrng driver)
• Register algorithms supporting tagged key
and generate black keys and encapsulate
them into black blobs
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide72 / 103
NXP Semiconductors
Security
Table 58. Device tree binding
PropertyTypeStatusDescription
compatibleStringRequiredfsl, sec-vX.Y (preferred) OR
fsl,secX.Y
Sample Device Tree crypto node
crypto@30000 {
compatible = "fsl,sec-v4.0";
fsl,sec-era = <2>;
#address-cells = <1>;
#size-cells = <1>;
reg = <0x300000 0x10000>;
ranges = <0 0x300000 0x10000>;
interrupt-parent = <&mpic>;
interrupts = <92 2>;
clocks = <&clks IMX6QDL_CLK_CAAM_MEM>,
<&clks IMX6QDL_CLK_CAAM_ACLK>,
<&clks IMX6QDL_CLK_CAAM_IPG>,
<&clks IMX6QDL_CLK_EIM_SLOW>;
clock-names = "mem", "aclk", "ipg", "emi_slow";
};
10.1.5 How to test the drivers
Crypto drivers could be validated in two modes: at boot time and at request. To enable crypto testing feature, the kernel needs to
be updated as follows.
Table 59. Kernel configuration
Kernel configurationDescription
--- Cryptographic API --->
[ ] Disable run-time self tests
[ ] Enable extra run-time crypto self tests
<M> Testing module
Section from boot log that specify where crypto test are made (If a boot test is passing with success, no information will be reported.
For algorithms with no tests available, a line in dmesg will be printed):
[ 4.647985] alg: No test for authenc(hmac(sha224),ecb(cipher_null)) (authenc-hmac-sha224-ecb-
cipher_null-caam)
[ 4.661181] alg: No test for authenc(hmac(sha256),ecb(cipher_null)) (authenc-hmac-sha256-ecb-
cipher_null-caam)
[ 4.671345] alg: No test for authenc(hmac(sha384),ecb(cipher_null)) (authenc-hmac-sha384-ecb-
cipher_null-caam)
[ 4.681486] alg: No test for authenc(hmac(sha512),ecb(cipher_null)) (authenc-hmac-sha512-ecb-
cipher_null-caam)
[ 4.691608] alg: No test for authenc(hmac(md5),cbc(aes)) (authenc-hmac-md5-cbc-aes-caam)
[ 4.699802] alg: No test for echainiv(authenc(hmac(md5),cbc(aes))) (echainiv-authenc-hmac-md5-cbc-
Deselect the feature that bypass crypto driver validation. By
default, Linux kernel is bypassing crypto driver validation.
Disable run-time self tests that normally take place at
algorithm registration.
Enable extra run-time self tests of registered crypto algorithms,
including randomized fuzz tests. This is intended for developer
use only, as these tests take much longer to run than the
normal self tests.
Enable testing module.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide73 / 103
NXP Semiconductors
Security
aes-caam)
[ 4.710445] alg: No test for echainiv(authenc(hmac(sha1),cbc(aes))) (echainiv-authenc-hmac-sha1-
cbc-aes-caam)
[ 4.720488] alg: No test for authenc(hmac(sha224),cbc(aes)) (authenc-hmac-sha224-cbc-aes-caam)
[ 4.734647] alg: No test for echainiv(authenc(hmac(sha224),cbc(aes))) (echainiv-authenc-hmac-
sha224-cbc-aes-caam)
[ 4.750504] alg: No test for echainiv(authenc(hmac(sha256),cbc(aes))) (echainiv-authenc-hmac-
sha256-cbc-aes-caam)
[ 4.762468] alg: No test for authenc(hmac(sha384),cbc(aes)) (authenc-hmac-sha384-cbc-aes-caam)
[ 4.771188] alg: No test for echainiv(authenc(hmac(sha384),cbc(aes))) (echainiv-authenc-hmac-
sha384-cbc-aes-caam)
[ 4.782380] alg: No test for echainiv(authenc(hmac(sha512),cbc(aes))) (echainiv-authenc-hmac-
sha512-cbc-aes-caam)
[ 4.792765] alg: No test for authenc(hmac(md5),cbc(des3_ede)) (authenc-hmac-md5-cbc-des3_ede-caam)
[ 4.801832] alg: No test for echainiv(authenc(hmac(md5),cbc(des3_ede))) (echainiv-authenc-hmac-md5-
cbc-des3_ede-caam)
[ 4.812814] alg: No test for echainiv(authenc(hmac(sha1),cbc(des3_ede))) (echainiv-authenc-hmac-
sha1-cbc-des3_ede-caam)
[ 4.823942] alg: No test for echainiv(authenc(hmac(sha224),cbc(des3_ede))) (echainiv-authenc-hmac-
sha224-cbc-des3_ede-caam)
[ 4.835465] alg: No test for echainiv(authenc(hmac(sha256),cbc(des3_ede))) (echainiv-authenc-hmac-
sha256-cbc-des3_ede-caam)
[ 4.846980] alg: No test for echainiv(authenc(hmac(sha384),cbc(des3_ede))) (echainiv-authenc-hmac-
sha384-cbc-des3_ede-caam)
[ 4.858497] alg: No test for echainiv(authenc(hmac(sha512),cbc(des3_ede))) (echainiv-authenc-hmac-
sha512-cbc-des3_ede-caam)
[ 4.869764] alg: No test for authenc(hmac(md5),cbc(des)) (authenc-hmac-md5-cbc-des-caam)
[ 4.877977] alg: No test for echainiv(authenc(hmac(md5),cbc(des))) (echainiv-authenc-hmac-md5-cbc-
des-caam)
[ 4.888078] alg: No test for echainiv(authenc(hmac(sha1),cbc(des))) (echainiv-authenc-hmac-sha1-
cbc-des-caam)
[ 4.898356] alg: No test for echainiv(authenc(hmac(sha224),cbc(des))) (echainiv-authenc-hmac-
sha224-cbc-des-caam)
[ 4.908994] alg: No test for echainiv(authenc(hmac(sha256),cbc(des))) (echainiv-authenc-hmac-
sha256-cbc-des-caam)
[ 4.919653] alg: No test for echainiv(authenc(hmac(sha384),cbc(des))) (echainiv-authenc-hmac-
sha384-cbc-des-caam)
[ 4.930292] alg: No test for echainiv(authenc(hmac(sha512),cbc(des))) (echainiv-authenc-hmac-
sha512-cbc-des-caam)
[ 4.940688] alg: No test for authenc(hmac(md5),rfc3686(ctr(aes))) (authenc-hmac-md5-rfc3686-ctr-
aes-caam)
[ 4.950372] alg: No test for seqiv(authenc(hmac(md5),rfc3686(ctr(aes)))) (seqiv-authenc-hmac-md5-
rfc3686-ctr-aes-caam)
[ 4.961281] alg: No test for seqiv(authenc(hmac(sha1),rfc3686(ctr(aes)))) (seqiv-authenc-hmac-sha1-
rfc3686-ctr-aes-caam)
[ 4.972281] alg: No test for authenc(hmac(sha224),rfc3686(ctr(aes))) (authenc-hmac-sha224-rfc3686-
ctr-aes-caam)
[ 4.982482] alg: No test for seqiv(authenc(hmac(sha224),rfc3686(ctr(aes)))) (seqiv-authenc-hmac-
sha224-rfc3686-ctr-aes-caam)
[ 4.993903] alg: No test for seqiv(authenc(hmac(sha256),rfc3686(ctr(aes)))) (seqiv-authenc-hmac-
sha256-rfc3686-ctr-aes-caam)
[ 5.005331] alg: No test for seqiv(authenc(hmac(sha384),rfc3686(ctr(aes)))) (seqiv-authenc-hmac-
sha384-rfc3686-ctr-aes-caam)
[ 5.016763] alg: No test for seqiv(authenc(hmac(sha512),rfc3686(ctr(aes)))) (seqiv-authenc-hmac-
sha512-rfc3686-ctr-aes-caam)
[ 5.028023] caam algorithms registered in /proc/crypto
[ 5.206167] caam 31400000.caam: caam pkc algorithms registered in /proc/crypto
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide74 / 103
NXP Semiconductors
10.2 Crypto algorithms support
• Algorithms supported in the Linux kernel scatterlist Crypto API
The Linux kernel contains various users of the Scatterlist Crypto API, including its IPsec implementation, sometimes referred
to as the NETKEY stack. The driver, after registering supported algorithms with the Crypto API, is therefore used to process
per packet symmetric crypto requests and forward them to the CAAM hardware. Since CAAM hardware processes requests
asynchronously, the driver registers asynchronous algorithm implementations with the crypto API: ahash, skcipher, and
aead with CRYPTO_ALG_ASYNC set in .cra_flags. Different combinations of hardware and driver software version support
different sets of algorithms, so searching for the driver name in /proc/crypto on the desired target system will ensure the
correct report of what algorithms are supported.
• Authenticated Encryption with Associated Data (AEAD) algorithms
These algorithms are used in applications where the data to be encrypted overlaps, or partially overlaps, the data to be
authenticated, as is the case with IPsec and TLS protocols. These algorithms are implemented in the driver such that the
hardware makes a single pass over the input data, and both encryption and authentication data are written out simultaneously.
The AEAD algorithms are mainly for use with IPsec ESP (however there is also support for TLS (1.x) record layer encryption
(KTLS Support)).
CAAM drivers currently supports offloading the following AEAD algorithms:
— "stitched" AEAD: all combinations of { NULL, CBC-AES, CBC-DES, CBC-3DES-EDE, RFC3686-CTR-AES } x HMAC-
{MD-5, SHA-1,-224,-256,-384,-512}
Security
— "true" AEAD: generic GCM-AES, GCM-AES used in IPsec: RFC4543-GCM-AES and RFC4106-GCM-AES
• Encryption algorithms
The CAAM driver currently supports offloading the following encryption algorithms.
• Authentication algorithms
The CAAM driver's ahash support includes keyed (hmac) and unkeyed hashing algorithms.
• Asymmetric (public key) algorithms
Currently, CAAM driver supports RSA-Encrypt and RSA-Decrypt togheter with pkcs1pad (rsa-caam,sha256) driver.
CAAM Job Ring backend driver (caam_jr) implements and uses the job ring interface (JRI) for submitting crypto API service
requests from the frontend drivers (caamalg, caamhash, caampkc, caamrng, caamkeyblob) to CAAM engine.
CAAM drivers have a few options, most notably hardware job ring size and interrupt coalescing. They can be used to fine-tune
performance for a particular use case.
The option Freescale CAAM Job Ring driver backend enables the Job Ring backend (caam_jr). The sub-option Job Ring Size
allows the user to select the size of the hardware job rings. If requests arrive at the driver enqueue entry point in a bursty
nature, the bursts' maximum length can be approximated. The user can set the greatest burst length to save performance and
memory consumption.
The sub-option Job Ring interrupt coalescing allows the user to select the use of the hardware's interrupt coalescing feature. Note
that the driver already performs IRQ coalescing in software, and zero-loss benchmarks have in fact produced better results with
this option turned off. If selected, two additional options become effective:
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide76 / 103
NXP Semiconductors
Security
•
Job Ring interrupt coalescing count threshold
(CRYPTO_DEV_FSL_CAAM_INTC_THLD) Device Drivers. Selects the
value of the descriptor completion threshold, in the range 1-256. A selection of 1 effectively defeats the coalescing feature,
and any selection equal or greater than the selected ring size will force timeouts for each interrupt.
•
Job Ring interrupt coalescing timer threshold
(CRYPTO_DEV_FSL_CAAM_INTC_TIME_THLD) Selects the value of the
completion timeout threshold in multiples of 64 CAAM interface clocks, to which, if no new descriptor completions occur
within this window (and at least one completed job is pending), then an interrupt will occur. This is selectable in the range
1-65535.
The options to register to Crypto API, hwrng API respectively, allow the frontend drivers to register their algorithm capabilities with
the corresponding APIs. They should be deselected only when the purpose is to perform Crypto API requests in software (on the
GPPs) instead of offloading them on CAAM engine.
caamhash
frontend (hash algorithms) may be individually turned off, since the nature of the application may be such that it prefers
software (core) crypto latency due to many small-sized requests.
caampkc
caamrng
frontend (public key / asymmetric algorithms) can be turned off too, if needed.
frontend (Random Number Generation) may be turned off in case there is an alternate source of entropy available to
the kernel.
caamkeyblob
frontend (algorithms supporting tagged key) can be turned off if tagged keys or blobs are not used.
10.3.1 Verifying driver operation and correctness
Other than noting the performance advantages due to the crypto offload, one can also ensure the hardware is doing the crypto
by looking for driver messages in dmesg. The driver emits console messages at initialization time:
[ 1.830397] caam 30900000.crypto: device ID = 0x0a16040100000000 (Era 9)
If the messages are not present in the logs, either the driver is not configured in the kernel, or no CAAM compatible device tree
node is present in the device tree.
10.3.2 Incrementing IRQs in /proc/interrupts
Given a time period when crypto requests are being made, the CAAM hardware will fire completion notification interrupts on the
corresponding Job Ring:
root@imx8qxpmek:~# cat /proc/interrupts | grep jr
418: 1059 0 0 0 GICv3 485 Level 31430000.jr2
419: 21 0 0 0 GICv3 486 Level 31440000.jr3
root@imx8qxpmek:~#
If the number of interrupts fired increment, then the hardware is being used to do the crypto. If the numbers do not increment, then
check the algorithm being exercised is supported by the driver. If the algorithm is supported, there is a possibility that the driver
is in polling mode (NAPI mechanism) and the hardware statistics in debugfs (inbound/outbound bytes encrypted/protected - see
below) should be monitored.
10.3.3 Verifying the 'self test' fields say 'passed' in /proc/crypto
An entry such as the one below means the driver has successfully registered support for the algorithm with the kernel crypto API:
name : cbc(des)
driver : cbc-des-caam
module : kernel
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide77 / 103
NXP Semiconductors
Security
priority : 3000
refcnt : 1
selftest : passed
internal : no
type : givcipher
async : yes
blocksize : 8
min keysize : 8
max keysize : 8
ivsize : 8
geniv : <built-in>
Note that although a test vector may not exist for a particular algorithm supported by the driver, the kernel will emit messages
saying which algorithms weren't tested, and mark them as 'passed' anyway:
[ 4.647985] alg: No test for authenc(hmac(sha224),ecb(cipher_null)) (authenc-hmac-sha224-ecb-
cipher_null-caam)
[ 4.661181] alg: No test for authenc(hmac(sha256),ecb(cipher_null)) (authenc-hmac-sha256-ecb-
cipher_null-caam)
[ 4.671345] alg: No test for authenc(hmac(sha384),ecb(cipher_null)) (authenc-hmac-sha384-ecb-
cipher_null-caam)
[ 4.681486] alg: No test for authenc(hmac(sha512),ecb(cipher_null)) (authenc-hmac-sha512-ecb-
cipher_null-caam)
10.4 OpenSSL offload
The Secure Socket Layer (SSL) protocol is the most widely deployed application protocol to protect data during transmission by
encrypting the data using ommonly used cipher algorithms such as AES, DES and 3DES. Apart from encryption, it also provides
message authentication services using hash/digest algorithms such as SHA1 and MD5. SSL is widely used in application web
servers (HTTP) and other applications such as SMTP POP3, IMAP, and Proxy servers, where protection of data in transit is
essential for data integrity. There are various versions of SSL protocol such as TLSv1.0, TLSv1.1, TLSv1.2, TLSv1.3, and DTLS
(Datagram TLS). This document describes NXP SSL acceleration solution on i.MX platforms using OpenSSL:
• OpenSSL software architecture
• Building OpenSSL with hardware offload support
• Examples of OpenSSL Offloading
10.4.1 OpenSSL software architecture
The SSL protocol is implemented as a library in OpenSSL - the most popular library distribution in Linux and BSD systems. The
OpenSSL library has several sub-components such as:
The following figure presents the general interconnect architecture for OpenSSL. Each relevant layer is represented with a clear
separation between Linux User Space and Linux Kernel Space.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide78 / 103
NXP Semiconductors
Security
Figure 4. OpenSSL Software stack architecture
10.4.2 OpenSSL's ENGINE interface
OpenSSL Crypto library provides Symmetric and Asymmetric (PKI) cipher support that is used in a wide variety of applications
such as OpenSSH, OpenVPN, PGP, IKE, and XML-SEC. The OpenSSL Crypto library provides software support for:
• Cipher algorithms
• Digest algorithms
• Random number generation
• Public Key Infrastructure
Apart from the software support, the OpenSSL can offload these functions to hardware accelerators through the ENGINE
interface. The ENGINE interface provides callback hooks that integrate hardware accelerators with the crypto library. The callback
hooks provide the glue logic to interface with the hardware accelerators. Generic offloading of cipher and digests algorithms
through Linux kernel is possible with cryptodev engine.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide79 / 103
NXP Semiconductors
Security
10.4.3 NXP solution for OpenSSL hardware offloading
The following layers can be observed in NXP's solution for OpenSSL hardware offloading:
• OpenSSL (user space) - implements the SSL protocol
• cryptodev-engine (user space) - implements the OpenSSL ENGINE interface; talks to cryptodev-linux (/dev/crypto) via
ioctls, offloading cryptographic operations in kernel
• cryptodev-linux (kernel space) - Linux module that translates ioctl requests from cryptodev-engine into calls to Linux
Crypto API
• Linux Crypto API (kernel space) - Linux kernel crypto abstraction layer
• CAAM driver (kernel space) - Linux device driver for the CAAM crypto engine
The following are offloaded in hardware in current BSP:
Typically, the imx-image-full includes the OpenSSL and cryptodev modules, but for other Yocto targets, users need to update the
conf file from the build directory. Update conf/local.conf by adding the following line:
10.4.5.1 Running OpenSSL benchmarking tests for symmetric ciphering and digest
In the speed test file, a series of performance tests are made to check the performance of the symmetric and digest operations.
The following is described in the OpenSSL test execution:
Disk encryption is a technology that protects information by converting it into unreadable code that cannot be deciphered easily
by unauthorized people. Disk encryption uses disk encryption software or hardware to encrypt every bit of data that goes on a disk
or disk volume. It is used to prevent unauthorized access to data storage. On i.MX Applications Processors, the disk encryption
scenarios could be implemented in various ways with different methods of key protection.
This section provides steps to run a transparent storage encryption at block level using DM-Crypt. The figure below presents the
software stack that implements disk encryption.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide81 / 103
NXP Semiconductors
Security
Figure 5. Disk Encryption use case
10.5.1 Enabling disk encryption support in kernel
By default, the kernel configuration file enables the Device Mapper configuration and Crypt Target support as modules. Therefore,
to enable disk encryption scenario, after the board is booted up, insert the following modules:
• keyutils: provides keyctl, which is required to manage Linux Key retention service.
• lvm2: provides dmsetup utility and libraries to manage device-mapper.
• e2fsprogs-mke2fs: contains necessary tools to create filesystems.
• util-linux: provides blockdev utility needed to read number of sectors from a volume.
10.5.3 DM-Crypt using CAAM's tagged key
In Linux Unified Key Setup (LUKS) mode, to generate the disk encryption key (master key), the user supplies a passphrase, which
is combined with a salt, and then a hash function is applied for a supplied number of rounds. When the user wants to mount an
encrypted volume, the passphrase should be supplied. An alternative could be providing a key file stored in an external drive
containing necessary decryption information. Those approaches are not convenient with embedded devices usage.
The aim of using DM-Crypt with CAAM’s tagged key is to suppress the mechanism of encrypting the master volume key with a
key derived from a user-supplied passphrase. DM-Crypt can also take advantages of tagged key to protect storage volumes from
offline decryption. In addition, the volume could only be opened by the devices that have the same OTPMK burned in the fuses.
For more details, see the Security Reference Manual for specific SoC.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide83 / 103
NXP Semiconductors
Security
The tagged key feature is based on CAAM’s black key mechanism. Black key protects user keys against bus snooping while
the keys are being written to or read from memory external to the SoC. CAAM supports two different black key encapsulation
schemes, which are AES-ECB and AES-CCM.
Regarding AES-ECB encryption, the data is a multiple of 16 bytes long and is intended for quick decryption.
The AES-CCM mode is not as fast as AES-ECB mode, but includes a “MAC tag” (integrity check value) that ensures the integrity
of the encapsulated key. A CCM-encrypted black key is always at least 12 bytes longer than the encapsulated key (nonce value
+ MAC tag).
Black keys are session keys; therefore, they are not power-cycles safe. CAAM's blob mechanism provides a method for protecting
user-defined data across system power cycles. It provides both confidentiality and integrity protection. The data to be protected
is encrypted so that it can be safely placed into non-volatile storage before the SoC is powered down.
The following diagram illustrates the changes that have been added to support full disk encryption using tagged key. The CAAM
driver registers new Cryptographic transformations in the kernel to use ECB and CBC blacken keys, tk(ecb(aes)) and tk(cbc(aes)).
The tk prefix refers to Tagged Key.
A Tagged Key is a black key that contains metadata indicating what it is and how to handle it.
Linux OS provides an in-kernel key management and retention facility called Keyrings. Keyring also enables interfaces to allow
accessing keys and performing operations such as add, update, and delete from user-space.
The kernel provides several basic types of keys including keyring, user, and logon.
The CAAM driver has associated a user-space application used to generate a tagged key and encapsulated it into a black blob.
This is also used to import a tagged key from a black blob.
$ ./caam-keygen
CAAM keygen usage: caam-keygen [options]
Options:
create <key_name> <key_enc> <key_mode> <key_val>
<key_name> the name of the file that will contain the black key.
A file with the same name, but with .bb extension, will contain the black blob.
<key_enc> can be ecb or ccm
<key_mode> can be -s or -t.
-s generate a black key from random with the size given in the next argument
-t generate a black key from a plaintext given in the next argument
<key_val> the size or the plaintext based on the previous argument (<key_mode>)
import <blob_name> <key_name>
<blob_name> the absolute path of the file that contains the blob
<key_name> the name of the file that will contain the black key.
By default, the keys and blobs are created in KEYBLOB_LOCATION, which is /data/caam/.
Later, CAAM Tagged Key is added into Linux Key Retention service and managed by user-space application such as keyctl. Black
blobs can be stored on any non-volatile storage.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide84 / 103
NXP Semiconductors
Security
Figure 6. DM-Crypt using CAAM Tagged key - overall architecture
Dmsetup (part of the libdevmapper package) is a powerful tool for performing very low-level configuration and is used to manage
encrypted volumes.
10.5.4 Usage
The following are the steps to perform a full disk encryption on i.MX devices.
1. After booting the device, make sure that cryptographic transformations using Tagged Key are registered.
5. Use dmsetup to create a new device-mapper device named encrypted for example, and specify the mapping table. The
table can be provided on stdin or as argument.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide86 / 103
NXP Semiconductors
The following is a breakdown of the mapping table:
• start means encrypting begins with sector 0.
• size is the size of the volume in sectors.
• blockdev gets the number of sectors of the device.
• target is crypt.
• cipher is set in Kernel Crypto API format to use Tagged Key. cipher set to capi:tk(cbc(aes))-plain and key set
to :36:logon:logkey: leads to use of the logon key with CAAM Tagged Key transformation.
• IV is the Initialization Vector defined to plain, initial vector, which is the 32-bit little-endian version of the sector number,
padded with zeros if necessary.
• key type is the Keyring key service type, set to Logon Key. 36 is the key size in bytes.
• key name is the key description to identify the key to load.
• IV offset is the value to add to sector number to compute the IV value.
• device is the path to device to be used as backend; it contains the encrypted data.
• offset represents encrypted data begins at sector 0 of the device.
• optional parameters represent the number of optional parameters.
• sector_size specifies the encryption sector size.
Security
For more detailed options and descriptions, refer to https://gitlab.com/cryptsetup/cryptsetup/-/wikis/DMCrypt.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide89 / 103
NXP Semiconductors
Chapter 11
Connectivity
This section describes the connectivity for Bluetooth wireless technology and Wi-Fi, as well as for USB type-C.
11.1 Connectivity for Bluetooth wireless technology and Wi-Fi
Bluetooth and Wi-Fi are supported on i.MX through on-board chip solutions and external hardware. The following table lists
the various on-board chips and external solutions. When i.MX uses the Azurewave CM358 uSD card, it requires a hardware
modification on boards.
Table 60. On-board chips and external solutions for Bluetooth and Wi-Fi support
SoCOn-board chipPCIeuSD card or uSD-to-M.2
baseboard
8QuadXPlus/8DXL-NXP 88W8997 (tested with
AW-CM276MA)
8QuadMax-NXP 88W8997 (tested with
AW-CM276MA)
8MQuad-NXP 88W8997 (tested with
AW-CM276MA)
8M Nano
8M Mini
7ULP--NXP 88W8987 (tested with
7Dual--NXP 88W8987 (tested with
6QuadPlus/Quad/Dual/Solo--NXP 88W8987 (tested with
6SLL/6UltraLite/6ULL/6ULZ--NXP 88W8987 (tested with
NXP 88W8987 (tested with
AzureWave AW-CM358SM)
NXP 88W8987 (tested with
AzureWave AW-CM358SM)
--
--
-
-
-
AW-CM358-uSD)
AW-CM358-uSD)
AW-CM358-uSD)
AW-CM358-uSD)
8M Plus-NXP 88W8997 (tested with
AW-CM276MA)
The wireless driver supports wpa_supplicant, which is a WEP/WPA/WPA2/WPA3 encryption authenticated tool.
• Wi-Fi driver: supports NXP 88W8987-based modules with SDIO interface, NXP 88W8997-based modules with
PCIe interface.
• Firmware
The NXP release package already includes all NXP, Wi-Fi/Bluetooth firmware. It requires to accept NXP license.
To run Wi-Fi, execute the following commands first and follow common commands below:
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide90 / 103
-
NXP Semiconductors
• For NXP 88w8997:
modprobe moal mod_para=nxp/wifi_mod_para.conf
• For NXP 88w8987:
modprobe moal mod_para=nxp/wifi_mod_para.conf
(Follow common step below)
• For Common Step, execute these commands using connman
The i.MX 6 boards require board rework to support the bluetooth WiFI enablement as well as running with the btwifi device tress.
The following is a list of the hardware modifications required and possibly conflicts caused by these modifications.
• i.MX 6QuadPlus/Quad/Dual/DualLite/Solo: See https://community.nxp.com/docs/DOC-94235. This change HAS a pin
conflict with: EPDC/SPI-NOR/GPIO-LED.
• i.MX 6SoloX: Install R328, and disconnect R327. Connect with SD2 slot and BLUETOOTH CABLE CONNECTOR J19. It
has no Pin conflict with other modules.
• i.MX 6SLL: Install R127, and double check to ensure R126 and R128 are installed. Connect with SD3 slot and
BLUETOOTH CABLE CONNECTOR J4. It has no Pin conflict with other modules.
• i.MX 6UL/ULL/ULZ: Install R1701. It has no Pin conflict with other modules.
11.2 Connectivity for USB type-C
The following describes the connectivity for USB type-C and power delivery connection on the i.MX 8QuadXPlus MEK board.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide91 / 103
NXP Semiconductors
Connectivity
• The Linux release includes USB type-C and PD stack, which is enabled by default. The specific power parameters are passed
in by DTS. The following fsl-imx8qxp-mek is an example:
typec_ptn5110: typec@50 {
compatible = "usb,tcpci";
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_typec>;
reg = <0x50>;
interrupt-parent = <&gpio1>;
interrupts = <3 IRQ_TYPE_LEVEL_LOW>;
ss-sel-gpios = <&gpio5 9 GPIO_ACTIVE_LOW>;
reset-gpios = <&pca9557_a 7 GPIO_ACTIVE_HIGH>;
src-pdos = <0x380190c8>;
snk-pdos = <0x380190c8 0x3802d0c8>;
max-snk-mv = <9000>;
max-snk-ma = <1000>;
op-snk-mw = <9000>;
port-type = "drp";
sink-disable;
default-role = "source";
status = "okay";
};
For power capability related configuration, users need to check the PD specification to see how to composite the PDO value.
To make it support power source role for more voltages, specify the source PDO. The i.MX 8QuadXPlus board can support
5 V and 12 V power supply.
• The Linux BSP of the Alpha and Beta releases on the i.MX 8QuadXPlus MEK platform only supports power source role for 5 V.
• Users can use /sys/kernel/debug/tcpm/2-0050 to check the power delivery state, which is for debugging purpose information.
• Booting only by type-C port power supply is not supported on the Alpha release.
11.3 NXP Bluetooth/Wi-Fi information
The NXP Bluetooth/Wi-Fi information is as follows:
• SoC version: 88W8987
• Wi-Fi firmware version: 16.92.10.p208
— 16: Major revision
— 92: Feature pack
— 10: Re ease version
— p207: Patch number
• Bluetooth-Bluetooth LE firmware version: 16.92.10.p208
— 16: Major revision
— 92: Feature pack
— 10: Release version
— p207: Patch number
• Driver version: MXM5X16210-MGPL
— 5X: Linux 5.x
— 16186: Release version
— p7: Patch number
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide92 / 103
NXP Semiconductors
— MGPL: General
Tested using perf version 2.0.5.
11.4 Certification
11.4.1 WFA certification
The following table lists the WFA certification.
Table 61. WFA certification
STACertification
STA802.11n
STA802.11ac
STAWPS2.0
STAPMF
STAWMM-PS
STAWPA3
Connectivity
For details, see Wi-Fi Alliance Derivative Certification (AN12976).
• DUT: 88W8987-Azurewave (Model: AW-CM358MA) with i.MX 8M Quad platform
• Client: Apple Macbook Air
• Channel: 6 | 36
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide93 / 103
NXP Semiconductors
Chapter 12
Virtualization
12.1 Xen overview
Xen project is a hypevisor using a microkernel design, providing services that allow multiple operating systems to run on the same
hardware with hardware capability. It is widely used on Cloud and is also used in embedded systems.
12.2 Basic architecture
The following figure shows the basic architecture of Xen.
Figure 7. Basic architecture of Xen
12.3 Xen xl
For detailed Xen xl usage, see https://xenbits.xen.org/docs/unstable/man/xl.cfg.5.html.
12.4 How to boot multiple operating systems on i.MX 8QuadMax EVK
To boot two Yocto operating systems, partition the SD card (no less than 16 GB) with three partitions. The first partition is FAT,
used to hold Xen, image, and imx8qm-mek-dom0.dtb. The second and third partitions are used to hold Yocto rootfs.
In the U-Boot stage:
• To boot from the SD card, run xenmmcboot.
• To boot from the network, run xennetboot.
After the first Linux OS is boot up, perform the following steps:
Jailhouse is a static partitioning hypervisor based on Linux OS. It can run bare-metal applications or other operating systems
besides Linux. It does not support virtual CPU schedule and only virtualizes those resources in software. This is essential for a
platform and cannot be partitioned in hardware. It configures CPUs and devices into different domains, called "cells".
When Jailhouse is activated, it runs bare-metal, which means, it takes full control over the hardware and needs no external
support. However, in contrast to other bare-metal hypervisors, it is loaded and configured by a normal Linux system. Its
management interface is based on the Linux infrastructure. Therefore, users need to boot Linux OS first, then enable Jailhouse,
and finally split off parts of the system's resources and assign them to additional cells. The following figure shows the workflow
from board power-on to multi-OS bootup. Jailhouse relies on the Linux OS (another boot loader) to do Jailhouse kernel loading
and initialization.
Figure 8. Workflow from board power-on to multi-OS bootup
To use the Jailhouse, perform the following steps:
1. Execute run jh_mmcboot or run jh_netboot in U-Boot stage.
— The second inmate Linux OS uses the first Linux UART as early console output. Its normal console uses its own
UART as console output.
— When the second inmate Linux boots up, execute ifconfig eth0 192.168.1.3 in the second inmate Linux OS.
And execute ifconfig eth[x] 192.168.1.4. For i.MX 8M Plus, [x] is 2. For the other platforms, [x] is 1. Then
you could use the network ping command to ping each other.
• IVSHMEM
a. Jailhouse cell creates /usr/share/jailhouse/cells/[soc]-ivshmem-demo.cell.
Replace [soc] with imx8mm/imx8mq/imx8mn/imx8mp/imx8qxp/imx8qm for the boards that you are using.
b. Jailhouse cell loads 1 /usr/share/jailhouse/inmates/ivshmem-demo.bin.
c. Jailhouse cell starts 1.
d. ivshmem-demo /dev/uio[x].
[x] normally the last one under your /dev/ directory.
• GIC
a. Jailhouse cell creates /usr/share/jailhouse/cells/[soc]-gic-demo.cell.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide99 / 103
NXP Semiconductors
Replace [soc] with imx8mm/imx8mq/imx8mn/imx8mp/imx8 for the boards that you are using.
b. Jailhouse cell loads 1 /usr/share/jailhouse/inmates/gic-demo.bin.
c. Jailhouse cell starts 1.
6. How to set up network for the second Linux OS using network.
a. After the first Linux OS boots up:
• sysctl -w net.ipv4.ip_forward=1
• sysctl -p /etc/sysctl.conf
b. After the root cell is enabled:
• iptables -A FORWARD -i eth1 -j ACCEPT
• iptables -A FORWARD -o eth1 -j ACCEPT
• iptables -t nat -A POSTROUTING -o eth0 -j MASQUERADE
c. After the second Linux OS boots up:
• ifconfig eth1 or 2 192.168.1.4 for the first Linux OS
• ifconfig eth0 192.168.1.5 for the second Linux OS
• ip route add default via 192.168.1.4 dev eth0 for the second Linux OS
Virtualization
ip route add 10.193.108.0/24 via 192.168.1.4 dev eth0 for the second Linux OS
• Now you could mount -t nfs 10.193.108.xx:/home/xxx/nfs /mnt.
i.MX Linux® User's Guide, Rev. LF5.10.9_1.0.0, 31 March 2021
User's Guide100 / 103
Loading...
+ hidden pages
You need points to download manuals.
1 point = 1 manual.
You can buy points or you can get point for every manual you upload.