UG10163

i.MX Linux User's Guide

Rev. LF6.6.36_2.1.0 — 30 September 2024 User guide

Document information
Information Content
Keywords i.MX, Linux, LF6.6.36_2.1.0
Abstract 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.
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 (RN00210).

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
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 platform
• i.MX 6DualLite SABRE-SD platform
• i.MX 6SoloX SABRE-SD platform
• i.MX 7Dual SABRE-SD 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 platform
• i.MX 8QuadXPlus MEK platform
• i.MX 8DualXLite EVK platform
• i.MX 8M Quad EVK platform
• i.MX 8M Mini EVK platform
• i.MX 8M Nano EVK platform
• i.MX 8M Plus EVK platform
• i.MX 8DualX MEK platform
• i.MX 8DualXLite Orangebox platform
• i.MX 8ULP EVK platform
• i.MX 91 EVK platform
• i.MX 91 QSB platform
• i.MX 93 EVK platform
• i.MX 93 QSB platform

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

2 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• i.MX 95 EVK platform

• i.MX 95 Verdin platform
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 platforms.
• SoloX or SX refers to the i.MX 6SoloX SABRE-SD platforms.
• 6UL refers to the i.MX 6UltraLite platform.
• 6ULL refers to the i.MX 6ULL platform.
• 6ULZ refers to the i.MX 6ULZ platform.
• 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.
• 8MP refers to the 8M Plus platform.
• 8DXL refers to the 8DualXLite platform.
• 8DX refers to the 8DualX platform.
• 8ULP refers to the i.MX 8Ultra Low Power platform.
• i.MX 91 refers to the i.MX 91 EVK platform.
• i.MX 93 refers to the i.MX 93 EVK 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 6 Family: 6QuadPlus, 6Quad, 6DualLite, 6SoloX, 6SLL, 6UltraLite, 6ULL, 6ULZ
• i.MX 7 Family: 7Dual, 7ULP
• i.MX 8 Family: 8QuadMax, 8QuadPlus, 8ULP
• i.MX 8M Family: 8M Plus, 8M Quad, 8M Mini, 8M Nano
• i.MX 8X Family: 8QuadXPlus, 8DXL, 8DXL OrangeBox, 8DualX
• i.MX 9 Family: i.MX 91, i.MX 93, i.MX 95
This release includes the following references and additional information.
• i.MX Linux Release Notes (RN00210) - Provides the release information.
• i.MX Linux User's Guide (UG10163) - Provides the information on installing U-Boot and Linux OS and using
i.MX-specific features.
• i.MX Yocto Project User's Guide (UG10164) - Describes the board support package for NXP development
systems using Yocto Project to set up host, install tool chain, and build source code to create images.
• i.MX Porting Guide (UG10165) - Provides the instructions on porting the BSP to a new board.
• i.MX Machine Learning User's Guide (UG10166) - Provides the machine learning information.
• i.MX DSP User's Guide (UG10167) - Provides the information on the DSP for i.MX 8.
• i.MX 8M Plus Camera and Display Guide (UG10168) - Provides the information on the ISP Independent
Sensor Interface API for the i.MX 8M Plus.
• i.MX Digital Cockpit Hardware Partitioning Enablement for i.MX 8QuadMax (UG10169) - Provides the i.MX
Digital Cockpit hardware solution for i.MX 8QuadMax.
• i.MX Graphics User's Guide (UG10159) - Describes the graphics features.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

3 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• Harpoon User's Guide (HRPNUG_3.1) - Presents the Harpoon release for i.MX 8M device family.
• i.MX Linux Reference Manual (RM00293) - Provides the information on Linux drivers for i.MX.
• i.MX VPU Application Programming Interface Linux Reference Manual (RM00294) - Provides the reference
information on the VPU API on i.MX 6 VPU.
• EdgeLock Enclave Hardware Security Module API (RM00284) - This document is a software reference
description of the API provided by the i.MX 8ULP, i.MX 93, and i.MX 95 Hardware Security Module (HSM)
solutions for the EdgeLock Enclave (ELE) Platform.
The quick start guides contain basic information on the board and setting it up. They are on the NXP website.
• SABRE Platform Quick Start Guide (IMX6QSDPQSG)
• i.MX 6UltraLite EVK Quick Start Guide (IMX6ULTRALITEQSG)
• i.MX 6ULL EVK Quick Start Guide (IMX6ULLQSG)
• i.MX 7Dual SABRE-SD Quick Start Guide (SABRESDBIMX7DUALQSG)
• i.MX 8M Quad Evaluation Kit Quick Start Guide (IMX8MQUADEVKQSG)
• i.MX 8M Mini Evaluation Kit Quick Start Guide (8MMINIEVKQSG)
• i.MX 8M Nano Evaluation Kit Quick Start Guide (8MNANOEVKQSG)
• i.MX 8QuadXPlus Multisensory Enablement Kit Quick Start Guide (IMX8QUADXPLUSQSG)
• i.MX 8QuadMax Multisensory Enablement Kit Quick Start Guide (IMX8QUADMAXQSG)
• i.MX 8M Plus Evaluation Kit Quick Start Guide (IMX8MPLUSQSG)
• i.MX 8ULP EVK Quick Start Guide (IMX8ULPQSG)
• i.MX 8ULP EVK9 Quick Start Guide (IMX8ULPEVK9QSG)
• i.MX 93 EVK Quick Start Guide (IMX93EVKQSG)
• i.MX 93 9x9 QSB Quick Start Guide (93QSBQSG)
Documentation is available online at nxp.com.
• i.MX 6 information is at nxp.com/iMX6series
• 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 91 information is at nxp.com/imx91.
• i.MX 93 information is at nxp.com/imx93
• i.MX 95 information is at nxp.com/imx95
The latest DDR configuration and test tools are available online at nxp.com and at NXP Community:
• i.MX 6/7:
i.MX 6/7 Series DDR Tool Release
• i.MX 8:
i.MX 8M Family DDR Tool Release
i.MX 8/8X Family DDR Tools Release
• i.MX 9 series:
Config Tools for i.MX Applications Processors

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

4 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 i.MX Yocto Project User's Guide (UG10164).
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.

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 converter 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.

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,

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

5 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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, and i.MX 8DXL, multiple 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 i.MX 8QuadMax/i.MX 8QuadXPlus B0 and i.MX 8DXL A1.
• (Optional) 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, multiple 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
On i.MX 8ULP, four elements are needed:
• imx-boot (built by imx-mkimage), which includes SPL, U-Boot, Arm Trusted Firmware, OP-TEE, uPower
Firmware, Sentinel Firmware, and Arm Cortex-M33 image
• Linux kernel image
• A device tree file (.dtb) for the board being used
• A root file system (rootfs) for the particular Linux image
On i.MX 93, multiple elements are needed:
• imx-boot (built by imx-mkimage), which includes SPL, U-Boot, Arm Trusted Firmware, OP-TEE, Sentinel
Firmware, DDR PHY Firmware
• Linux kernel image
• (Optional) Arm Cortex-M33 image
• A device tree file (.dtb) for the board being used
• A root file system (rootfs) for the particular Linux image
On i.MX 95, multiple elements are needed:
• imx-boot (built by imx-mkimage), which includes SPL, U-Boot, Arm Trusted Firmware, OP-TEE, ELE+V2X
Firmware, DDR PHY Firmware, System Manager, DDR OEI, TCM OEI (optional, but mandatory with the
Cortex-M7 image), Arm Cortex-M7 image (optional)
• Linux kernel image

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

6 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• 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 and i.MX 9 require 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.
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 (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 6.6.36 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. Filenames 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.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

7 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

On i.MX 8, i.MX 8M, i.MX 8ULP, i.MX 93, and i.MX 91, 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 filename 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|wic]. 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 1.5.125 or higher from https://github.com/NXPmicro/mfgtools/releases.

4.2.2 Using UUU

To use the UUU for i.MX 6, i.MX 7, i.MX 8, and i.MX 9, follow the instructions below:
1. Connect a USB cable from a computer to the USB OTG/TYPE C (or Micro-B, depending on board) port on
the board for download link.
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 "Section 3" in this document.
4. Set the boot pin to serial download mode mode. See Section "Section 4.5.11" in this document.
To use the UUU for i.MX 8ULP EVK, follow the instructions below:
• To burn single-boot image and rootfs to eMMC, run the following command:
uuu -b emmc_all imx-boot-imx8ulpevk-sd.bin-flash_singleboot_m33
<rootfs.wic.zst>
• To burn single-boot image to FlexSPI2 NOR flash, run the following command:
uuu -b qspi imx-boot-imx8ulpevk-fspi.bin-flash_singleboot_m33_flexspi
• To burn dual-boot image and rootfs to eMMC and FlexSPI0 NOR flash, perform the following steps:
1. Prepare imx-boot-imx8ulpevk-sd.bin-flash_singleboot_m33, imx-boot-imx8ulpevk-
sd.bin-flash_dualboot, imx-boot-imx8ulpevk-sd.bin-flash_dualboot_m33, and
<rootfs.wic>.
2. Update the UUU script file uuu_8ulp_dual.auto with the file path and name of the images above.
3. Run uuu mfgtools/scripts/samples/uuu_8ulp_dual.auto.
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>

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

8 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

The following command decompresses zst file and writes into eMMC:

uuu -b emmc_all <bootloader> <rootfs.wic.zst/*>

The following command executes downloading and bootloader (SPL and U-Boot) by USB:

uuu -b spl <bootloader>

The following command burns boot image into FlexSPI NAND:

uuu -b spi_nand <nand_uuu_fw> <bootloader>

The following 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 (for example, 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.
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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

9 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 1. Image layout

Start address (sectors) Size (sectors) Format Description
0x400 bytes (2) 0x9FFC00 bytes (20478) RAW i.MX 6 and i.MX 7 U-Boot and reserved
area
0x8400 (66) 0x9F7C00 (20414) RAW i.MX 8M Quad and i.MX 8M Mini imx-boot
reserved area
0x8000 (64) 0x9F8000 (20416) RAW i.MX 8QuadMax/8QuadXPlus/8M Nano/
8M Plus/8DXL/8DualX/8ULP, i.MX 93, and
i.MX 95
0xa00000 bytes (20480) 500 MB (1024000) FAT Kernel Image and DTBs
0x25800000 bytes Remaining space Ext3/Ext4 Rootfs
(1228800)

4.3.1 Preparing the card

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.
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.

$ sudo dd if=<image name>.wic of=/dev/sdx bs=1M && sync

The entire contents of the SD card are replaced. If the SD card is larger than 4 GB, the additional space is not
accessible.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

10 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.

$ sudo dd if=<U-Boot image> of=/dev/sdx bs=1k seek=<offset> conv=fsync

Where offset is:

• 1 - for i.MX 6 or i.MX 7
• 33 - for i.MX 8QuadMax A0, i.MX 8QuadXPlus A0, and i.MX 8M Quad, and i.MX 8M Mini
• 32 - for i.MX 8QuadXPlus B0, i.MX 8QuadMax B0, i.MX 8DualX, i.MX 8DXL, i.MX 8M Nano, i.MX 8M Plus,
i.MX 8ULP, and i.MX 9
The sectors of SD/eMMC before the “offset” KB are reserved. It may include the partition table.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

11 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 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:
For i.MX 6 and i.MX7, use this command:

$ sudo dd if=zImage_imx_v7_defconfig of=/dev/sdx bs=512 seek=2048 conv=fsync

For i.MX 8, use this command:

sudo dd if=Image-imx8qmsabreauto.bin of=/dev/sdx bs=512 seek=2048 conv=fsync

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.
Choose a command for your board:

$ sudo dd if=zImage-imx6qp-sabreauto.dtb of=/dev/sdx bs=512 seek=20480

conv=fsync
$ sudo dd if=zImage-imx6qp-sabresd.dtb of=/dev/sdx bs=512 seek=20480 conv=fsync
$ sudo dd if=zImage-imx6q-sabreauto.dtb of=/dev/sdx bs=512 seek=20480 conv=fsync
$ sudo dd if=zImage-imx6q-sabresd.dtb of=/dev/sdx bs=512 seek=20480 conv=fsync
$ sudo dd if=zImage-imx6sl-evk.dtb of=/dev/sdx bs=512 seek=20480 conv=fsync
$ sudo dd if=zImage-imx7d-sdb.dtb of=/dev/sdx bs=512 seek=20480 conv=fsync

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:

$ sudo dd if=zImage-imx6ul-14x14-evk.dtb of=/dev/sdx bs=512 seek=20480

conv=fsync

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

12 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

$ sudo dd if=zImage-imx6ull-14x14-evk.dtb of=/dev/sdx bs=512 seek=20480

conv=fsync

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).

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.

zst to /home/user/rootfs:

$ cd /home/user/rootfs
$ tar -jxvf imx-image-multimedia-imx7ulpevk.tar.zst

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.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

13 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

4.4.1.1 Flashing an Arm Cortex-M4 image on QuadSPI

i.MX 6SoloX SABRE-SD, 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 LF6.6.36_2.1.0_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-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 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.
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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

14 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.
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 LF6.6.36_2.1.0_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.
U-Boot > fatload mmc 0:1 0x62000000 sdk20-app.img;
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 Section 4.3.

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:
U-Boot > env default -f -a U-Boot > saveenv U-Boot > reset
2. Configure the U-Boot environment for network communications. The following 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 fdtfile <your dtb image name on the TFTP server>
The user can set a fake MAC address through ethaddr environment 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

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

15 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

4. Query the information about the MMC/SD card.

U-Boot > mmc devU-Boot > mmcinfo
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.
blk# = (microSD Offset)/(SD block length) = 0x100000/0x200 = 0x800

cnt = (image Size)/(SD block length) = 0x800000/0x200 = 0x4000

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} ${fdtfile}
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.
9. Boot the board.
U-Boot > setenv bootcmd_mmc 'run bootargs_base mmcargs;mmc dev;mmc
read ${loadaddr} 0x800 0x4000;mmc read ${fdt_addr} 0x5000 0x800;bootz
${loadaddr} - ${fdt_addr}'
U-Boot > setenv bootcmd 'run bootcmd_mmc'
U-Boot > saveenv

4.4.1.3 Using eMMC

There is an eMMC chip on i.MX SABRE boards, i.MX 8 MEK and EVK boards, i.MX 8M EVK boards, and i.MX
8ULP EVK boards. It is accessed through SDHC4 on i.MX 6 SABRE boards, SDHC3 on i.MX 7Dual SABRE-SD
board, SDHC1 on i.MX 8 MEK/EVK boards and i.MX 8M EVK boards, and SDHC0 on i.MX 8ULP EVK board.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

16 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.
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 LF6.6.36_2.1.0_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 fdtfile <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
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:
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

17 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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
### select mmc dev 0 （USDHC0） on the i.MX 8ULP EVK
U-boot > mmc dev 0
### 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} ${fdtfile}
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 indexes 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;
mmc read ${loadaddr} 0x800 0x4000; mmc read ${fdt_addr} 0x5000 0x800;bootz
${loadaddr} - ${fdt_addr} '
U-Boot > setenv bootcmd 'run mmcboot'
U-Boot > saveenv
9. Boot up the system through the rootfs in eMMC, using the CLAA WVGA panel as display:
• For i.MX 6 boards:
U-Boot > setenv mmcargs 'setenv bootargs ${bootargs}
root=/dev/mmcblk3p2 rootwait rw video=mxcfb0:dev=lcd,CLAA-WVGA,if=RGB565
ip=dhcp'
• For i.MX 7Dual SABRE boards:
U-Boot > setenv mmcargs 'setenv bootargs ${bootargs}
root=/dev/mmcblk2p2 rootwait rw video=mxcfb0:dev=lcd,CLAA-WVGA,if=RGB565
ip=dhcp'
10. Boot up the system through rootfs in eMMC, using HDMI as display:
• For i.MX 6 boards:
U-Boot > setenv mmcargs 'setenv bootargs ${bootargs} root=/dev/mmcblk3p2
rootwait rw video=mxcfb0:dev=hdmi,1920x1080M@60,if=RGB24'
• For i.MX 7Dual SABRE boards:
U-Boot > setenv mmcargs 'setenv bootargs ${bootargs} root=/dev/mmcblk2p2
rootwait rw video=mxcfb0:dev=hdmi,1920x1080M@60,if=RGB24'
• 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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

18 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.
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 Section 4.4.2 or Section 4.3.

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, 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 LF6.6.36_2.1.0_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-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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

19 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 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.
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 Section 4.3.4. Then boot from
the SD card.
3. TFTP the U-Boot image.
tftpboot ${loadaddr} u-boot.imx
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

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

20 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x00080bff
Device Boot Start End Blocks Id System
4. As described in Section 4.6, 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)
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!
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

21 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Calling ioctl() to re-read $SD partition table

5. Check the partitions (see above) to determine the name of the partition. $PARTITION is used here to
indicate the partition to be formatted. Format the MMC/SD partitions as ext3 or ext4 type. For example, to
use ext3:
root@ ~$ mkfs.ext3 /dev/$PARTITION
mke2fs 1.42 (29-Nov-2011)
Filesystem label=
OS type: Linux
Block size=4096 (log=2)
Fragment size=4096 (log=2)
248992 inodes, 994184 blocks
49709 blocks (5.00%) reserved for the super user
First data block=0
Maximum filesystem blocks=1019215872
31 block groups
32768 blocks per group, 32768 fragments per group
8032 inodes per group
Superblock backups stored on blocks:
32768, 98304, 163840, 229376, 294912, 819200, 884736
Writing inode tables: done
Creating journal (16384 blocks): done
Writing superblocks and filesystem accounting information: done
This filesystem will be automatically checked every 20 mounts or
180 days, whichever comes first. Use tune2fs -c or -i to override.
6. Copy the rootfs contents to the MMC/SD card. The name may vary from the one used below. Check the
directory for the rootfs desired. (Copy the *.ext2 to NFS rootfs).
mkdir /mnt/tmpmnt
mount -t ext3 -o loop /imx-image-multimedia.ext3 /mnt/tmpmnt
cd /mnt
mkdir mmcblk0p1
mount -t ext3 /dev/$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 Section 4.7 to boot the image from
the MMC/SD card.
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 filename are synchronized with the filename pointed to by the U-Boot
environment variables: fdtfile and image. Use the print command under U-Boot to display these two
environment variables. For example:
print fdtfile 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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

22 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x3302445d
Device Boot Start End Blocks Id System
/dev/sdb1 4096 106495 51200 83 Linux
Command (m for help): n
Partition type:
p primary (1 primary, 0 extended, 3 free)
e extended
Select (default p): p
Partition number (1-4, default 2): 2
First sector (2048-30318591, default 2048): 106496
Last sector, +sectors or +size{K,M,G} (106496-30318591, default 30318591):
Using default value 30318591
Command (m for help): p
Disk /dev/sdb: 15.5 GB, 15523119104 bytes
64 heads, 32 sectors/track, 14804 cylinders, total 30318592 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x3302445d
Device Boot Start End Blocks Id System
/dev/sdb1 4096 106495 51200 83 Linux
/dev/sdb2 106496 30318591 15106048 83 Linux
Command (m for help): w
The partition table has been altered!
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 with 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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

23 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 SABRE-SD boards.

Table 2. Booting from SD1 on i.MX 7Dual SABRE-SD

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW2 OFF OFF ON OFF OFF OFF OFF OFF
SW3 ON OFF - - - - - -

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

Switch D1 D2 D3 D4
SW1 ON OFF OFF ON

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

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW2 OFF OFF ON ON OFF OFF - -

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

Switch D1 D2 D3 D4
SW2 ON ON OFF OFF

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

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW6 ON OFF OFF OFF OFF OFF ON OFF

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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

24 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 7. Booting from TF on i.MX 6UltraLite EVK and i.MX 6ULL EVK
Switch D1 D2 D3 D4
SW601 OFF OFF ON OFF
SW602 ON OFF - -

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

Switch D1 D2 D3 D4
SW801 ON ON OFF OFF
SW802 ON OFF - -

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

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW1101 OFF ON OFF OFF OFF ON ON OFF
SW1102 OFF OFF ON ON OFF ON OFF OFF

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

Switch D1 D2 D3 D4
SW1101 ON ON OFF OFF

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.

Table 11. Booting from TF on i.MX 8M Plus EVK

Switch D1 D2 D3 D4
SW4 OFF OFF ON ON

The following table shows the DIP switch settings for booting from the USDHC2 slot.

Table 12. Booting from USDHC2 on i.MX 93 11x11 EVK and i.MX 91 11x11 EVK
Switch D1 D2 D3 D4
SW1301 OFF ON OFF OFF

Table 13. Booting from USDHC2 on i.MX 93 9x9 QSB

Switch D1 D2 D3 D4
SW601 OFF OFF ON ON

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

25 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 14. Booting from USDHC2 on i.MX 91 9x9 QSB

Switch D1 D2 D3 D4
SW3 OFF OFF ON ON

Table 15. Booting from USDHC2 on i.MX 95 19x19 EVK and i.MX 95 15x15 EVK
Switch D1 D2 D3 D4
SW7 ON OFF ON ON

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 16. Booting from an SD card in slot SD3 on i.MX 6 SABRE-AI

Switch D1 D2 D3 D4 D5 D6 D7 D8 D9 D10
S1 X X X OFF ON X X X X X
S2 X OFF ON OFF - - - - - -
S3 OFF OFF ON OFF - - - - - -

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 17. Booting from an MMC card in Slot SD3 on i.MX 6SoloX SABRE-AI
Switch D1 D2 D3 D4 D5 D6 D7 D8
S4 OFF ON OFF X OFF OFF ON OFF
S3 X OFF OFF OFF ON ON OFF OFF
S1 OFF OFF ON OFF - - - -

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 18. Booting from an SD card in slot SD3 on i.MX 6 SABRE-SD

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW6 OFF ON OFF OFF OFF OFF ON OFF

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 19. Booting from an SD card in slot SD4 on i.MX 6SoloX SABRE-SD
Switch D1 D2 D3 D4 D5 D6 D7 D8
SW10 OFF OFF OFF OFF OFF OFF OFF OFF

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

26 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 19. Booting from an SD card in slot SD4 on i.MX 6SoloX SABRE-SD...continued
Switch D1 D2 D3 D4 D5 D6 D7 D8
SW11 OFF OFF ON ON ON OFF OFF OFF
SW12 OFF ON OFF OFF OFF OFF OFF OFF

Table 20. Booting from an MMC card in slot SD4 on i.MX 6SoloX SABRE-SD
Switch D1 D2 D3 D4 D5 D6 D7 D8
SW10 OFF OFF OFF OFF OFF OFF OFF OFF
SW11 OFF OFF ON ON ON OFF OFF OFF
SW12 OFF ON ON OFF OFF OFF OFF OFF

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 21. Booting from eMMC on i.MX 6 SABRE-SD

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW6 ON ON OFF ON OFF ON ON OFF

i.MX 7Dual is different from i.MX 6. The eMMC uses the SD3 pin connections from the i.MX 7Dual processor.

Table 22. Booting from eMMC on i.MX 7Dual SABRE-SD

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW2 OFF ON OFF ON OFF OFF OFF OFF
SW3 ON OFF - - - - - -

The following table shows the boot switch settings to boot from eMMC4.4 on the i.MX 7ULP EVK boards.

Table 23. Booting from eMMC on i.MX 7ULP EVK

Switch D1 D2 D3 D4
SW1 ON OFF OFF OFF

The following table shows the boot switch settings to boot from eMMC5.0 on the i.MX 8QuadMax MEK boards.

Table 24. Booting from eMMC on i.MX 8QuadMax MEK

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW2 OFF OFF OFF ON OFF OFF - -

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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

27 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 25. Booting from eMMC on i.MX 8QuadXPlus MEK

Switch D1 D2 D3 D4
SW2 OFF ON OFF OFF

The following table shows the boot switch settings to boot from eMMC5.0 on the i.MX 8M Quad EVK boards.

Table 26. Booting from eMMC on i.MX 8M Quad EVK

Switch D1 D2 D3 D4
SW801 OFF OFF ON OFF

The following table shows the boot switch settings to boot from eMMC5.1 on the i.MX 8M Mini EVK boards.

Table 27. Booting from eMMC on i.MX 8M Mini EVK

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW1101 OFF ON ON ON OFF OFF ON OFF
SW1102 OFF OFF OFF OFF ON OFF ON OFF

The following table shows the boot switch settings to boot from eMMC5.1 on the i.MX 8M Nano EVK boards.

Table 28. Booting from eMMC on i.MX 8M Nano EVK

Switch D1 D2 D3 D4
SW1101 OFF ON OFF OFF

The following table shows the boot switch settings to boot from eMMC5.1 on the i.MX 8M Plus EVK boards.

Table 29. Booting from eMMC on i.MX 8M Plus EVK

Switch D1 D2 D3 D4
SW4 OFF OFF OFF ON

The following table lists the boot switch settings to boot from eMMC5.1 on the i.MX 8ULP EVK.

Table 30. Singleboot booting from eMMC on i.MX 8ULP EVK

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW5 OFF OFF OFF OFF OFF OFF OFF ON

Table 31. Dualboot booting from eMMC for A35 on i.MX 8ULP EVK
Switch D1 D2 D3 D4 D5 D6 D7 D8
SW5 OFF ON OFF OFF OFF OFF OFF ON

Table 32. Booting from eMMC on i.MX 93 11x11 EVK and i.MX 91 11x11 EVK
Switch D1 D2 D3 D4
SW1301 OFF OFF OFF OFF

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

28 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 33. Booting from eMMC on i.MX 93 9x9 QSB

Switch D1 D2 D3 D4
SW601 OFF OFF ON OFF

Table 34. Booting from eMMC on i.MX 91 9x9 QSB

Switch D1 D2 D3 D4
SW3 OFF OFF ON OFF

Table 35. Booting from eMMC on i.MX 95 19x19 EVK and i.MX 95 15x15 EVK
Switch D1 D2 D3 D4
SW7 ON OFF ON OFF

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 36. Booting from SATA on i.MX 6 SABRE-SD

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW6 OFF OFF OFF OFF OFF ON OFF OFF

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 37. Booting from NAND on i.MX 6 SABRE-AI

Switch D1 D2 D3 D4 D5 D6 D7 D8 D9 D10
S1 OFF OFF OFF ON OFF OFF OFF OFF OFF OFF
S2 OFF OFF OFF ON - - - - - -
S3 OFF OFF ON OFF - - - - - -

The following table shows the DIP switch settings needed to boot from NAND for i.MX 7Dual SABRE-SD
boards.

Table 38. Booting from NAND on i.MX 7Dual SABRE-SD

Switch D1 D2 D3 D4 D5 D6 D7 D8 D9 D10
S2 OFF ON ON X X X X OFF - -
S3 ON OFF X X X X X X - -

The following table shows the DIP switch settings needed to boot from NAND for i.MX 8M Mini DDR4 EVK
boards.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

29 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 39. Booting from NAND on i.MX 8M Mini DDR4 EVK

Switch D1 D2 D3 D4 D5 D6 D7 D8 D9 D10
SW1101 OFF ON OFF OFF OFF OFF OFF ON - -
SW1102 OFF OFF OFF ON ON ON ON OFF - -

Table 40. Booting from FlexSPI NAND on i.MX 91 9x9 QSB

Switch D1 D2 D3 D4
SW3 OFF ON OFF ON
SW4 ON X X X

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 41. Booting from SPI-NOR on i.MX 6 SABRE-AI

Switch D1 D2 D3 D4 D5 D6 D7 D8 D9 D10
S1 X X X X X X X X X X
S2 ON ON OFF OFF OFF OFF OFF OFF OFF OFF
S3 OFF OFF ON OFF - - - - - -

4.5.9 Booting from EIM (Parallel) NOR

The following table shows the DIP switch settings to boot from NOR.

Table 42. Booting from EIM NOR on i.MX 6 SABRE-AI

Switch D1 D2 D3 D4 D5 D6 D7 D8 D9 D10
S1 X X X OFF OFF ON X X X X
S2 X OFF OFF OFF - - - - - -
S3 OFF OFF ON OFF - - - - - -

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 43. Booting from QuadSPI on i.MX 6SoloX SABRE-SD

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW10 OFF OFF OFF OFF OFF OFF OFF OFF
SW11 OFF OFF OFF OFF OFF OFF OFF OFF
SW12 OFF OFF OFF ON ON OFF OFF OFF

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

30 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 44. Booting from QuadSPI on i.MX 6SoloX SABRE-AI

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW4 OFF OFF OFF OFF ON OFF OFF OFF
SW3 OFF OFF OFF OFF OFF OFF OFF OFF
SW1 OFF OFF ON OFF - - - -

Table 45. Booting from QuadSPI on i.MX 7Dual SABRE-SD

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW2 ON OFF OFF OFF OFF OFF OFF OFF
SW3 ON OFF - - - - - -

Table 46. Booting from QuadSPI on i.MX 6UltraLite EVK and i.MX 6ULL EVK
Switch D1 D2 D3 D4
SW601 OFF OFF OFF OFF
SW602 ON OFF - -

Table 47. Booting from FlexSPI on i.MX 8QuadMax MEK

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW2 OFF OFF OFF ON ON OFF - -

Table 48. Booting from FlexSPI on i.MX 8QuadXPlus MEK

Switch D1 D2 D3 D4
SW2 OFF ON ON OFF

Table 49. Booting from FlexSPI on i.MX 8M Mini LPDDR4 EVK

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW1101 OFF ON OFF OFF OFF OFF OFF OFF
SW1102 OFF ON OFF OFF OFF OFF OFF ON

Table 50. Booting from QuadSPI on i.MX Nano EVK

Switch D1 D2 D3 D4
SW601 OFF OFF OFF OFF
SW602 ON OFF - -

Table 51. Booting from QuadSPI on i.MX 8M Plus EVK

Switch D1 D2 D3 D4
SW4 OFF ON ON OFF

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

31 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 52. Singleboot booting from FlexSPI NOR on i.MX 8ULP EVK
Switch D1 D2 D3 D4 D5 D6 D7 D8
SW5 OFF OFF OFF OFF OFF ON OFF ON

Table 53. Dualboot booting from FlexSPI for A35 on i.MX 8ULP EVK
Switch D1 D2 D3 D4 D5 D6 D7 D8
SW5 OFF ON OFF OFF OFF ON OFF ON

Table 54. Booting from M.2 FlexSPI NOR on i.MX 91 11x11 EVK
Switch D1 D2 D3 D4
SW1301 OFF ON OFF ON

Table 55. Booting from M.2 FlexSPI NOR on i.MX 91 9x9 QSB
Switch D1 D2 D3 D4
SW3 OFF ON OFF OFF
SW4 OFF X X X

Table 56. Booting from FlexSPI NOR on i.MX 95 19x19 EVK

Switch D1 D2 D3 D4
SW7 ON ON OFF OFF

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 57. Setup for the Manufacturing Tool on i.MX 6 SABRE-AI

Switch D1 D2 D3 D4
S3 OFF ON OFF OFF

Table 58. Setup for the Manufacturing Tool on i.MX 7Dual SABRE-SD
Switch D1 D2 D3 D4
S3 OFF ON - -

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

32 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 59. Setup for Manufacturing Tool on i.MX 6UltraLite EVK and i.MX 6ULL EVK
Switch D1 D2
SW602 OFF ON

Table 60. Setup for Manufacturing Tool on i.MX 7ULP EVK

Switch D1 D2 D3 D4
SW1 OFF ON - -

Table 61. Setup for Manufacturing Tool on i.MX 8M Quad EVK

Switch D1 D2
SW802 OFF ON

Table 62. Setup for Manufacturing Tool on i.MX 8M Plus EVK

Switch D1 D2 D3 D4
SW4 OFF OFF ON OFF

Table 63. Setup for Manufacturing Tool on i.MX 8M Mini EVK

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW1101 ON OFF X X X X X X
SW1102 X X X X X X X X

Table 64. Setup for Manufacturing Tool on i.MX 8M Nano EVK

Switch D1 D2 D3 D4
SW1101 ON OFF OFF OFF

Table 65. Setup for Manufacturing Tool on i.MX 8QuadMax MEK

Switch D1 D2 D3 D4 D5 D6
SW2 OFF OFF ON OFF OFF OFF

Note:
The following settings are same for the i.MX 8DualX MEK and i.MX 8DXL EVK boards (8DXL EVK uses SW1).

Table 66. Setup for Manufacturing Tool on i.MX 8QuadXPlus MEK

Switch D1 D2 D3 D4
SW2 ON OFF OFF OFF

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

33 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 67. Setup for Manufacturing Tool on i.MX 8ULP EVK

Switch D1 D2 D3 D4 D5 D6 D7 D8
SW2 OFF OFF OFF OFF OFF OFF ON OFF

Table 68. Setup for Manufacturing Tool on i.MX 93 11x11 EVK and i.MX 91 11x11 EVK
Switch D1 D2 D3 D4
SW1301 ON ON OFF OFF

Table 69. Setup for Manufacturing Tool on i.MX 93 9x9 QSB

Switch D1 D2 D3 D4
SW601 OFF OFF OFF ON

Table 70. Setup for Manufacturing Tool on i.MX 91 9x9 QSB

Switch D1 D2 D3 D4
SW3 OFF OFF OFF ON

Table 71. Setup for Manufacturing Tool on i.MX 95 19x19 EVK and i.MX 95 15x15 EVK
Switch D1 D2 D3 D4
SW7 ON OFF OFF ON

Note:
If the SD card with bootable image is plugged in SD2 (baseboard), ROM will not fall back into the serial
download mode.

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 i.MX Yocto Project User's Guide (UG10164). In the following
command, set 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.
DISTRO=fsl-imx-fb MACHINE=Target-Machine bitbake core-image-minimal -c
populate_sdk
Note:
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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

34 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• 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 it can be placed
anywhere on the host machine.
On the host machine, the following are the steps to build U-Boot and Kernel.
Toolchain Configuration:
• For i.MX 6 and i.MX 7 builds on the host machine, set the environment with the following command before
building.
source /opt/fsl-imx-fb/6.6-nanbield/environment-setup-cortexa9hf-neon-poky-linux-gnueab
export ARCH=arm

• For i.MX 8 and i.MX 9 builds on the host machine, set the environment with the following command before
building.
source /opt/fsl-imx-xwayland/6.6-nanbield/environment-setup-aarch64-poky-linux
export ARCH=arm64

U-Boot:
Download source by cloning with:

git clone https://github.com/nxp-imx/uboot-imx -b lf_v2023.04

cd uboot-imx

• 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.
make clean
make mx6ull_14x14_evk_defconfig
make
• 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
• For i.MX 8M Quad EVK:
make distclean
make imx8mq_evk_defconfig
make
• For i.MX 8M LPDDR4 EVK:
make distclean
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

35 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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
• For i.MX 8ULP EVK board:
make distclean
make imx8ulp_evk_defconfig
make
• For i.MX 93 11x11 EVK board:
make distclean
make imx93_11x11_evk_defconfig
make
• For i.MX 93 9x9 QSB board:
make distclean
make imx93_9x9_qsb_defconfig
make

• For i.MX 91 11x11 EVK board:

make distclean
make imx91_11x11_evk_defconfig
make
• For i.MX 95 19x19 EVK board:
make distclean
make imx95_19x19_evk_defconfig
make
• For i.MX 95 15x15 EVK board:
make distclean
make imx95_15x15_evk_defconfig
make

Kernel:
Download source by cloning with:

git clone https://github.com/nxp-imx/linux-imx -b lf-6.6.y

cd linux-imx

• To build the kernel in the standalone environment for i.MX 6 and i.MX 7, execute the following commands:
make imx_v7_defconfig
make

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

36 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• To build the kernel in the standalone environment for i.MX 8 and i.MX 9, execute the following commands:
make imx_v8_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.
The fused modules are as follows:
• G0: TSC, ADC2, FLEXCAN1, FLEXCAN2, FREQ_MON, TEMP_MON, VOLT_MONLCDIF, CSI, ENET2,
CAAM, USB_OTG2, SAI23, BEE, UART5678, PWM5678, ECSPI34, I2C34, GPT2, and EPIT2.
• G1: TSC, ADC2, FLEXCAN2, FREQ_MON, TEMP_MON, VOLT_MON, LCDIF, CSI, ENET2, and BEE.
• G2: FREQ_MON, TEMP_MON, VOLT_MON, and BEE.
• G3: No fused module.
U-Boot configuration changes:
G0:

/* #define CONFIG_VIDEO */
#define CONFIG_FEC_ENET_DEV 0
/* #define CONFIG_CMD_BEE */
#define CONFIG_USB_MAX_CONTROLLER_COUNT 1

G1:

/* #define CONFIG_VIDEO */
#define CONFIG_FEC_ENET_DEV 0
/* #define CONFIG_CMD_BEE */

G2:

/* #define CONFIG_CMD_BEE */

G3: No change.

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 (mx8qmb0-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/.
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/.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

37 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

4. Copy the SECO firmware container images (mx8qxpb0-ahab-container.img and mx8qxpc0-ahab-

container.img) to imx-mkimage/iMX8QM/.
5. Run make SOC=iMX8QX flash to generate flash.bin for i.MX 8QuadXPlus B0, and run make
SOC=iMX8QX REV=C0 flash to generate flash.bin for i.MX 8QuadXPlus C0.
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 and mx8dxlb0-ahab-
container.img) to imx-mkimage/iMX8DXL/.
5. Run make SOC=iMX8DXL flash to generate flash.bin for i.MX 8DXL A1, and run make
SOC=iMX8DXL REV=B0 flash to generate flash.bin for i.MX 8DXL B0.
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 72. Matrix table for targets of i.MX 8QuadMax, i.MX 8QuadXPlus, and i.MX 8DXL
- OP-TEE U-Boot SPL Cortex-M4
flash_spl Yes Yes Yes No
flash No Yes No No
flash_linux_m4 Yes Yes Yes Yes
flash_regression_ No Yes No Yes
linux_m4

For i.MX 8ULP EVK, to build imx-boot image by using imx-mkimage, perform the following steps:
1. Copy u-boot.bin from u-boot/u-boot.bin and u-boot-spl.bin from u-boot/spl/u-boot-
spl.bin to imx-mkimage/iMX8ULP/.
2. Copy bl31.bin from Arm Trusted firmware (imx-atf) to imx-mkimage/iMX8ULP/.
3. Copy the image of Sentinel firmware container mx8ulpa0-ahab-container.img to imx-mkimage/
iMX8ULP/.
4. Copy the image of uPower firmware image upower.bin to imx-mkimage/iMX8ULP/.
5. Copy the Cortex-M33 image m33_image.bin to imx-mkimage/iMX8ULP/.
6. If using OP-TEE, copy tee.bin to imx-mkimage/iMX8ULP/. The bl31.bin copied in Step 2 must be
built with OP-TEE SPD enabled.
7. Run make SOC=iMX8ULP flash_singleboot_m33 to generate flash.bin.
Note: For the location where the binaries for Sentinel/SECO/uPower/M33/V2X firmwares are available to
download, see the Table "BSP and multimedia standard packages" in the i.MX Linux Release Notes (RN00210).
The following table list the imx-mkimage targets used on i.MX 8ULP.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

38 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 73. imx-mkimage targets used on i.MX 8ULP

Boot type Cortex-A35 Cortex-M33 SW5[8:1]
Single Boot (Default) Boot Cortex-A35 + Cortex-M33 from eMMC: 1000_xx00 Single Boot-
make SOC=iMX8ULP flash_singleboot_m33 eMMC

Boot Cortex-A35 + Cortex-M33 from FlexSPI NOR: 1010_xx00 Single Boot-

make SOC=iMX8ULP flash_singleboot_m33_flexspi NOR

Dual Boot make SOC=iMX8ULP flash_ For RAM target: 1000_0010 A35-eMMC/
dualboot make SOC=iMX8ULP M33-NOR
make SOC=iMX8ULP flash_ flash_dualboot_m33For Flash target: 1010_0010 A35-Nor/
dualboot_flexspi make SOC=iMX8ULP flash_ M33-NOR
dualboot_m33_xip
Low Power - 1000_00x1 A35-eMMC/
Boot M33-NOR
- 1010_00x1 A35-Nor/
M33-NOR

For i.MX 8M EVK, 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), imx8mm-evk.dtb (for i.MX 8M Mini LPDDR4 EVK),
imx8mm-ddr4-evk.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 the 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 the 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 the 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 the firmware-imx
package to imx-mkimage/iMX8M/.
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).
To boot with eMMC fasboot on i.MX 8M Quad EVK and i.MX 8M Mini LPDDR4 EVK, use
flash_evk_emmc_fastboot target.
For i.MX 93 A1 and i.MX 91, to build imx-boot image by using imx-mkimage, perform the following steps:

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

39 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

1. Copy u-boot.bin from u-boot/u-boot.bin and u-boot-spl.bin from uboot/ spl/u-boot-

spl.bin to imx-mkimage/iMX93/.
2. Copy bl31.bin from Arm Trusted firmware (imx-atf) to imx-mkimage/iMX93/.
3. Copy the image of Sentinel firmware container mx93a1-ahab-container.img to imx-mkimage/iMX9/.
4. If using OP-TEE, copy tee.bin to imx-mkimage/iMX93/. The bl31.bin copied in Step 2 must be built
with OP-TEE SPD enabled.
5. Run the following command to generate flash.bin.
• For i.MX 93 A1:
make SOC=iMX93 flash_singleboot
• For i.MX 91:
make SOC=iMX91 flash_singleboot
To generate boot image for FlexSPI NAND boot on i.MX 91, run the following command:
make SOC=iMX91 flash_singleboot_spinand
Note: Two binaries are generated:
• flash.bin is the boot image burned into the SPI NAND.
• flash_fw.bin is the NAND UUU FW image used by the UUU tool.
For i.MX 95, to build imx-boot image by using imx-mkimage, perform the following steps:
Note: For i.MX 93, GDET is disabled after the ROM is booted with flash_singleboot. To enable GDET
permanently, use flash_singleboot_gdet to generate the image. To enable GDET only during the ELE API
call, use flash_singleboot_gdet_auto.
1. Copy u-boot.bin from u-boot/u-boot.bin and u-boot-spl.bin from uboot/spl/u-
bootspl.bin to imx-mkimage/iMX95/.
2. Copy tee.bin from OP-TEE to imx-mkimage/iMX95/.
3. Copy bl31.bin from Arm Trusted firmware (imx-atf) to imx-mkimage/iMX95/.
4. Copy the image of ELE and V2X container mx95a1-ahab-container.img to imx-mkimage/iMX95/.
5. Copy the System Manager image M33_image.bin to imx-mkimage/iMX95/.
6. Copy the SDK image M7_image.bin to imx-mkimage/iMX95/.
7. Copy the OEI images oei-m33-ddr.bin and oei-m33-tcm.bin to imx-mkimage/iMX95/.
8. Copy the DDR PHY firmware images (LPDDR5): lpddr5_dmem_qb_v202311.bin,
lpddr5_dmem_v202311.bin, lpddr5_imem_qb_v202311.bin, lpddr5_imem_v202311.bin, or
firmware images (LPDDR4x): lpddr4x_dmem_qb_v202311.bin, lpddr4x_dmem_v202311.bin,
lpddr4x_imem_qb_v202311.bin, lpddr4x_imem_v202311.bin to imx-mkimage/iMX95.
9. Select one of following commands to generate flash.bin:
• On i.MX 95 19x19 EVK, for all images, run:
make SOC=iMX95 OEI=YES flash_all
• To build images for i.MX 95 15x15 EVK, add the LPDDR_TYPE=lpddr4x option to the build command.
For example, to include all images, run:
make SOC=iMX95 OEI=YES LPDDR_TYPE=lpddr4x flash_all
• For Cortex-A55 images, and without Cortex-M7, run:
make SOC=iMX95 OEI=YES flash_a55
• For Cortex-A55 images for FlexSPI NOR boot, and without Cortex-M7, run:
make SOC=iMX95 OEI=YES flash_a55_flexspi

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

40 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 use 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.
For example:

mtdparts=gpmi-nand:64m(boot),16m(kernel),16m(dtb),-(rootfs)

4.6.3 Parallel NOR flash memory map

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 i.MX .dtsidevice 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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

41 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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:

mtdparts=spi32766.0:768k(uboot),8k(env),128k(dtb),-(kernel)

4.6.5 QuadSPI flash memory map

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:

mtdparts=21e4000.qspi:1m(uboot),8m(kernel),1m(dtb),-(user)

U-Boot has the mapping below to help in accessing the QuadSPI flash in U-Boot for non-parallel mode.

Table 74. U-Boot mapping for QuadSPI

Device on hardware Device in U-Boot Memory address in U-Boot Remark
QuadSPI1 Port A CS0 sf probe 0:0 on i.MX 6SoloX SABRE-AI 0x60000000 -
board, i.MX 7Dual SABRE-SD board, i. 0x08000000
MX 6UltraLite EVK board, i.MX 8QuadMax
MEK, i.MX 8QuadXPlus MEK, and i.MX
8DXL EVK
QuadSPI1 Port B CS0 sf probe 1:0 on i.MX 6 SoloX SABRE-AI 0x68000000 -
board
QuadSPI2 Port A CS0 sf probe 0:0 on i.MX 6SoloX SABRE-SD 0x70000000 -
board
QuadSPI2 Port B CS0 sf probe 1:0 on i.MX 6SoloX SABRE-SD 0x78000000 -
board

Table 75. U-Boot mapping for FlexSPI for i.MX 8ULP

Device on hardware Device in U-Boot Memory address in U-Boot Remark
Flexspi0 PortA CS0 sf probe 0:0 0x04000000 -
Flexspi2 PortA CS0 sf probe 2:0 0x60000000 -

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 Section 4.4 or
Section 4.3. If you have not set up your Serial Terminal, see Section 3.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

42 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.
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
baud rate and the port can be modified. Therefore, for NFS, it might be ttymxc3.

U-Boot > setenv consoleinfo 'console=ttymxc2,115200'

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 (RN00210). 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,1920x1080
M@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
• U-Boot > setenv displayinfo 'video=mxcfb0:mxcfb0:dev=lcd,if=RGB565
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 76. Board-specific default values

Variable 6Quad, 6 6SoloX SD 7Dual 6UltraLite, 7ULP EVK 8Quad 8ULP 8M Quad/8M i.MX 93, i.MX 95 Description
QuadPlus, SABRE-SD 6ULL, and Max, 8 Mini EVK i.MX 91
and 6Dual 6ULZ EVK QuadXPlus,
Lite SABRE 8DualX, and
(AI and SD) 8DXL
loadaddr 0x12000000 0x80800000 0x80800000 0x80800000 0x60800000 0x80200000 0x80400000 0x40400000 0x80400000 0x90400000 Address in
the memory

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

43 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 76. Board-specific default values...continued

Variable 6Quad, 6 6SoloX SD 7Dual 6UltraLite, 7ULP EVK 8Quad 8ULP 8M Quad/8M i.MX 93, i.MX 95 Description
QuadPlus, SABRE-SD 6ULL, and Max, 8 Mini EVK i.MX 91
and 6Dual 6ULZ EVK QuadXPlus,
Lite SABRE 8DualX, and
(AI and SD) 8DXL
the kernel
are loaded to

fdt_addr 0x18000000 0x83000000 0x83000000 0x83000000 0x63000000 0x83000000 0x83000000 0x43000000 0x83000000 0x93000000 Address in
the memory
the device
tree code are
copied to

In addition, fdtfile 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 fdtfile 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.
• U-Boot > setenv rootfsinfo 'root=/dev/nfs ip=dhcp nfsroot=${serverip}:
${nfsroot},v3,tcp'
• U-Boot > setenv rootfsinfo 'root=/dev/nfs ip=dhcp weim-nor nfsroot=${serverip}:
${nfsroot},v3,tcp'
• U-Boot > setenv rootfsinfo 'ubi.mtd=4 root=ubi0:rootfs rootfstype=ubifs rootwait
rw mtdparts=gpmi-nand:64m(boot),16m(kernel),16m(dtb),-(rootfs)'
• U-Boot > setenv rootfsinfo 'root=/dev/mmcblk0p2 rootwait rw'

Special settings
i.XM 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:

U-Boot > setenv bootargsset 'setenv bootargs ${consoleinfo} ${rootfsinfo}

${displayinfo} ${special}'

The executed boot command is then as follows. Arguments vary by device.

U-Boot > setenv bootcmd 'run bootargsset; {settings-for-device}; bootz

${loadaddr} - ${fdt_addr}'

4.7.1 Running the image from NAND

NAND can be found on i.MX 6 SABRE-AI boards.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

44 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.

U-Boot > setenv bootcmd 'run bootargsset; nand read ${loadaddr} 0x1000000
0x800000; nand read ${fdt_addr} 0x2000000 0x100000; bootz ${loadaddr} -
${fdt_addr}'

4.7.2 Running Linux OS from Parallel NOR

Parallel NOR is available on i.MX 6 SABRE-AI boards. The following procedure can be used to boot the system
from Parallel NOR.
1. Assume that the kernel image starts at address 0xc0000 bytes.
2. At the U-Boot prompt, set up these variables:
U-Boot > setenv bootcmd 'run bootargsset; cp.b 0x80c0000 ${loadaddr}
0x800000;cp.b 0x80a0000 ${fdt_addr} 0x20000;bootz ${loadaddr} - ${fdt_addr}
'

4.7.3 Running the Linux OS image from QuadSPI

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 ${fdt_addr} 0x800000 0x800; bootz ${loadaddr} -
${fdt_addr} '

4.7.4 Running the Arm Cortex-M4/7/33 image

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).
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

45 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.
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 (RM00293).
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

Note:
If your image has no resource table, such as NXP hello_world.bin, clear the resource table area.
Otherwise, Linux OS may shows the garbage value.
• For i.MX 8M Mini/Nano/Quad LPDDR4 EVK: run #mw 0xb80ff000 0 4 to clear the garbage resource table
area.
• For i.MX 8M Plus LPDDR4 EVK: run #mw 0x550ff000 0 4 to clear the garbage resource table area.
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}

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

46 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

u-boot=> run prepare_mcore

u-boot=> bootaux 0x7e0000

There are two methods to start the remote core: U-Boot bootaux and Linux remoteproc.
Whether to use bootaux or remoteproc to start the remote core, use remoteproc to stop or start the
Cortex-M core for debug purposes only. It is not recommended to stop the Cortex-M core from Linux OS in a
production system.
If you choose to use remoteproc to start the remote core directly, execute run prepare_mcore in U-Boot
before starting the Linux OS.
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:
flash_linux_m4: $(MKIMG) mx8qmb0-ahab-container.img scfw_tcm.bin u-boot-spl.bin
m4_image.bin m4_1_image.bin u-boot-atf-container.img
./$(MKIMG) -soc QM -rev B0 -dcd skip -append mx8qmb0-ahab-container.img -c -
flags 0x00200000 -scfw scfw_tcm.bin -ap u-boot-spl.bin a53 0x00100000 -p3 -m4
m4_image.bin 0 0x34FE0000 -p4 -m4 m4_1_image.bin 1 0x38FE0000 -out flash.bin
cp flash.bin boot-spl-container.img
@flashbin_size=`wc -c flash.bin | awk '{print $$1}'`; \
pad_cnt=$$(((flashbin_size + 0x400 - 1) / 0x400)); \
echo "append u-boot-atf-container.img at $$pad_cnt KB"; \
dd if=u-boot-atf-container.img of=flash.bin bs=1K seek=$
$pad_cnt;
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

On the i.MX 93 11x11 EVK, use bootaux:

=> fatload mmc 1:1 ${loadaddr} imx93-11x11-

evk_m33_TCM_rpmsg_lite_str_echo_rtos.bin
15948 bytes read in 4 ms (3.8 MiB/s)
=> cp.b ${loadaddr} 0x201e0000 ${filesize}
=> bootaux 0x1ffe0000 0
## Starting auxiliary core stack = 0x20020000, pc = 0x0FFE05D5...

Pass clk_ignore_unused in bootargs when using bootaux to kick Arm Cortex-M33. This would increase
overall consumption.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

47 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

For i.MX 93, users can directly start/stop Cortex-M33 using remoteproc as follows:

[ 138.693391] remoteproc remoteproc0: stopped remote processor imx-rproc

root@imx93evk:~# echo ~/imx93_m33_TCM_rpmsg_lite_str_echo_rtos_imxcm33.elf > /
sys/devices/platform/imx93-cm33/remoteproc/remoteproc0/firmware
root@imx93evk:~# echo start > /sys/devices/platform/imx93-cm33/remoteproc/
remoteproc0/state
[ 162.361023] remoteproc remoteproc0: powering up imx-rproc
[ 162.368617] remoteproc remoteproc0: Booting fw image
imx93_m33_TCM_rpmsg_lite_str_echo_rtos_imxcm33.elf, size 175644
[ 162.916805] remoteproc0#vdev0buffer: assigned reserved memory node
vdevbuffer@a4020000
[ 162.926429] virtio_rpmsg_bus virtio0: rpmsg host is online
[ 162.935336] remoteproc0#vdev0buffer: registered virtio0 (type 7)
[ 162.941986] remoteproc0#vdev1buffer: assigned reserved memory node
vdevbuffer@a4020000
[ 162.951071] virtio_rpmsg_bus virtio1: rpmsg host is online
[ 162.956649] virtio_rpmsg_bus virtio1: creating channel rpmsg-virtual-tty-
channel addr 0x1e
[ 162.962559] remoteproc0#vdev1buffer: registered virtio1 (type 7)
[ 162.971179] remoteproc remoteproc0: remote processor imx-rproc is now up
root@imx93evk:~#
root@imx93evk:~# echo stop > /sys/devices/platform/imx93-cm33/remoteproc/
remoteproc0/state
[ 172.114287] remoteproc remoteproc0: stopped remote processor imx-rproc

4.7.5 Linux OS login

The default login user name 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 MS-DOS 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 77. Linux uSDHC relationships

uSDHC mmcroot
uSDHC 1 mmcblk0*
uSDHC 2 mmcblk1*
uSDHC 3 mmcblk2*
uSDHC 4 mmcblk3*

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

48 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.
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 mmcpart 1

U-Boot > setenv loadfdt 'fatload mmc ${mmcdev}:${mmcpart} ${fdt_addr}
${fdtfile}'
U-Boot > setenv loadkernel 'fatload mmc ${mmcdev}:${mmcpart} ${loadaddr} zImage'
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 Network File System (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
fdtfile <your dtb image name on the TFTP server>
U-Boot > setenv
rootfsinfo 'setenv bootargs ${bootargs} root=/dev/nfs ip=dhcp \
nfsroot=${serverip}:/data/rootfs_home/rootfs_mx6,v3,tcp'
U-Boot > setenv bootcmd_net 'run rootfsinfo; dhcp ${image}; dhcp ${fdt_addr} \
${fdtfile}; booti ${loadaddr} - ${fdt_addr}'
U-Boot > setenv bootcmd 'run bootcmd_net'

Note: If the MAC address has not been burned into the fuses, set the MAC address to use the network in U-
Boot.

# eth0:
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

49 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

setenv ethaddr xx:xx:xx:xx:xx:xx

# eth1:
setenv eth1addr xx:xx:xx:xx:xx:xx

4.8 Arm SystemReady-IR

4.8.1 Arm SystemReady-IR ACS compliance test

To run the SystemReady-IR ACS on the i.MX 8M Mini EVK board, refer to the Arm SystemReady-IR ACS from:
https://github.com/ARM-software/arm-systemready
1. Clone the ACS Git.
2. Flash the pre-build release image or image built from source code on to the U-Disk, SD, or eMMC. If the
target storage is also boot storage, flash the boot image (flash.bin) thereafter the ACS image. For
example, on the U-Disk:
sudo dd if=ir_acs_live_image.img of=/dev/sde bs=64M;sync
3. Boot the board. The ACS test will automatically run up.
4. There will be manual step for the power-off test from ACS. Just re-power on the board when run into that
test case.
The test result will be stored on the storage.
5. Use the SCT_Parser to analyze the ACS result. The SCT_Parser can be obtained from: https://github.com/
vstehle/SCT_Parser
6. Copy the acs_results\sct_results\Overall\Summary.ekl file from the RESULT partition in
U-disk to the SCT_Parser folder. Run python3 parser.py --config EBBR.yaml Summary.ekl
EBBR.seq to get the final report.

4.8.2 Capsule update

Use the following command to do the capsule update:
• For SD:
U-Boot > env set dfu_alt_info "mmc 1=1 raw 0x42 0x2000"
• For eMMC:
U-Boot > env set dfu_alt_info "mmc 2=1 raw 0x42 0x2000 mmcpart 1"
U-Boot > efidebug boot add 0 Boot0000 mmc 1:1 capsule1.bin;efidebug boot next 0
U-Boot > setenv serverip 10.192.242.218;dhcp $loadaddr capsule1.bin;fatwrite
mmc 1:1 ${loadaddr} /EFI/UpdateCapsule/capsule1.bin 0x${filesize}
U-Boot > setenv -e -nv -bs -rt -v OsIndications =0x04
U-Boot > efidebug capsule disk-update
reset

Do not interrupt U-Boot. Let the board run into grub. Before grub runs, it should update the bootloader
automatically and remove capsule1.bin. And reboot the board again. The board will boot up with the
updated U-Boot.

4.8.3 Linux distro installation

Fedora34:
1. Download the Fedora IoT 34 version.
2. Burned into the SD card.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

50 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

3. Boot from eMMC.

4. Install distro from the SD card to eMMC.
OpenSUSE:
1. Download the OpenSUSE Tumbleweed version.
2. Burn ISO into the USB disk.
3. Boot from eMMC.
4. Install from ISO to eMMC.

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

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 i.MX Linux Reference Manual (RM00293)
contains information on the power modes that are 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 i.MX Linux
Reference Manual (RM00293).
Use one of the following methods to wake up the system from suspend mode.
• The debug UART can be set as a wake-up source with:
echo enabled > /sys/class/tty/ttymxc0/power/wakeup
Note:
It is ttylp0 for i.MX 8QuadXPlus and i.MX 8QuadMax, and ttyLP0 for i.MX 8DXL.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

51 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

To identify the current debug UART and configure it as a wake-up source:

echo enabled > /sys/class/tty/$(tty | cut -d'/' -f 3)/power/wakeup
• 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:
cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies
• To get the available scaling governors:
cat /sys/devices/system/cpu/*/cpufreq/scaling_available_governors
• To check the current CPU frequency:
cat /sys/devices/system/cpu/*/cpufreq/cpuinfo_cur_freq
The frequency is displayed depending on the governor set.
• To check the maximum frequency:
cat /sys/devices/system/cpu/*/cpufreq/cpuinfo_max_freq
• To check the minimum frequency:
cat /sys/devices/system/cpu/*/cpufreq/cpuinfo_min_freq

These commands set a constant CPU frequency:

• Use the maximum frequency:
echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
• Use the current frequency to be the constant frequency:
echo userspace > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
• 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:
echo userspace > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
echo <frequency> > /sys/devices/system/cpu/cpu0/cpufreq/scaling_setspeed

6.3 Bus frequency scaling

This release does not support the bus frequency scaling feature on i.MX 7ULP EVK.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

52 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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

Execute the following command for i.MX 8ULP to enable system level voltage and frequency scaling:

ifconfig eth0 down

systemctl stop weston
echo 1 > /sys/devices/platform/imx8ulp-lpm/enable

On most systems, the chip enters low power IDLE mode after the above two commands are executed.
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 (RM00293) has more information on the bus frequency in the chapter about
DVFS.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

53 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 (RN00210).

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.

Table 78. Command mapping

Variable $GSTL $PLAYBIN $GPLAY $GSTINSPECT
GStreamer 1.x gst-launch-1.0 playbin or playbin3 gplay-1.0 gst-inspect-1.0

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 or export PLAYBIN=playbin3
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 i.MX Linux Release Notes (RN00210) for
i.MX-specific options, or at "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.

7.3.1 Playback use cases

Playback use cases include the following:

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

54 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• Audio-only playback
• Video-only playback
• Audio/Video file playback
• Multichannel audio playback
• Other methods for playback
• Video playback to multiple displays

7.3.1.1 Audio-only playback

An audio-only playback command uses this format:

$GSTL filesrc location=$clip_name [typefind=true] ! [$id3parse] ! queue !

$audio_parser_plugins
! $audio_decoder_plugin ! $audio_sink_plugin

If the file to be played contains an ID3 header, use the ID3 parser. If the file does not have an ID3 header, this
has no effect.
This example plays an MP3 file in the audio jack output.

$GSTL filesrc location=test.mp3 ! id3demux ! queue ! mpegaudioparse ! beepdec !

pulsesink

7.3.1.2 Video-only playback

$GSTL filesrc location=test.video typefind=true ! $capsfilter !

$demuxer_plugin ! queue max-size-time=0 ! $video_decoder_plugin !
$video_sink_plugin

This is an example of an MP4 container with H.264 encoding format video file playback:

$GSTL filesrc location=test.mp4 typefind=true ! video/quicktime ! aiurdemux !

h264parse ! queue max-size-time=0 ! v4l2h264dec ! autovideosink

This is an example of VPU downscaling from 1920x1080 to 720x480 on i.MX 95:

$GSTL filesrc location=1080p.mp4 ! video/quicktime ! aiurdemux ! h264parse !

v4l2h264dec ! "video/x-raw,width=720,height=480" ! autovideosink

7.3.1.3 Audio/Video file playback

This is an example of a command to play a video file with audio:

$GSTL filesrc location=test_file typefind=true ! $capsfilter

! $demuxer_plugin name=demux demux.
! queue max-size-buffers=0 max-size-time=0 ! $video_decoder_plugin
! $video_sink_plugin demux.
! queue max-size-buffers=0 max-size-time=0 ! $audio_decoder_plugin
! $audio_sink_plugin

This is an example of an AVI file:

$GSTL filesrc location=test.avi typefind=true ! video/x-msvideo

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

55 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

! aiurdemux name=demux demux.

! queue max-size-buffers=0 max-size-time=0 ! $video_decoder_plugin
! autovideosink demux.
! queue max-size-buffers=0 max-size-time=0 ! beepdec
! alsasink

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 PipeWire is enabled, see Section 7.4.

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

Supports the PCM, AC3, and DDP format audio pass-through function by the S/PDIF interface.

$GPLAY /mnt/sdcard/test.wav --audio-sink="alsasink

device=iec958:CARD=imxaudioxcvr,AES3=0x01 pass-through=true"

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 i.MX Porting Guide (IMXBSPPG).

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:

$GSTL $PLAYBIN uri=file:///$file1 $PLAYBIN uri=file:///$file2 video-

sink="overlaysink
display-master=false display-slave=true"

7.3.1.6.2 Routing the same video to different displays

A video can be displayed on multiple displays with a command as follows:

$GSTL $PLAYBIN uri=file:///$filename video-sink="overlaysink display-slave=true"

7.3.1.6.3 Multiple videos overlay

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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

56 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Video 1 Video 2

Video 5

Video 3 Video 4

aaa-052989

Figure 2. Multiple video overlay

gst-launch-1.0 playbin uri=file://$FILE1

video-sink="overlaysink overlay-width=512 overlay-height=384"
playbin uri=file://$FILE2 flags=0x41
video-sink="overlaysink overlay-left=512 overlay-width=512 overlay-
height=384"
playbin uri=file://$FILE3 flags=0x41
video-sink="overlaysink overlay-top=384 overlay-width=512 overlay-
height=384"
playbin uri=file://$FILE4 flags=0x41
video-sink="overlaysink overlay-left=512 overlay-top=384 overlay-width=512
overlay-height=384"
playbin uri=file://$FILE5 flags=0x41
video-sink="overlaysink overlay-left=352 overlay-top=264 overlay-width=320
overlay-height=240 zorder=1"

7.3.2 Audio encoding

Here are some examples for MP3 encoding.

$GSTL filesrc location=test.wav ! wavparse ! lamemp3enc

! filesink location=output.mp3

7.3.3 Video encoding

The following commands provide 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.
For i.MX 6, use the following command:

$GSTL filesrc location=test.yuv

! videoparse format=2 witdh=$WIDTH height=$HEIGHT framerate=30/1
! vpuenc_xxx ! $MUXER ! filesink location=$output

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

57 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

For i.MX 8M Mini/8M Plus and i.MX 95, use the following command:

$GSTL filesrc location=test.yuv

! rawvideoparse format=2 witdh=$WIDTH height=$HEIGHT framerate=30/1
colorimetry=bt709
! v4l2xxxenc $PARSE ! $MUXER ! filesink location=$output

• The target encoder codec type can be:

– MPEG4, H.263, H.264, and MJPEG for i.MX 6
– H.264, VP8 for i.MX 8M Mini
– H.264, HEVC for i.MX 8M Plus and i.MX 95
• The vpuenc_xxx can be:
– vpuenc_mpeg4, vpuenc_h263, vpuenc_h264, and vpuenc_jpeg for i.MX 6
• The v4l2xxxenc can be:
– v4l2h264enc and v4l2vp8enc for i.MX 8M Mini
– v4l2h264enc and v4l2h265enc for i.MX 8M Plus and i.MX 95
• VPU encoder is v4l2h264enc on i.MX 8QuadMax and 8QuadXPlus.
• The $PARSE can be set to h264parse or h265parse if the encoding format is H.264 or H.265. For other
formats, $PARSE is not needed.
• The $MUXER can be set to qtmux, matroskamux, mp4mux, avimux, or flvmux.
• Different muxers support different encoded codec types. Use $GSTINSPECT and $MUXER to see the
capabilities of the muxer to be used.
The following table lists the encoding bitrate modes for i.MX 8M Mini/8M Plus.

Table 79. Encoding bitrate modes

V4L2_CID_MPEG_ Video bitrate mode Encoded file
VIDEO_FRAME_RC_
ENABLE
1 V4L2_MPEG_VIDEO_BITRATE_MODE_CBR Bitrate takes effect limited by MIN_QP and
MAX_QP.
V4L2_CID_MPEG_VIDEO_BITRATE
V4L2_CID_MPEG_VIDEO_H264_MIN_QP
V4L2_CID_MPEG_VIDEO_H264_MAX_QP
1 V4L2_MPEG_VIDEO_BITRATE_MODE_VBR Same as the above case but with bigger bitrate
variance.
0 V4L2_MPEG_VIDEO_BITRATE_MODE_CBR/ Qp takes effect.
V4L2_MPEG_VIDEO_BITRATE_MODE_VBR V4L2_CID_MPEG_VIDEO_H264_I_FRAME_QP
V4L2_CID_MPEG_VIDEO_H264_P_FRAME_QP
V4L2_CID_MPEG_VIDEO_H264_B_FRAME_QP

Use extra-controls property after v4l2xxxenc to specify the value for different codec controls. v4l2-ctl
--device device_path --list-ctrls shows all video device's controls and their values.
For example:

gst-launch-1.0 filesrc location=test.yuv ! rawvideoparse format=2

width=$WIDTH height=$HEIGHT colorimetry=bt709 ! v4l2h264enc extra-
controls="encode,frame_level_rate_control_enable=1,h264_mb_level_rate_control=1,video_bitr
filesink location=output.h264

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

58 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

7.3.4 Transcoding
Transcoding is converting a file from one video encoding to another.
VPU video encoding only works on SoC with VPU encoder support.
For i.MX 6 family with VPU, use the following command:

$GSTL filesrc location=$filename typefind=true ! $capsfilter ! aiurdemux

! vpudec ! imxvideoconvert_ipu ! $CAPS1 ! vpuenc_xxx ! matroskamux ! filesink
location=720p.mkv

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, and vpuenc_jpeg.
For example:

gst-launch-1.0 filesrc location=$FILE.mp4 typefind=true ! video/quicktime !

aiurdemux !
vpudec ! imxvideoconvert_ipu ! video/x-raw,format=NV12,width=1280,height=720 !
vpuenc_h264 ! [h264parse] ! matroskamux ! filesink location=$FILE.mkv

For i.MX 8QuadMax/8QuadXPlus, use the following command:

gst-launch-1.0 filesrc location=$FILE.mp4 typefind=true ! video/quicktime !

aiurdemux ! v4l2h264dec ! queue ! imxvideoconvert_g2d ! queue ! videoconvert !
queue ! v4l2h264enc ! [h264parse] ! matroskamux ! filesink location=$FILE.mkv

For i.MX 8M Mini/8M Plus and i.MX 95, use the following command:

gst-launch-1.0 filesrc location=$FILE.mp4 typefind=true ! video/quicktime !

aiurdemux ! v4l2h264dec ! queue ! v4l2h264enc ! [h264parse] ! matroskamux !
filesink location=$FILE.mkv

Note:
• 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 and i.MX 95, h264parse/h265parse should be added before some mux, because the VPU supports only
the byte-stream output.

7.3.5 Audio recording

Audio recording from EARC or S/PDIF on i.MX 8M Plus.
Since Multimedia version 4.7.0, PCM and compressed format eARC recording pipeline are unified into the
following commandline:

gst-launch-1.0 -v alsasrc provide-clock=false device=hw:1,0 ! spdifdemux !

decodebin ! playsink audio-sink="alsasink device=sysdefault:CARD=wm8960audio
buffer-time=40000 sync=false"

Note: The imxaudioxcvr card number is 1.

For backward compatibility, the following legacy pipelines are still supported:

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

59 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

For PCM format:

gst-launch-1.0 -v alsasrc device=sysdefault:CARD=imxaudioxcvr ! audio/

x-raw,format=S16LE,channels=2,rate=48000 ! playsink audio-sink="alsasink
device=sysdefault:CARD=wm8960audio buffer-time=40000"

For compressed format:

gst-launch-1.0 alsasrc device=sysdefault:CARD=imxaudioxcvr ! audio/x-

raw,format=S16LE,channels=2,rate=48000 ! queue max-size-buffers=0 max-size-
bytes=0 max-size-time=0 ! spdifdemux ! decodebin ! playsink audio-sink="alsasink
device=sysdefault:CARD=wm8960audio buffer-time=40000 sync=false"

Note:
Run the following command to work in EARC mode:

amixer -c imxaudioxcvr cset numid=1,iface=MIXER,name='XCVR Mode' 'eARC'

The following examples show how to make an MP3 or WMA audio recording.
• MP3 recording
$GSTL pulsesrc num-buffers=$NUMBER blocksize=$SIZE ! lamemp3enc
! filesink location=output.mp3

Note:
The recording duration is calculated as $NUMBER * $SIZE * 8 / (samplerate * channel * bit
width).
Therefore, to record 10 seconds of a stereo channel sample with a 44.1K sample rate and a 16 bit width, use
the following command:

$GSTL pulsesrc num-buffers=430 blocksize=4096 ! 'audio/x-raw, rate=44100,

channels=2' ! lamemp3enc
! filesink location=output.mp3

7.3.6 Video recording

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 i.MX BSP Porting Guide
(IMXBSPPG).
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 and i.MX 9 supports opensource plugin v4l2src as camera source.
Use the $GSTINSPECT command to obtain more information about the codec property.
An example of recording might look like this:

$GSTL $V4L2SRC device=$DEVICE num-buffers=300 ! $INPUT_CAPS ! queue !

$video_encoder_plugin ! [h264parse] ! $MUXER ! filesink location=output.
$EXTENSION

• $V4L2SRC can be imxv4l2src, or v4l2src according to the SoC.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

60 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• $DEVICE could be set to /dev/video, /dev/video0, or /dev/video1 according to the system video input
device.
• $INPUT_CAPS should be set to video/x-
raw,format=(string)NV12,width=1920,height=1080,framerate=(fraction)30/1.
• $MUXER can be set to qtmux, matroskamux, mp4mux, avimux, or flvmux.
• $EXTENSION is filename extension according to the muxer type.
Refer to Section Section 7.3.3 to choose the correct $video_encoder_plugin.

7.3.7 Audio/Video recording

This is an example of a command used to record audio and video together:

$GSTL –e $V4L2SRC device=$DEVICE ! $INPUT_CAPS ! queue ! $video_encoder_plugin !

[h264parse] ! queue ! mux. pulsesrc ! 'audio/x-raw, rate=44100, channels=2' !
lamemp3enc ! queue ! mux. $MUXER name=mux ! filesink location= output.
$EXTENSION

• $V4L2SRC can be imxv4l2src or v4l2src according to SoC.

• $INPUT_CAPS should be set to video/x-raw, format=(string)NV12, width=1920,
height=1080, framerate=(fraction)30/1.
• $MUXER can be set to qtmux, matroskamux, mp4mux, avimux, or flvmux.
Refer to Section Section 7.3.3 to choose the correct $video_encoder_plugin.
Common parameters are as follows:
• -e indicates to send EOS when the user presses Ctrl+C to avoid output corruption.
• $EXTENSION is the filename extension according to the multiplexer type.

7.3.8 Camera preview

For i.MX 95 smart camera case (AP1302/AR0144), set up the media pipeline before using the GST v4l2src
plugin to capture. The following is an example about how to set up the pipeline:

media-ctl -l "'ap130x.2-003c':2->'csidev-4ad30000.csi':0 [1]"

media-ctl -l "'csidev-4ad30000.csi':1 -> '4ac10000.syscon:formatter@20':0 [1]"
media-ctl -V "'ap130x.2-003c':2 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'csidev-4ad30000.csi':0 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'4ac10000.syscon:formatter@20':0 [fmt: UYVY8_1X16/1920x1080
field:none]"
media-ctl -V "'crossbar':2 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'mxc_isi.0':0 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'mxc_isi.1':0 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'mxc_isi.2':0 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'mxc_isi.3':0 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'mxc_isi.4':0 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'mxc_isi.5':0 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'mxc_isi.6':0 [fmt: UYVY8_1X16/1920x1080 field:none]"
media-ctl -V "'mxc_isi.7':0 [fmt: UYVY8_1X16/1920x1080 field:none]"

Note: The entity name may vary.

This example displays what the camera sees. It is only available on platforms with a camera.

$GSTL v4l2src ! 'video/x-raw, format=(string)$FORMAT,

width=$WIDTH, height=$HEIGHT, framerate=(fraction)30/1'

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

61 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

! v4l2sink

Camera preview example:

$GSTL v4l2src device=/dev/video1 ! 'video/x-raw,

format=(string)UYVY,width=640,height=480,framerate=(fraction)30/1'
! autovideosink

Parameter comments:
• 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 Libcamera

7.3.9.1 Overview
Note: This section is specific to i.MX 95.
Libcamera is an open source camera stack and framework developed by the Linux media community in
collaboration with the industry. It is a user space library that relies on the existing Linux kernel drivers and API.
It aims to abstract the complexity of the camera subsystem and provide the users with a unified and intuitive
interface for the camera operation.
It supports single and multi-camera use cases. Supported cameras can be either an internal camera typically
connected through a MIPI CSI bus, or USB camera exposing a UVC class.
Native Libcamera applications make direct use of the interfaces exported by the library. A GStreamer adapter is
also supported by the framework: It implements a GStreamer plugin available to applications to use as a source
element for a GStreamer pipeline.

Native GStreamer Native Android

Applications libcamera based V4L2 camera
application application application framework

GStreamer
Libcamera V4L2
source Android HAL
adapters compatibility
plugin

Libcamera
Libcamera
framework
User space
Kernel
V4L2
Media device Video device
Subdevice
aaa-056250

Figure 3. Libcamera software stack

In this release, i.MX 95 supports both raw and smart cameras:

• Raw cameras require using the Neo ISP, to decode and post process the raw image output by the sensor. In
the libcamera framework, a specific ISP is supported through the dedicated NXP Neo ISP pipeline handler
and its associated Image Processing Algorithm (IPA).
• Smart cameras provide a decoded image in various formats. In the libcamera framework, the NXP ISI pipeline
handler aims to handle such cameras.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

62 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Libcamera public API

Libcamera library

Camera devices
Camera device
manager

NXP NXP Neo ISP UVC Other Helpers

NXP Neo
ISI pipeline pipeline handler pipeline pipeline support classes
IPA
handler handler handlers...

USB
Camera CSI ISI ISP
UVC class
aaa-056251

Figure 4. NXP pipeline handlers integration in libcamera

The pipeline handlers’ roles are to implement the platform-specific handling of the media devices from the
hardware pipeline:
• For the Neo ISP pipeline case, it corresponds to the camera, CSI, ISI, and ISP devices.
• For the ISI pipeline case, it corresponds to the camera, CSI, and ISI devices.
They are responsible for the enumeration and configuration of those devices and circulating the image and
metadata buffers between the hardware and software parties.
The IPA is the component implementing the camera control algorithms (White Balance, Auto Exposure Control,
etc.). Algorithms from the IPA process on a camera frame basis the statistics metadata output by the ISP. They
compute and reconfigure the necessary parameters in real time to optimize the ISP and camera operation.
UVC camera support is standard. Its pipeline handler comes with the Libcamera framework.

7.3.9.2 Neo ISP Image Processing Algorithm

Neo ISP libcamera pipeline is used when its pipeline handler is configured as the matching pipeline for the
camera enumeration procedure. See Section 7.3.9.4. During its operation, the Neo ISP pipeline handler
interacts with an Image Processing Algorithm (IPA) relevant to that specific pipeline and ISP. Such IPA
module embeds the 3A control algorithms in charge of the real-time programming of the ISP and the sensor
configuration.
In the libcamera framework, an IPA module is a shared library. That library may be built in-tree (open-source).
Then, it can be bound to the pipeline handler and executed in the same process but a different thread from
libcamera. Alternatively, an IPA implementation can be located out-of-tree giving the option to be licensed as
a closed-source . In that case, IPA can be bound to the pipeline handler but runs in isolation in a separate
process. In both cases, pipeline handler and IPA interwork through the same Inter Processor Communication
(IPC) interface. See libcamera IPA Writer's Guide for more details.
This release contains an open-source IPA implementation for the Neo ISP pipeline, which can be used for
enablement and demonstration purposes.

7.3.9.2.1 NXP Neo IPA

7.3.9.2.1.1 Algorithms
IPA consists of several algorithms. Some of them are used for static configuration, only executing once at the
first frame to configure an ISP block in a fixed way for the whole duration of the camera stream. Conversely,

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

63 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

some algorithms are dynamic, processing the ISP statistics of each frame of the camera stream, to produce an
updated configuration for the ISP or the sensor, or both, applied for the subsequent frames.
The following table lists the NXP Neo ISP algorithms supported in that release.

Table 80. NXP Neo ISP algorithms

Algorithm Comment Type
AGC/AEC (Automatic Gain/ Controls the sensor’s gain and exposure based on ISP histograms Dynamic
Exposure Control) (STAT unit).
AWB (Automatic White Balance) Controls the ISP White Balance correction (OB WB unit) based on Dynamic
the Color Temperature Unit statistics.
DRC (Dynamic Range Configures the DRC unit Globat LUT and gain. Static
Compression )
HDR Decompression Configures the HDR Decompression unit, necessary with the Static
sensors companding the data.
RGBIr Configures the RGBIR and the IR Compression units, required Static
with sensors having a RGBIr matrix.

Note: For detailed ISP hardware specification information, contact your NXP Account Manager or Field
Representative.

7.3.9.2.1.2 Camera support

The support of several camera sensors is provided as a reference for the IPA and the pipeline handler, as
documented in the relevant sections of this libcamera chapter.
Adding the support for a new camera sensor involves updating the IPA with the necessary information to
handle the specifics of that camera, namely:
• Implementation of a CameraHelper object tailored for that camera module
• Creation of a calibration file.

CameraHelper overview
CameraHelper is a helper class used by NXP Neo IPA, whose purpose is to abstract the specifics of that
camera.
Relevant files are in the <libcamera>/src/ipa/nxp/cam_helper source tree directory. It contains
essentially:
• The base class definition (camera_helper.h) and implementation (camera_helper.cpp)
• The camera-specific subclasses (camera_helper_<camera model>.cpp) that may override some
methods of the base class for customization purposes
• The meson configuration file (meson.build) that lists the source files and dependencies to be included for
the build
To support a supplementary camera, an associated CameraHelper subclass must be implemented. That
typically involves the creation of a dedicated source file and updating the meson.build accordingly.

Calibration file
Some IPA algorithms require the usage of parameters that must be configured specifically for a given camera
model. Those configurable values are listed in a dedicated configuration file named <camera model>.yaml.
The IPA module loads at run-time the calibration file relevant to the camera model used for the camera session.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

64 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

The calibration files are present in the libcamera source tree in the directory <libcamera>/src/ipa/nxp/
neo/data. To support a new camera sensor, a dedicated calibration file shall be created there, and the local
meson.build file is updated accordingly so that this file gets installed.
A calibration file is a YAML document that consists of a mapping listing the parameters relevant to each
algorithm. Some algorithms do not require any parameters. Algorithms can also be omitted from the mapping if
they are optional or purposely disabled.
The following table gives a high-level view of the parameters associated to each algorithm.

Table 81. Parameters associated to each algorithm

Algorithm Parameter Comment
AGC/AEC (Automatic Gain/ Histogram scaling factors -
Exposure Control)
AWB (Automatic White Balance) None -
DRC (Dynamic Range • DRC Global lookup table -
Compression ) • DRC Global gain
HDR Decompression Decompression data: knee points, offsets, and ratios Companded data
only
RGBIr • Head Color (CFA matrix selection) RGBIr sensors only
• IR crosstalk
• IR pixels compression

Configuration details for those parameters, when applicable, are documented in the relevant algorithm
implementations, present in the directory <libcamera>/src/ipa/nxp/neo/algorithms.
The context for those parameters is highly related to the Neo ISP pipeline and blocks definition. For detailed ISP
hardware specification information, contact your NXP Account Manager or Field Representative.

7.3.9.3 Camera modules

This release supports the following camera modules:
• X-MX95MBDESER01 (MAX96724 deserializer) + 4 modules X-MX95MBCAM10001 (Omnivision OX03C10
camera + MAX9295 serializer)
– Multi (4 units) 2MP cameras setup
• X-IMX95-OS08A20 (Omnivision OS08A20 camera)
– Single 8MP camera setup
• X-MX95MBDES1003
– Single 5MP RGBIr camera setup
• XRPI-CAM-MINISAS, XRPI-CAM (combo AR0144 sensor + AP1302 ISP)
– Single 1MP smart camera setup with ISP integrated
Operation of these modules requires usage of a dedicated device tree (dtb file). A device tree is provided for
each setup to enable the camera module(s), the ISP, and the IMX-MIPI-HDMI (IT6263) on port LVDS0 or the
IMX-MIPI-DSI (adv7535) on port MIPI DSI.

Table 82. Camera modules

Camera Device tree
X-MX95MBDESER01 imx95-19x19-evk-ox03c10-isp-it6263-lvds0.dtb
4 modules imx95-15x15-evk-ox03c10-isp-adv7357.dtb
X-MX95MBCAM10001 imx95-19x19-verdin-ox03c10-isp-lt8912.dtb

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

65 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 82. Camera modules...continued

Camera Device tree
X-IMX95-OS08A20 imx95-19x19-evk-os08a20-isp-it6263-lvds0.dtb
imx95-15x15-evk-os08a20-isp-adv7535.dtb
imx95-19x19-verdin-os08a20-isp-lt8912.dtb
X-MX95MBDES1003 imx95-19x19-evk-ox05b1s-isp-it6263-lvds0.dtb
imx95-15x15-evk-ox05b1s-isp-adv7535.dtb
imx95-19x19-verdin-ox05b1s-isp-lt8912.dtb
XRPI-CAM-MINISAS imx95-19x19-evk-adv7535-ap1302.dtb
imx95-15x15-verdin-adv7535-ap1302.dtb
imx95-19x19-verdin-adv7535-ap1302.dtb
XRPI-CAM imx95-15x15-evk-adv7535-ap1302-rpi.dtb

Usage examples are given in the subsequent sections, requiring camera-dependent environment variables.
Environment variables are as follows:
• W: sensor width
• H: sensor height
• RAW: sensor mbus code
Note: Only valid for raw camera sensors.
• CAMERAn: libcamera unique/persistent identifier for camera instance number n.
The following table lists the variables to be set for each setup.

Table 83. Variables for each setup

Camera Environment variables
X-MX95MBDESER01 W=1920
4 modules H=1280
RAW=SBGGR16
X-MX95MBCAM10001 CAMERA0="/base/soc/bus@42000000/i2c@42530000/max96724@27/i2c-mux/i2c@0/mx95mbcam@40"
CAMERA1="/base/soc/bus@42000000/i2c@42530000/max96724@27/i2c-mux/i2c@1/mx95mbcam@40"
CAMERA2="/base/soc/bus@42000000/i2c@42530000/max96724@27/i2c-mux/i2c@2/mx95mbcam@40"
CAMERA3="/base/soc/bus@42000000/i2c@42530000/max96724@27/i2c-mux/i2c@3/mx95mbcam@40"

X-IMX95-OS08A20 W=3840
H=2160
RAW=SBGGR12
CAMERA0="/base/soc/bus@42000000/i2c@42530000/os08a20_mipi@36"

X-MX95MBDES1003 W=2592
H=1944
CAMERA0="/base/soc/bus@42000000/i2c@42530000/ox05b1s@36"
RAW=SGRBG10

XRPI-CAM-MINISAS W=1280
XRPI-CAM H=800
CAMERA0="/base/soc/bus@42000000/i2c@42530000/ap1302_mipi@3c"

7.3.9.4 Libcamera configuration

Libcamera library embeds support for multiple pipelines handlers. Explicit pipeline match ordering is possible
using an environment variable. For instance, to select the NXP Neo ISP pipeline first, then the NXP ISI pipeline:

export LIBCAMERA_PIPELINES_MATCH_LIST='nxp/neo,imx8-isi'

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

66 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

To add some verbosity in the console log, for instance:

export LIBCAMERA_LOG_LEVELS='NxpNeo:DEBUG,ISI:DEBUG'

The environment variables supported by Libcamera are documented here.

7.3.9.5 GStreamer pipelines

This section lists the sample GStreamer pipelines relying on the libcamerasrc source element.
Command lines assume that the environment variables have been set beforehand for the targeted camera.

7.3.9.5.1 GStreamer pipeline single camera preview

This is a pipeline example to capture the frames from a single camera for display on the video sink:

DISPLAY_W=640
DISPLAY_H=480
gst-launch-1.0 \
libcamerasrc camera-name="${CAMERA0}" ! \
video/x-raw, format=YUY2 ! \
queue ! \
waylandsink window-width=${DISPLAY_W} window-height=${DISPLAY_H}

7.3.9.5.2 GStreamer multiple camera composition and preview

This is a pipeline example to capture the frames from 4 cameras, and display the composited output on the
video sink:

DISPLAY_W=640
DISPLAY_H=480
gst-launch-1.0 -v \
imxcompositor_g2d name=comp \
sink_0::xpos=0 sink_0::ypos=0 \
sink_0::width=${DISPLAY_W} sink_0::height=${DISPLAY_H} \
sink_1::xpos=0 sink_1::ypos=${DISPLAY_H} \
sink_1::width=${DISPLAY_W} sink_1::height=${DISPLAY_H} \
sink_2::xpos=${DISPLAY_W} sink_2::ypos=0 \
sink_2::width=${DISPLAY_W} sink_2::height=${DISPLAY_H} \
sink_3::xpos=${DISPLAY_W} sink_3::ypos=${DISPLAY_H} \
sink_3::width=${DISPLAY_W} sink_3::height=${DISPLAY_H} ! \
waylandsink \
libcamerasrc camera-name="${CAMERA0}" ! \
video/x-raw,format=YUY2 ! queue ! comp.sink_0 \
libcamerasrc camera-name="${CAMERA1}" ! \
video/x-raw,format=YUY2 ! queue ! comp.sink_1 \
libcamerasrc camera-name="${CAMERA2}" ! \
video/x-raw,format=YUY2 ! queue ! comp.sink_2 \
libcamerasrc camera-name="${CAMERA3}" ! \
video/x-raw,format=YUY2 ! queue ! comp.sink_3

7.3.9.5.3 GStreamer RGB/YUV and Infrared output captures (RGBIr sensor)

DISPLAY_W=640
DISPLAY_H=480
gst-launch-1.0 \
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

67 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

libcamerasrc camera-name="${CAMERA0}" name=src \

src.src ! video/x-raw, format=YUY2 ! queue ! \
waylandsink window-width=${DISPLAY_W} window-height=${DISPLAY_H} \
src.src_0 ! video/x-raw, format=GRAY8 ! queue ! imxvideoconvert_g2d ! \
waylandsink window-width=${DISPLAY_W} window-height=${DISPLAY_H}

7.3.9.5.4 Gstreamer single smart camera with 2 streams composition and preview

DISPLAY_W=640
DISPLAY_H=480
DISPLAY_W2=320
DISPLAY_H2=240
gst-launch-1.0 -v \
libcamerasrc camera-name="${CAMERA0}" name=src \
src.src ! \
video/x-raw,format=YUY2,width=1280,height=800 ! \
queue ! \
waylandsink window-width=${DISPLAY_W} window-height=${DISPLAY_H} \
src.src_0 ! \
video/x-raw,format=NV12,width=640,height=400 ! \
queue ! \
waylandsink window-width=${DISPLAY_W2} window-height=${DISPLAY_H2}

7.3.9.6 cam test application

Libcamera comes with the cam test application that uses the native libcamera API. When using the NXP
Neo pipeline handler, cam gives the option to capture the camera frame before (raw stream) or after the ISP
decoding. The frames captured can be either displayed through the drm/kms device (single camera only) or
stored to the file system.
To display the capture frames, the weston service shall be disabled first to release the drm/kms device, using
the command:

systemctl stop weston

In addition, writing frames to the file system generates a huge amount of data that is not absorbable in real
time. In that case, because of the delay induced on the application to recycle its buffers, periodic frame loss is
expected to occur.
Sample command lines in this section assume that the environment variables have been set beforehand for the
targeted camera.
1. List the detected camera.
cam -l
2. List the stream formats for a camera.
cam --camera 1 -I
3. Capture 1000 decoded frames and display over HDMI.
cam \
--camera 1 \
--stream width=${W},height=${H},pixelformat=YUYV \
--capture=1000 \
--display=HDMI-A-1

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

68 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

4. Capture 20 raw frames and save to the file.

cam \
--camera 1 \
--stream width=${W},height=${H},pixelformat=${RAW} \
--capture=20 \
--file=frame-#.raw
5. Capture simultaneously 20 raw and RGB888 decoded frames and save to the file.
cam \
--camera 1 \
--stream width=${W},height=${H},pixelformat=${RAW},role=raw \
--stream width=${W},height=${H},pixelformat=BGR888,role=video \
--capture=20 \
--file=frame-#.raw
Note: Cam pixel formats use the DRM convention: DRM BGR888 corresponds to V4L2 RGB888.
6. Capture 10 decoded frames and save to the file (multicamera case).
cam \
--camera 1 \
--stream width=${W},height=${H},pixelformat=YUYV \
--file=frame-#.yuv \
--capture=10 \
--camera 2 \
--stream width=${W},height=${H},pixelformat=YUYV \
--file=frame-#.yuv \
--capture=10 \
--camera 3 \
--stream width=${W},height=${H},pixelformat=YUYV \
--file=frame-#.yuv \
--capture=10 \
--camera 4 \
--stream width=${W},height=${H},pixelformat=YUYV \
--file=frame-#.yuv \
--capture=10
7. Capture RGB/YUV and Infrared frames and save to file (RGBIr sensor).
cam \
--camera 1 \
--stream width=${W},height=${H},pixelformat=YUYV,role=video \
--stream width=${W},height=${H},pixelformat=R8,role=video \
--capture=20 \
--file=frame-#.bin
8. Capture 2 streams with different formats and downscaling (smart sensors).
W2=$((W/2))
H2=$((H/2))
cam \
--camera 1 \
--stream width=${W},height=${H},pixelformat=YUYV \
--stream width=${W2},height=${H2},pixelformat=BGR888 \
--capture=20 \
--file=frame-#.bin

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

69 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

7.3.9.7 Limitations

7.3.9.7.1 Neo ISP pipeline handler limitations

The following limitations apply:
• ISP is operated in memory-to-memory mode only. This enables both single and multi-camera operations.
However, the streaming mode operation (direct connection from CSI to ISP), which is a system optimization
for a single camera, is not implemented.
• The Neo ISP has no scaler. Therefore, the stream configuration size must use one of the native sensor sizes
(modes). Stream role is only used to select raw frames capture.
• Pipeline handler restricts usage of the sensor size whose raw output format has the highest number of bits per
sample.
• Single-camera topologies are automatically detected. Multi-camera topologies typically involve intricate
graphs that currently require a platform configuration file (config.yaml). The mx95mbcam setup with 4
cameras can be used as an example.
• Control algorithms (IPA) have a minimal support limited to Automatic/Manual White Balancing and Automatic/
Manual Gain/Exposure Correction. There is no color correction, or other control algorithm enabled.

7.3.9.7.2 ISI pipeline handler limitations

The following limitations apply:
• Support is limited to single-camera instance, with up to 2 streams.
• Stream resolution is bounded to the sensor native resolution.

7.3.9.7.3 UVC (USB camera) pipeline handler limitations

The following limitations apply:
• Buffers output by the UVC driver cannot be processed by GStreamer waylandsink element. Element
glimagesink should be used as a replacement.

7.3.10 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.

7.3.11 Web camera

The following command line is an example of how to record and transfer web camera input.

$GSTL v4l2src device=/dev/video1 ! $video_encoder_plugin ! rtph264pay ! udpsink

host=$HOST_IP

HOST_IP is the IP/multicast group to send the packets to.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

70 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

This command line is an example of how to receive and display web camera input.

$GSTL udpsrc ! buffer-size=204800 (example number) application/x-rtp !

rtph264depay ! $video_decoder_plugin ! autovideosink

7.3.12 HTTP streaming

The HTTP streaming includes the following:
• Manual pipeline
$GSTL souphttpsrc location= http://SERVER/test.avi ! typefind
! aiurdemux name=demux demux. ! queue max-size-buffers=0 max-size-time=0
! $video_decoder_plugin ! $video_sink_plugin demux. ! queue max-size-
buffers=0 max-size-time=0
! beepdec ! $audio_sink_plugin
• PLAYBIN
$GSTL $PLAYBIN uri=http://SERVER/test.avi
• GPLAY
$GPLAY http://SERVER/test.avi

7.3.13 HTTP live streaming

The HLS streaming includes the following:
• PLAYBIN
$GSTL $PLAYBIN uri=http://SERVER/test.m3u8
• GPLAY
$GPLAY http://SERVER/test.m3u8

7.3.14 MPEG-DASH streaming

The MPEG-DASH streaming includes the following:
• PLAYBIN
$GSTL $PLAYBIN uri=http://SERVER/test.mpd
• GPLAY
$GPLAY http://SERVER/test.mpd

Note: Supports TS, MP4, and WebM container format segment currently.

7.3.15 Real Time Streaming Protocol (RTSP) playback

Use the following command to see the GStreamer RTP depacketize plugins:

$GSTINSPECT | grep depay

RTSP streams can be played with a manual pipeline or by using playbin. The format of the commands is as
follows.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

71 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• Manual pipeline
$GSTL rtspsrc location=$RTSP_URI name=source
! queue ! $video_rtp_depacketize_plugin ! $vpu_dec ! $video_sink_plugin
source.
! queue ! $audio_rtp_depacketize_plugin ! $audio_parse_plugin ! beepdec !
$audio_sink_plugin
• 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.
Manual pipeline example:

$GSTL rtspsrc location=rtsp://10.192.241.11:8554/test name=source

! queue ! rtph264depay ! avdec_h264 ! overlaysink source.
! queue ! rtpmp4gdepay ! aacparse ! beepdec ! pulsesink

Playback does not exit automatically in GStreamer 1.x, if buffer-mode is set to buffer in the rtpsrc plugin.

7.3.16 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:

$GSTL udpsrc do-timestamp=false uri=$UDP_URI caps="video/mpegts"

! aiurdemux streaming_latency=400 name=d d. ! queue ! $vpu_dec

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

72 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

! queue ! $video_render_sink sync=true d. ! queue ! beepdec !

$audio_sink_plugin sync=true

$GSTL udpsrc do-timestamp=false uri=udp://10.192.241.255:10000 caps="video/

mpegts"
! aiurdemux streaming_latency=400 name=d d. ! queue ! vpudec
! queue ! overlaysink sync=true d. ! queue ! beepdec ! pulsesink sync=true

The format for an RTP MPEGTS streaming command is covered as follows:

$GSTL udpsrc do-timestamp=false uri=$RTP_URI caps="application/x-rtp"

! rtpmp2tdepay ! aiurdemux streaming_latency=400 name=d d. ! queue !
$vpu_dec
! queue ! $video_render_sink sync=true d. ! queue ! beepdec !
$audio_sink_plugin sync=true

$GSTL udpsrc do-timestamp=false uri=udp://10.192.241.255:10000

caps="application/x-rtp"
! rtpmp2tdepay ! aiurdemux streaming_latency=400 name=d d.
! queue ! vpudec ! queue ! overlaysink sync=true d. ! queue ! beepdec
! pulsesink sync=true

7.3.17 RTSP streaming server

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:
$yocto_root/build/tmp/work/cortexa9hf-vfp-neon-poky-linux-gnueabi/
gstreamer1.0-rtsp-server/$version/
build/examples/
5. Flash the image.
Copy test-uri into /usr/bin in the rootfs on your board and assign execute permission to it.
Some information on running the tool is as follows:
• Command:
test-uri $RTSP_URI
For example:
test-uri file:///home/root/temp/TestSource/mp4/1.mp4

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

73 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• Server address:
rtsp://$SERVER_IP:8554/test
For example:
rtsp://10.192.241.106:8554/test
• Client operations supported are Play, Stop, Pause, Resume, and Seek.

7.3.18 Video conversion

There are four video conversion plugins, imxvideoconvert_ipu, imxvideoconvert_g2d,
imxvideoconvert_ocl, and imxvideoconvert_pxp. All of them, except for imxvideoconvert_ocl,
can be used to perform video color space conversion, resize, and rotate. imxvideoconvert_ipu can
also be used to perform video deinterlacing. imxvideoconvert_ocl can be used to perform video color
space conversion, downscale, color range conversion, and deinterlace. 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.
imxvideoconvert_ocl supports the following conversions:
• RGBA32 to NV12, supports output buffer of BT601 and BT709, limited and full range.
• RGBX32 to NV12, supports output buffer of BT601 and BT709, limited and full range.
• NV12 to NV12, supports input and output buffer of BT601 and BT709, limited and full range.
• NV12 to RGB24, supports input buffer of BT601 and BT709, limited and full range.
• YUYV to RGB24, supports input buffer of BT601 and BT709, limited and full range.
• RGBA32 to RGB24.
• NV12TILE to NV12, only for Amphion VPU decoder, including deinterlace, full range to limited range.
• NV15TILE to NV12, only for Amphion VPU decoder, including deinterlace, full range to limited range.
• imxvideoconvert_ocl also supports the downscale function. The input format can be NV12 YUY2 RGBA
I420 and the output format is fixed to RGB. Currently, the width and height of the video source need 16
aligned and the width and height of video output need 4 aligned and 2 aligned respectively.
• imxvideoconvert_ocl also support dewarp and dewarp downscale functions. For dewarp function, the
input format can be BGRx, RGBA, RGBx, BGR, RGB, NV12, and YUY2. The output format must be the same
as the input format. For dewarp downscale function, the input format can be BGRx, RGBA, RGBx, BGR,
RGB, NV12, and YUY2, and the output format is fixed to RGB.
• The autovideoconver plugin can make single or multiple video convert plugins to perform color space
conversion (CSC) according to the specific input and output video formats. It preferentially selects the
supported i.MX plugin according to the specific board. For some conversions, a single plugin may not support
and often need to combine multiple plugins to complete. It can combine those plugins automatically to achieve
the required function.
Use gst-inspect-1.0 to get each converter's capability and supported input and output formats. Note that
imxvideoconvert_g2d can only perform color space converting to RGB space.
Color Space Conversion (CSC)

gst-launch-1.0 videotestsrc ! video/x-raw,format=NV12 ! imxvideoconvert_{xxx} !

video/x-raw,format=RGB16 ! ximagesink display=:0

Color Space Conversion (CSC) with OpenCL-based Converter

gst-launch-1.0 videotestsrc ! video/x-raw,format=RGBA ! imxvideoconvert_ocl !

video/x-raw,format=NV12 ! waylandsink

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

74 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Color Range Conversion with OpenCL-based Converter

gst-launch-1.0 videotestsrc ! video/x-raw,format=NV12,colorimetry=1:3:0:0 !

imxvideoconvert_ocl ! video/x-raw,format=NV12,colorimetry=2:3:0:0 ! waylandsink

Downscale with OpenCL-based Converter

gst-launch-1.0 videotestsrc ! video/x-raw,format=NV12,width=640,height=480 !

imxvideoconvert_ocl ! video/x-raw,format=RGB,width=300,height=300 !
videoconvert ! waylandsink

Combine usage for G2D and OpenCL-based Converter

gst-launch-1.0 videotestsrc ! video/x-raw,format=NV12,width=1920,height=1080 !

imxvideoconvert_g2d ! video/x-raw,width=300,height=300 ! imxvideoconvert_ocl !
video/x-raw,format=RGB ! videoconvert ! waylandsink

Resize

gst-launch-1.0 videotestsrc ! video/x-raw,format=NV12,width=800,height=600 !

imxvideoconvert_{xxx} ! video/x-raw, width=640, height=480 ! ximagesink
display=:0

Rotate

gst-launch-1.0 videotestsrc ! imxvideoconvert_{xxx} rotation=2 ! ximagesink

display=:0

Image Dewarp with OpenCL-based Converter on i.MX 95

imxvideoconvert_ocl supports the dewarp and dewarp downscale function on the i.MX 95 board. It
supports RGB, NV12, and YUY2 format for input.
imxvideoconvert_ocl can probe and parse the header data automatically in the map file, which contains
the initialization parameters. You can enable video warp and configure the dewarp map file, which contains the
header data as follows. The map file is the same one used by g2d dewarp.
For the imxvideoconvert_ocl dewarp function, only absolute coordinate mode is supported. The map-
format parameter can only be configured to 0 (coordinate mode), the map file must contain the header data, and
the coordinate data must be the absolute coordinate. Each pixel has 32-bit coordinate data.
Dewarp example:

gst-launch-1.0 videotestsrc ! video/x-raw,format=NV12,width=1920,height=1080 !

imxvideoconvert_ocl video-warp-enable=true video-warp-
coordfile=dewarp_map_file ! video/x-raw,format=NV12,width=1920,height=1080 !
videoconvert ! waylandsink

Dewarp downscale example:

gst-launch-1.0 videotestsrc ! video/x-raw,format=NV12,width=1920,height=1080 !

imxvideoconvert_ocl video-warp-enable=true video-warp-
coordfile=dewarp_map_file ! video/x-raw,format=RGB,width=300,height=300 !
videoconvert ! waylandsink

Image Dewarp with DPU G2D on i.MX 95

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

75 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

imxvideoconvert_g2d supports the Dewarp function on the i.MX 95 board. It supports both the RGB
format and packed YUV422 format, such as YUY2. NV12 format dewarp is not support since there is hardware
limitation for DPU.
• imxvideoconvert_g2d can probe and parse the header data automatically in the map file, which contains
the initialization parameters. You can enable video warp and configure the Dewarp map file, which contains
the header data as follows. Example:
gst-launch-1.0 videotestsrc ! video/x-raw,format=RGBA,width=1920,height=1080 !
imxvideoconvert_g2d video-warp-enable=true video-warp-coord-
file=dewarp_map_file ! video/x-raw,format=RGBA,width=1920,height=1080 !
videoconvert ! waylandsink
• If the map file has no header data, which contains the initialization parameters, configure these parameters
manually as follows:
– The map-format parameter configures three Dewarp modes, which are 0 (coordinate mode), 1 (Delta
mode), and 2 (Delta increment mode).
– The width and height parameters configure the actual input video width and height information.
– The bpp parameter means bit per pixel and configures the memory layout of each x/y value in coordinate
buffers.
– The arb_start_x and arb_start_y parameters configure the start value of of each x/y, which are used
only in Delta mode or Delta increment mode.
– The arb_delta_xx, arb_delta_xy, arb_delta_yx, and arb_delta_yy parameters configure the
initial Delta value, which are used only in Delta increment mode. Example:
gst-launch-1.0 videotestsrc ! video/x-
raw,format=RGBA,width=1920,height=1080 !
imxvideoconvert_g2d video-warp-enable=true video-warp-coord-
file=dewarp_map_file
video-warp-extra-controls="c,map-format=2,width=1920,height=1080,bpp=8,
arb_start_x=0x286c,arb_start_y=0x16a6,arb_delta_xx=0x0e,arb_delta_xy=0xfc,
arb_delta_yx=0xfc,arb_delta_yy=0x14" !
video/x-raw,format=RGBA,width=1920,height=1080 ! videoconvert ! waylandsink

Color Space Conversion (CSC) with autovideoconvert

gst-launch-1.0 videotestsrc ! video/x-raw,format=RGBA ! autovideoconvert !

video/x-raw,format=YUY2 ! videoconvert ! waylandsink

Video crop with i.MX with G2D

gst-launch-1.0 videotestsrc ! videocrop top=10 bottom=10 right=10 left=10 !

imxvideoconvert_g2d videocrop-meta-enable=true ! queue ! ximagesink display=:0

Deinterlacing with i.MX with IPU

gst-launch-1.0 playbin uri=file://$FILE video-sink="imxvideoconvert_ipu

deinterlace=3 ! ximagesink display=:0 sync=false"

Transcoding with i.MX with VPU

gst-launch-1.0 filesrc location=$FILE.mp4 typefind=true ! video/

quicktime ! aiurdemux ! vpudec ! imxvideoconvert_ipu ! video/x-
raw,format=NV12,width=1280,height=720 ! vpuenc_h263 ! avimux ! filesink
location=$FILE.avi

Combination with i.MX with IPU or VPU

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

76 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

It is possible to combine CSC, resize, rotate, and deinterlace at one time. Both imxvideoconvert_ipu and
imxvideoconvert_g2d can be used at the same time in a pipeline. The following is an example:

gst-launch-1.0 videotestsrc ! video/x-

raw,format=I420,width=1280,height=800,interlace-mode=interleaved !
imxvideoconvert_ipu rotation=2 deinterlace=3 ! video/
x-raw,format=NV12,width=800,height=600 ! vpuenc_h264 !
vpudec ! imxvideoconvert_g2d rotation=3 ! video/x-
raw,format=RGB16,width=640,height=480 ! ximagesink sync=false display=:0

7.3.19 Video composition

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. imxcompositor_g2d can also
perform dewarp for each input stream on i.MX 95.
• Composite two videos into one.
gst-launch-1.0 imxcompositor_{xxx} name=comp sink_1::xpos=160
sink_1::ypos=120 ! overlaysink videotestsrc ! comp.sink_0 videotestsrc !
comp.sink_1
• Composite two videos into one with red background color.
gst-launch-1.0 imxcompositor_{xxx} background=0x000000FF name=comp
sink_1::xpos=160 sink_1::ypos=120 ! overlaysink videotestsrc ! comp.sink_0
videotestsrc ! comp.sink_1
• Composite two videos into one with CSC, resize, and rotate.
gst-launch-1.0 imxcompositor_{xxx} name=comp sink_0::width=640
sink_0::height=480
sink_1::xpos=160 sink_1::ypos=120 sink_1::width=640 sink_1::height=480
sink_1::rotate=1 !
video/x-raw,format=RGB16 ! overlaysink videotestsrc !
video/x-raw,format=NV12,width=320,height=240 ! comp.sink_0 videotestsrc !
video/x-raw,format=I420,width=320,height=240 ! comp.sink_1
• Composite three videos into one with CSC, resize, rotate, alpha, z-order, and keep aspect ratio.
gst-launch-1.0 imxcompositor_{xxx} name=comp sink_0::width=640
sink_0::height=480
sink_0::alpha=0.5 sink_0::z-order=3 sink_1::alpha=0.8 sink_1::z-order=2
sink_1::xpos=160
sink_1::ypos=120 sink_1::width=640 sink_1::height=480 sink_1::rotate=1
sink_2::xpos=320
sink_2::ypos=240 sink_2::width=500 sink_2::height=500 sink_2::alpha=0.6
sink_2::keep-ratio=true ! video/x-raw,format=RGB16 ! overlaysink
videotestsrc !
video/x-raw,format=NV12,width=320,height=240 ! comp.sink_0 videotestsrc !
video/x-raw,format=I420,width=320,height=240 ! comp.sink_1 videotestsrc !
video/x-raw,format=RGB16,width=320,height=240 ! comp.sink_2
• Composite four videos into one with CSC, resize, Alpha blend, and dewarp:
gst-launch-1.0 imxcompositor_g2d name=comp background=0x00FFFF00 \
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

77 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

sink_0::xpos=0 sink_0::ypos=0 sink_0::width=960 sink_0::height=524

sink_0::video-warp-enable=true sink_0::video-warp-coord-file=dewarp_map_file \
sink_1::xpos=40 sink_1::ypos=40 sink_1::width=960 sink_1::height=524
sink_1::alpha=0.5 sink_1::video-warp-enable=true sink_1::video-warp-coord-
file=dewarp_map_file \
sink_2::xpos=80 sink_2::ypos=80 sink_2::width=960 sink_2::height=524 \
sink_3::xpos=160 sink_3::ypos=160 sink_3::width=960 sink_3::height=524 !
waylandsink sync=false \
videotestsrc ! video/x-raw,format=YUY2,width=1920,height=1282 ! queue !
comp.sink_0 \
videotestsrc ! video/x-raw,format=YUY2,width=1920,height=1282 ! queue !
comp.sink_1 \
videotestsrc ! video/x-raw,format=YUY2,width=1920,height=1282 ! queue !
comp.sink_2 \
videotestsrc ! video/x-raw,format=YUY2,width=1920,height=1282 ! queue !
comp.sink_3

7.4 PipeWire input/output settings

This section provides the instructions and commands for PipeWire.

7.4.1 PipeWire settings

This section provides the steps for PipeWire settings.
Note: Make sure to start the PipeWire, WirePlumber, and PipeWire-pulse services at first:

systemctl --user --now enable pipewire wireplumber pipewire-pulse

1. Use the wpctl status command to list all the available audio sinks and sources:
$ wpctl status
A list of available audio sinks/sources is displayed:
Audio
├─ Devices:
│ 36. Built-in Audio [alsa]
│ 37. Built-in Audio [alsa]
│ 38. Built-in Audio [alsa]
│ 39. Built-in Audio [alsa]
│
├─ Sinks:
│ 40. Built-in Audio Mono [vol: 0.40]
│ * 42. Built-in Audio Analog Surround 5.1 [vol: 0.40]
│ 45. Built-in Audio Digital Stereo (IEC958) [vol: 0.40]
│
├─ Sink endpoints:
│
├─ Sources:
│ 41. Built-in Audio Mono [vol: 1.00]
│ * 43. Built-in Audio Analog Surround 5.1 [vol: 1.00]
│ 44. Built-in Audio Stereo [vol: 1.00]
│ 46. Built-in Audio Digital Stereo (IEC958) [vol: 1.00]
│
├─ Source endpoints:
│
└─ Streams:

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

78 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

2. Use wpctl inspect <object-id> to display the information about the specified object.
For example, wpctl inspect 36 can check this audio card information.
3. Use wpctl set-default <object-id> to set the default audio sink and source according to the ID
number of sinks or sources in the list shown above.
Multichannel output support settings
For those boards that need to output multiple channels, perform the following steps to enable the multichannel
output profile:
1. Use pw-cli info <object-id> to find <param-id> of EnumProfile.
* params: (4)
* 8 (Spa:Enum:ParamId:EnumProfile) r-
* 9 (Spa:Enum:ParamId:Profile) rw
* 12 (Spa:Enum:ParamId:EnumRoute) r-
* 13 (Spa:Enum:ParamId:Route) rw
Here, 8 is the <param-id> of EnumProfile.
2. Use pw-cli enum-params <object-id> <param-id> to check the profile index of the card.
Here, <object-id> is the ID number of audio devices shown by wpctl status.
...
Object: size 464, type Spa:Pod:Object:Param:Profile (262151), id
Spa:Enum:ParamId:EnumProfile (8)
Prop: key Spa:Pod:Object:Param:Profile:index (1), flags 00000000
Int 3
Prop: key Spa:Pod:Object:Param:Profile:name (2), flags 00000000
String "output:analog-surround-71+input:analog-surround-51"
Prop: key Spa:Pod:Object:Param:Profile:description (3), flags 00000000
String "Analog Surround 7.1 Output + Analog Surround 5.1 Input"
...
Here, 3 is the profile index.
3. Use wpctl set-profile <object-id> <profile index> to set the profile for particular features.
4. Use pw-play xx.wav to play a multichannel wave file.

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_ACCEPTED = "commercial"
2. Build gstreamer1.0-libav.
$ bitbake gstreamer1.0-libav
3. Build the rootfs image.
$ bitbake <image_name>

8 Audio

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

79 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

8.1 DSP support

DSP support is provided on specific i.MX 8QuadXPlus, i.MX 8QuadMax, i.MX 8M Plus, and i.MX 8ULP SoCs.

8.1.1 HiFi 4 DSP framework

Supporting HiFi 4 on a custom board is documented in the i.MX DSP User's Guide (UG10167).

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.
For details about toolchains, supported platforms, binary packaging, and quick setup of audio scenarios, see
SOF User Guide on NXP i.MX 8 platforms.

8.2 HDMI eARC support

eARC is supported on the i.MX 8M Plus EVK board.
The procedure enables audio, which is input through the imxaudioxcvr card and played back through the
wm8960audio card.
Make sure there is a headset plugged into the i.MX 8M Plus EVK audio jack. The procedure is as follows:
1. On i.MX 8M Plus EVK, set eARC mode. The default mode is SPDIF:
amixer -c 0 cset numid=1 2
2. On i.MX 8M Plus EVK, use alsamixer to set Headphone and Playback controls to the maximum values
for the wm8960-audio card.
3. On i.MX 8M Plus EVK, start audio recording on the imxaudioxcvr card and playback on the
wm8960audio card.
arecord -Dsysdefault:CARD=imxaudioxcvr -c2 -r48000 -fS32_LE -twav | aplay -
Dsysdefault:CARD=wm8960audio
4. Make sure Digital Audio Out from the TV is PCM. Then, set eARC mode to ON and check audio on
the headset connected to the i.MX 8M Plus EVK jack. The settings on the TV should be as shown in the
following figure.
Audio Output

Digital Audio Out PCM

eARC mode ON

Figure 5. eARC support

Note: The procedure has been tested on Sony and LG TV.

The firmware on the eARC RX (i.MX 8M Plus EVK) waits until HPD=high is sensed. Therefore, start recording
and playback on i.MX 8M Plus before enabling eARC mode on the TV as described in Steps 3 and 4 above.
This makes the behavior more predictable on the RX side. In addition, set the subsequent eARC mode to OFF
then to ON on the TV while keeping arecord … | aplay … running on i.MX 8M Plus EVK. Check whether
you can hear audio in the headset after subsequent eARC mode is set to ON on the TV.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

80 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Besides, enabling the complete eARC feature, per the HDMI 2.1 specification, is more of a system-level
application that integrates different layers and modules of the CEC driver, DRM, HDMI controller driver, EDID,
and eARC driver modules.

8.3 Low power voice UI

8.3.1 Introduction
The Cortex-M core on the i.MX 8M Plus, i.MX 8M Mini, i.MX 8ULP, i.MX 93, and i.MX 95 platforms can be used
in an AsymmetricMultiprocessing (AMP) architecture for a low power voice User Interface (UI) solution.
The voice activity detection and wake work engine shall use the lowest power core of the i.MX 8M, i.MX 93, and
i.MX 95, so that the Cortex-A cluster and related peripherals can remain in sleep mode for most of the “active
listening” time.
Upon successful detection of a wake word, the Cortex-M core shall wake up the Cortex-A domain for better
acoustic performance and further voice processing.
There are three components needed for this solution:
• Audio Front End (AFE)
• Linux drivers
• Cortex-M Image
Note: The i.MX 8ULP platform is added in the supported list. However, for i.MX 8ULP, there is no low-power
wake word detection on Cortex-M. The Cortex-M is only for providing the path of the PDM microphone. AFE
and Linux drivers are supported.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

81 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Components in NXP
Voice Enablement scope

i.MX
Voice Processing

SDK / FreeRTOS VoiceSeeker VIT Voice Command

AFE
Low Power Wake Word local
VoiceSpot WW buffer AFE Wrapper VoiceSpot WW

VoiceSeeker Audio
User space Playback

Linux kernel
Main app
RPMsg RPMsg ALSA

ALSA
Drivers buffer Drivers

SAI SAI Cortex- DDR Memory & HW

Mx Cortex-Ax Peripherals

PDM I2S

Figure 6. Low-power voice solution architecture

8.3.2 Standard voice solution

In addition to the low power voice UI solution described in the previous section, a standard voice solution is also
available. This solution does not leverage a Cortex-M image and therefore takes the microphone and speaker
inputs/outputs directly from the Linux kernel (through drivers and the ALSA library). It also leverages VIT wake
word detection engine (in place of VoiceSpot in the low power voice UI solution).

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

82 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Components in NXP
Voice Enablement scope

i.MX
Voice Processing

VoiceSeeker VIT Voice Command

AFE
AFE Wrapper VIT WW

Audio
Playback User space

Linux kernel

ALSA

Drivers

SAI SAI Memory & HW

Cortex-Ax
Peripherals

PDM I2S

Figure 7. Standard voice UI solution architecture

8.3.3 Audio Front End (AFE)

To precisely detect human language, the audio signal from the microphone must be clean, without echo, noise,
or other disturbances. To achieve this, a microphone array is typically used, with multiple, (usually) interleaved
microphone signals input to the embedded device. Such a compound signal is then fed into a signal processor
(commonly known as an Audio Front End), which filters out the noise, echo, and other disturbances. The output
from the signal processor is then the desired single channel, clean microphone audio, which is used for further
processing (wake-word detection and natural language processing).

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

83 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

To interface audio with the Linux OS, the Advanced Linux Sound Architecture (ALSA) library is used. The
following figure shows the audio architecture.
ALSA logical
ALSA physical (loopback)
streams streams

Audio front end Filtered

Physical source (AFE) / signal Virtual source microphone
processor samples

Reference
(playback)
samples

Playback
Physical sink Virtual sink
samples

aaa-052980

Figure 8. Audio front end architecture

The AFE code is on GitHub: https://github.com/nxp-imx/nxp-afe/. Use the Git tag corresponding to the current
Linux BSP release.
The AFE deliveries are in Yocto rootfs/unit_tests/nxp-afe/ and /usr/lib/nxp-afe/.
• /unit_tests/nxp-afe/afe: The main application of the AFE
• /unit_tests/nxp-afe/TODO.md: User guide document
• /unit_tests/nxp-afe/asound.conf_imx8mm: Used for i.MX 8M Mini default and RPMsg dtb
• /unit_tests/nxp-afe/asound.conf_imx8mp: Used for i.MX 8M Plus default and RPMsg dtb
• /unit_tests/nxp-afe/asound.conf_imx8mp_revb4: Used for i.MX 8M Plus EVKB4 default and
RPMsg dtb
• /unit_tests/nxp-afe/asound.conf_imx93: Used for i.MX 93 default and RPMsg dtb for both EVK and
QSB boards
• /unit_tests/nxp-afe/asound.conf_imx95: Used for i.MX 95 EVK default and RPMsg dtb
In addition to the AFE, the Yocto BSP integrates VoiceSeeker (a multi-microphone voice control audio front-
end signal processing solution), VoiceSpot (a small memory and MIPS profile wake word engine supporting the
"Hey NXP" voice trigger word), and VIT for local voice command recognition. These deliveries are available on
GitHub: https://github.com/nxp-imx/imx-voiceui. This contains:
• VoiceSeeker_wrapper folder contains the code used for generating the shared library used by the AFE
wrapper. Check the TODO.md in the nxp-afe repository.
• Voice_UI_Test_app folder contains the code used for generating the voice UI application called
"voice_ui_app". This application contains the VoiceSpot wake-word engine and VIT for local voice
command recognition. This application uses the output of the AFE for voice detection (wake-word and voice
commands).
See the README file in this GitHub repository, which explains how to build the VoiceSeeker library and the
Voice UI Test application. Once these are built, the user shall copy the following files to Yocto rootfs:
• Copy release/Config.ini to /unit_tests/nxp-afe/.
• Copy release/HeyNXP_1_params.bin to /unit_tests/nxp-afe/.
• Copy release/HeyNXP_en-US_1.bin to /unit_tests/nxp-afe/.
• Copy release/libvoiceseekerlight.so.2.0 to /usr/lib/nxp-afe/libvoiceseekerlight.so.
2.0, symbol link libvoiceseekerlight.so to libvoiceseekerlight.so.2.0.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

84 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• Copy release/voice_ui_app to /unit_tests/nxp-afe/.

root@imx8mpevk:/unit_tests/nxp-afe# ls -l
-rw-r--r-- 1 weston weston 294 Mar 9 2018 Config.ini
-rw-r--r-- 1 weston weston 56 Mar 9 2018 HeyNXP_1_params.bin
-rw-r--r-- 1 weston weston 32812 Mar 9 2018 HeyNXP_en-US_1.bin
-rw-r--r-- 1 weston weston 4130 Mar 9 2018 TODO.md
-rwxr-xr-x 1 weston weston 67752 Mar 9 2018 afe
-rw-r--r-- 1 weston weston 1642 Mar 9 2018 asound.conf_imx8mm
-rw-r--r-- 1 weston weston 1661 Mar 9 2018 asound.conf_imx8mp
-rw-r--r-- 1 weston weston 1660 Mar 9 2018 asound.conf_imx8mp_revb4
-rw-r--r-- 1 weston weston 1642 Mar 9 2018 asound.conf_imx93
-rwxr-xr-x 1 weston weston 3414200 Mar 9 2018 voice_ui_app

root@imx8mpevk:/usr/lib/nxp-afe# ls -l
lrwxrwxrwx 1 weston weston 19 Mar 9 2018 libdummyimpl.so ->
libdummyimpl.so.1.0
-rw-r--r-- 1 weston weston 67576 Mar 9 2018 libdummyimpl.so.1.0
lrwxrwxrwx 1 weston weston 26 Mar 9 2018 libvoiceseekerlight.so ->
libvoiceseekerlight.so.2.0
-rw-r--r-- 1 weston weston 331104 Mar 9 2018 libvoiceseekerlight.so.2.0

After this, follow the steps in /unit_tests/nxp-afe/TODO.md to perform a test. The typical test method is
as follows:
• ./voice_ui_app &
• ./afe libvoiceseekerlight &
• aplay test.wav &
• arecord -d10 -fS32_LE -r16000 -c1 voiceseeker_afe_on.wav
The voice_ui_app binary enables the following VIT commands:
• MUTE
• NEXT
• SKIP
• PAIR DEVICE
• PAUSE
• STOP
• POWER OFF
• POWER ON
• PLAY MUSIC
• PLAY GAME
• WATCH CARTOON
• WATCH MOVIE
When users say "Hey NXP, power on", the "Hey NXP" wakes up the system, and the "power on" command is
detected.

trigger = 2, trigger_sample = 0, start_sample = -12812, stop_sample = -812,

score = 28
keyword_start_offset_samples = 12812
ITER = 633
- Voice Command detected 8 POWER ON

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

85 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

There is a configuration file called Config.ini through which user can choose the wake-word engine, select
the VIT language or implement other settings. It is a part of the standard voice solution.

$ cat /unit_tests/nxp-afe/Config.ini
[AFEConfig]
WWDectionDisable = 0
WakeWordEngine = VoiceSpot
DebugEnable = 0
# Declare mic coordinates if not using default value for supported i.MX EVKs
#mic0 = 35.0, 15.15, 0.0
#mic1 = 17.5, -15.15, 0.0
#mic2 = -17.5, -15.15, 0.0
#mic3 = -35.0, 15.15, 0.0
VoiceSpotModel = HeyNXP_en-US_1.bin
VoiceSpotParams = HeyNXP_1_params.bin
VITLanguage = English

• WWDectionDisable
Disables/Enables the wake-word and command detection.
– 0 - By default, enables the wake-word and command detection.
– 1 - Disables wake-word and command detection. voice_ui_app does not work.
• WakeWordEngine
This configuration depends on setting WWDectionDisable to 0. Selects if the voice_ui_app uses
VoiceSpot or VIT for wake-word detection.
– VoiceSpot - By default, use VoiceSpot to detect the wake-word.
– VIT - Use VIT to detect wake-word.
• DebugEnable
– 0 - By default, no debug recordings are made.
– 1 - Enables recording the AFE input/output streams for debugging and tuning the RefSignalDelay.
• RefSignalDelay
Used to calibrate the reference signal delay when using VoiceSeeker’s Acoustic Echo Cancellation (AEC).
The AEC enabled library is delivered with controlled access through Flexera. Please contact [email protected]
for more information.
• Mic coordinates
XYZ coordinates of the microphones in millimeters. The origin of the coordinate axis can be chosen as
convenient. Users can set custom values if not using the default value for the supported i.MX EVKs.
• VoiceSpotModel/VoiceSpotParams
The two parameters depend on setting WakeWordEngine to "VoiceSpot". They are used to specify the wake-
word model and parameters used by VoiceSpot.
• VITLanguage
This configuration depends on setting WakeWordEngine to "VIT". Selects the VIT language used to detect
the wake-word and commands.
– English - by default, uses English.
– Mandarin - uses Mandarin.
– Spanish - uses Spanish.
– German - uses German.
– Japanese - uses Japanese.
– Korean - uses Korean.
– Turkish - uses Turkish.
– Italian - uses Italian.
– French - uses French.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

86 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Notes on VIT model:

– VIT Wake-word and commands can be updated with generating new VIT Model thanks to the VIT Model
generation online tool.
– To have the new VIT Model considered by the application, the VIT model has be updated in ./vit/
i.MX8M_A53/Lib/ or /vit/i.MX9X_A55/Lib/ and Voice_ui_app recompiled.
– Same VIT model is used by the Voice_ui_app in low power or standard configuration. In the low power
configuration, since VIT is used in voice commands mode only, VIT wake-word information is not considered
by the VIT engine.
Note:
• For i.MX 93, the low-power voice UI solution uses digital microphones and codec already mounted on the
EVK. The add-on Audio hat is not supported.
• In this release, the VoiceSpot library is released by Flexera. Contact the NXP marketing to get it if you need
it. In default release, VIT is used for wakeword detection. This change is applied for all platforms that support
AFE.

8.3.4 Linux drivers

The primary Linux drivers used by the voice solutions are as follows:
• Loopback sound card: sound/drivers/aloop.c
• RPMsg sound card:
– sound/soc/fsl/fsl_rpmsg.c
– sound/soc/fsl/imx-pcm-rpmsg.c
– sound/soc/fsl/imx-rpmsg.c

8.3.5 Cortex-M Image

8.3.5.1 Application name

The Cortex-M application low_power_wakeword uses VoiceSeeker and VoiceSpot. When suspended, the Linux
OS is resumed only when the “Hey NXP” wake-word is recognized.
This application is provided for i.MX 8M Mini, i.MX 8M Plus, i.MX 93, and i.MX 95:
• imx8mm_m4_TCM_low_power_wakeword.bin
• imx8mp_m7_TCM_low_power_wakeword.bin
• imx93-11x11-evk_m33_TCM_low_power_wakeword.bin
• imx93-9x9-qsb_m33_TCM_low_power_wakeword.bin
• imx95-19x19-evk_m7_TCM_low_power_wakeword_sm_cm7.bin

8.3.5.2 Board setup

To set up the board, perform the following steps:
1. Ensure the appropriate Cortex-M application is copied to the “boot” partition of your SD Card. The
low_power_wakeword application should already exist from the public Yocto builds.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

87 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Note: For i.MX 95, use a system manager image built with the imx95alt configuration, or any similar
configuration that gives the necessary permission to the Cortex-M core to run this application.
2. Boot the board, stop in U-Boot prompt, and run the commands below:
a. Chose the appropriate device tree:
• i.MX 8M Mini:
setenv fdtfile imx8mm-evk-rpmsg-wm8524-lpv.dtb
• i.MX 8M Plus:
setenv fdtfile imx8mp-evk-rpmsg-lpv.dtb
• i.MX 93 EVK:
setenv fdtfile imx93-11x11-evk-rpmsg-lpv.dtb
• i.MX 93 QSB:
setenv fdtfile imx93-9x9-qsb-rpmsg-lpv.dtb
• i.MX 95 EVK:
setenv fdtfile imx95-19x19-evk-rpmsg-lpv.dtb
b. Load the Cortex-M image from Flash to TCM and boot the core before booting the Linux OS.
• i.MX 8M Mini and i.MX 8M Plus:
setenv lpv 'fatload mmc 1:1 0x48000000 <application_name>; cp.b
0x48000000 0x7e0000 0x40000; bootaux 0x7e0000;'
setenv bootcmd 'run prepare_mcore;run lpv;'${bootcmd}
• i.MX 93 EVK and QSB:
setenv lpv 'fatload mmc 1:1 0x80000000 <application_name>; cp.b
0x80000000
0x201e0000 0x20000; bootaux 0x1ffe0000 0;'
setenv bootcmd 'run lpv;'${bootcmd}
setenv mmcargs 'setenv bootargs ${jh_clk} console=${console} root=
${mmcroot} clk-imx93.mcore_booted snd_pcm.max_alloc_per_card=134217728'
• i.MX 95 EVK:
setenv lpv 'fatload mmc 1:1 0x90400000 <application_name>; cp.b
0x90400000 0x203c0000 0x40000; bootaux 0x00000000 1;'
setenv bootcmd 'run lpv;'${bootcmd}
setenv mmcargs 'setenv bootargs ${jh_clk} console=${console} root=
${mmcroot} clk-imx95.mcore_booted snd_pcm.max_alloc_per_card=134217728
pd_ignore_unused'
c. Save the changes above.
saveenv
3. Reboot the board, and the Cortex-M will be automatically started before the Linux OS. This can be checked
on the Cortex-M console.
4. Apply the appropriate ALSA configuration. After the Linux OS has booted, from the console:
• i.MX 8M Mini:
cp /unit_tests/nxp-afe/asound.conf_imx8mm /etc/asound.conf
• i.MX 8M Plus:
cp /unit_tests/nxp-afe/asound.conf_imx8mp /etc/asound.conf

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

88 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• i.MX 93 EVK and QSB:

cp /unit_tests/nxp-afe/asound.conf_imx93 /etc/asound.conf
• i.MX 95:
cp /unit_tests/nxp-afe/asound.conf_imx95 /etc/asound.conf
• Reboot to apply the changes above.
5. Before starting any audio application, e.g., arecord, aplay, afe, load the audio loopback driver. From a
Linux OS console:
modprobe snd-aloop

8.3.5.3 Execution
Once started, users have no direct actions to control the Cortex-M application. It automatically executes
appropriate actions according to the Linux state:
• When Linux OS is active and an application is recording audio: The Cortex-M application is acting as a data
pump, getting audio data from the microphones and providing them to ALSA drivers through RPMsg.
• When Linux OS is suspended while an application is recording audio: Audio data are processed locally on
Cortex-M (by VoiceSeeker/VoiceSpot). The data is also stored in a ring-buffer. Once voice or the wake-word
is detected (depending on the application), the Linux OS is automatically resumed, data from the ring buffer is
sent to ALSA (so the Linux OS also gets the wakeword), and then the data-pump is re-started.
• Suspending Linux OS: Cortex-M only has the possibility to resume the Linux OS, not to suspend it. Instead,
the Linux OS should be suspended by user-space action (the decision to suspend cannot be based only
on voice. It should also consider all the other potential user applications running on the Linux OS). For test
purposes, this can be forced by the user by entering the following command to the Linux console.
echo mem > /sys/power/state

8.3.6 Power consumption notes

• For i.MX 8M Plus, i.MX 8M Mini, i.MX 93 QSB, and i.MX 95 EVK:
These applications using the public Yocto release demonstrate the voice UI mechanism described above to
suspend the Linux OS and wake up on voice activity or a wakeword, but they still have a much higher power
consumption than expected.
Some changes are required in both the Cortex-M application and the Yocto image to achieve as low as
possible power consumption during the “Linux suspended” state. They are delivered with controlled access
through Flexera. For more information, contact your NXP representative.
• For i.MX 93 EVK:
The plublic Yocto release already contains power-optimized applications demonstrating very low power
consumption in "Linux suspended" state.

8.3.7 Using Linux remoteproc (i.MX 93 and i.MX 95)

On i.MX 93, the low_power_wakeword application can be started on the Cortex-M core by using the Linux
remotreproc driver (see Section 4.7.4). In this use case, the board setup described in Section 8.3.5.2 is the
same, except for Step b "Load Cortex-M image" that should be skipped.
Then, when Linux OS is running, users can load and start the low_power_wakeword application on Cortex-M
with the following commands:

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

89 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• i.MX 93 EVK:
echo imx93-11x11-evk_m33_TCM_low_power_wakeword.elf > /sys/class/remoteproc/
remoteproc0/firmware
echo start > /sys/class/remoteproc/remoteproc0/state
• i.MX 93 QSB:
echo imx93-9x9-qsb_m33_TCM_low_power_wakeword.elf > /sys/class/remoteproc/
remoteproc0/firmware
echo start > /sys/class/remoteproc/remoteproc0/state
• i.MX 95 EVK:
echo imx95-19x19-evk_m7_TCM_low_power_wakeword_sm_cm7.elf > /sys/class/
remoteproc/
remoteproc1/firmware
echo start > /sys/class/remoteproc/remoteproc1/state

Once started, the usage is the same as when the application is started by U-Boot.
If needed, users can also stop and restart remoteproc with the following commands:

echo stop > /sys/class/remoteproc/remoteproc0/state

echo start > /sys/class/remoteproc/remoteproc0/state

Note: Replace remoteproc0 by remoteproc1 in the commands above for i.MX 95.

8.4 Conversa Integration

Conversa is a comprehensive full-duplex, hands-free telephony speech processing suite providing total control
of uplink and downlink audio. Conversa-embedded multimicrophone, full-duplex voice processing delivers the
ultimate in sound quality for speakerphone and headset applications.
The Conversa processing suite is integrated to the i.MX 8M Plus using the NXP Audio Front End (AFE)
wrapper. The i.MX 8M Plus works as an external USB speaker and microphone with voice processing capability.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

90 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

NXP Linux BSP

User space

USB Gadget
NXP Audio Front End
composite
Conversa Tuning
thread

VCOM UAC 2.0 TX RX

Linux kernel
ALSA

Drivers

SAI SAI
USB
Memory & HW
Cortex-Ax Peripherals

PDM I2S

Figure 9. Conversa Integration

Note: The Conversa processing suite is a restricted-access solution. Contact your NXP representative for more
details.

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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

91 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 i.MX Graphics User's
Guide (UG10159). This SDK is only supported for hardware that has OpenGLES hardware acceleration.

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.
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 (RN00210) 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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

92 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• 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.
– – 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
• 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.

9.4 Qt 6
Qt 6 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 (UG10164).

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).

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

93 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.

Linux user space

Cryptodev-Linux
AF_ALGAPI /dev/hwrng
API
Linux kernel
Cryptodev-Linux af_alg

Crypto-API Power Shared

management operations

caamhash caamalg
Descripter HW desc
lib (CCSR regs)
sha 3des/des
caampkc caamrng
Descripter Init
md5 aes rsa hwrng construct lib functions

caam_ jr ctrl

JR Rl

CAAM
i.MX
aaa-052988

Figure 10. 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 84. Source files

Source File Description Module name
ctrl.[c,h] Init (global settings, RNG, power management, etc.) caam
desc.h HW description (CCSR registers, etc.) N/A

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

94 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 84. Source files...continued

Source File Description Module name
desc_constr.h Inline append - descriptor construction library N/A
caamalg_desc.[c,h] (Shared) Descriptors library (symmetric encryption, AEAD) caamalg_desc
caamhash_desc.[c,h] (Shared) Descriptors library (HASH) caamhash_desc
caamrng.c RNG (runtime) N/A
caamkeyblob_desc. Descriptors library (black keys and blobs) caamkeyblob_desc
[c,h]
jr.[c,h] JRI backend caam_jr
caamalg.c JRI frontend (symmetric encryption, AEAD) N/A
caamhash.c JRI frontend (hashing) N/A
caampkc.c, pkc_desc.c JRI frontend (public key cryptography) N/A
caamkeyblob.[c,h] JRI frontend (black keys and blobs) N/A
caamkeygen.c IOCTL calls for key and blob generation/import N/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 Section 10.1.2 for the list of module names and Section Section 10.1.4 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.

Table 85. Kernel configuration tree view

Kernel configuration tree view option Description
Enable CAAM device driver:
---Cryptographic API --->
[*] Hardware crypto devices ---> • Basic platform driver: Freescale CAAM-
<*>CAAM/SNVS Security Violation Handler Multicore platform driver backend
(EXPERIMENTAL) • Backends/interfaces: Freescale CAAM Job
<*>Freescale CAAM-Multicore platform driver Ring driver backend - JRI
backend • Frontends/crypto algorithms: symmetric
[ ] Enable debug output in CAAM driver encryption, AEAD, "stitched" AEAD; Register
<*> Freescale CAAM Job Ring driver backend algorithm implementations with the Crypto
---> API - via JRI (caamalg driver)
(9) Job Ring size • Register hash algorithm implementations
[ ] Job Ring interrupt coalescing with Crypto API - hashing (only via JRI -
[*] Register algorithm implementations caamhash driver)
with the Crypto API • Register public key cryptography
[*] Register hash algorithm implementations with Crypto API -
implementations with Crypto API asymmetric/public key (only via JRI -
[*] Register public key cryptography caampkc driver)
implementations with Crypto API
[*] Register caam device for hwrng API • Register CAAM device for hwrng API - HW
[*] Register tagged key cryptography RNG (only via JRI - caamrng driver)
implementations with Crypto API
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

95 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 85. Kernel configuration tree view

Kernel configuration tree view option Description
[ ] Test caam rng • Register algorithms supporting tagged key
[*] CAAM Secure Memory / Keystore API and generate black keys and encapsulate
(EXPERIMENTAL) them into black blobs
(7) Size of each keystore slot in
Secure Memory
<M> CAAM Secure Memory - Keystore Test/
Example (EXPERIMENTAL)
<M> Freescale Job Ring UIO support

Table 86. Device tree binding

Property Type Status Description
compatible String Required fsl, 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 87. Kernel configuration

Kernel configuration Description
Deselect the feature that bypass crypto driver validation. By
--- Cryptographic API ---> default, Linux kernel is bypassing crypto driver validation.
[ ] Disable run-time self tests Disable run-time self tests that normally take place at
[ ] Enable extra run-time crypto algorithm registration.
self tests
<M> Testing module 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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

96 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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-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)

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

97 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

[ 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
[ 5.157622] caam_jr 31430000.jr2: registering rng-caam
[ 5.206167] caam 31400000.caam: caam pkc algorithms registered in /proc/
crypto

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 a head 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}
– "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 together with pkcs1pad (rsa-caam,
sha256) driver.
• Algorithms supported by CAAM drivers
root@imx8mqevk:~# cat /proc/crypto | grep caam driver : pkcs1pad(rsa-
caam,sha256) driver : rsa-caam driver : cmac-aes-caam driver : xcbc-aes-caam
driver : md5-caam driver : hmac-md5-caam driver : sha256-caam driver : hmac-
sha256-caam driver : sha224-caam driver : hmac-sha224-caam driver : sha1-
caam driver : hmac-sha1-caam driver : seqiv-authenc-hmac-sha256-rfc3686-ctr-
aes-caam driver : authenc-hmac-sha256-rfc3686-ctr-aes-caam driver : seqiv-
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

98 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

authenc-hmac-sha224-rfc3686-ctr-aes-caam driver : authenc-hmac-sha224-rfc3686-

ctr-aes-caam driver : seqiv-authenc-hmac-sha1-rfc3686-ctr-aes-caam driver :
authenc-hmac-sha1-rfc3686-ctr-aes-caam driver : seqiv-authenc-hmac-md5-
rfc3686-ctr-aes-caam driver : authenc-hmac-md5-rfc3686-ctr-aes-caam driver :
echainiv-authenc-hmac-sha256-cbc-des-caam driver : authenc-hmac-sha256-cbc-
des-caam driver : echainiv-authenc-hmac-sha224-cbc-des-caam driver : authenc-
hmac-sha224-cbc-des-caam driver : echainiv-authenc-hmac-sha1-cbc-des-caam
driver : authenc-hmac-sha1-cbc-des-caam driver : echainiv-authenc-hmac-md5-
cbc-des-caam driver : authenc-hmac-md5-cbc-des-caam driver : echainiv-authenc-
hmac-sha256-cbc-des3_ede-caam driver : authenc-hmac-sha256-cbc-des3_ede-caam
driver : echainiv-authenc-hmac-sha224-cbc-des3_ede-caam driver : authenc-hmac-
sha224-cbc-des3_ede-caam driver : echainiv-authenc-hmac-sha1-cbc-des3_ede-caam
driver : authenc-hmac-sha1-cbc-des3_ede-caam driver : echainiv-authenc-hmac-
md5-cbc-des3_ede-caam driver : authenc-hmac-md5-cbc-des3_ede-caam driver :
echainiv-authenc-hmac-sha256-cbc-aes-caam driver : authenc-hmac-sha256-cbc-
aes-caam driver : echainiv-authenc-hmac-sha224-cbc-aes-caam driver : authenc-
hmac-sha224-cbc-aes-caam driver : echainiv-authenc-hmac-sha1-cbc-aes-caam
driver : authenc-hmac-sha1-cbc-aes-caam driver : echainiv-authenc-hmac-md5-
cbc-aes-caam driver : authenc-hmac-md5-cbc-aes-caam driver : authenc-hmac-
sha256-ecb-cipher_null-caam driver : authenc-hmac-sha224-ecb-cipher_null-caam
driver : authenc-hmac-sha1-ecb-cipher_null-caam driver : authenc-hmac-md5-ecb-
cipher_null-caam driver : gcm-aes-caam driver : rfc4543-gcm-aes-caam driver :
rfc4106-gcm-aes-caam driver : ecb-arc4-caam driver : ecb-des3-caam driver :
tk-ecb-aes-caam driver : ecb-aes-caam driver : ecb-des-caam driver : rfc3686-
ctr-aes-caam driver : ctr-aes-caam driver : cbc-des-caam driver : cbc-3des-caam
driver : tk-cbc-aes-caam driver : cbc-aes-caam root@imx8mqevk:~#

10.3 CAAM Job Ring backend driver specifications

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:
• 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.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

99 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 frontend (public key / asymmetric algorithms) can be turned off too, if needed.
caamrng 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)

[ 1.837113] caam 30900000.crypto: job rings = 2, qi = 0
[ 1.849949] caam algorithms registered in /proc/crypto
[ 1.855972] caam 30900000.crypto: caam pkc algorithms registered in /proc/
crypto
[ 1.865564] caam_jr 30901000.jr: registering rng-caam
[ 1.870766] Device caam-keygen registered

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
priority : 3000
refcnt : 1
selftest : passed
internal : no
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

100 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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 commonly 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:
• SSL protocol library
• SSL protocol library Crypto library (Symmetric and Asymmetric cipher support, digest support, etc.)
• Certificate Management
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.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

101 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Linux user space

Customer
Apache Nginx
applications

OpenSSL®

SSLAPI

Handshake Record layer RNG

OpenSSL® Lib-crypto/EVP API

Cyptodev-Linux API /dev/hwrng

Cyptodev-Linux

Cyptodev-API Linux kernel

Caamalg

Digest Symmetric asym asym rng

caam_ jr

JR

CAAM
i.MX
aaa-052990

Figure 11. 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.

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) through ioctls, offloading cryptographic operations in the kernel

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

102 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• cryptodev-linux (kernel space): Linux module that translates ioctl requests from cryptodev-engine into
calls to Linux Crypto API
• AF_ALG is a netlink-based in the kernel asynchronous interface that adds an AF_ALG address family
introduced in 2.6.38.
• 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:
• Symmetric Ciphering operations - AES (CBC, ECB), 3DES (CBC, ECB)
• Digest Operations - SHA (1, 256, 384, 512), MD5
• Public Key Operations - RSA Sign (1k, 2k, 4k) / RSA Verify (1k, 2k, 4k)

10.4.4 Deploying OpenSSL into rootfs

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:

CORE_IMAGE_EXTRA_INSTALL+="cryptodev-module openssl-bin"

Restart the build procedure:

bitbake imx-image-full

10.4.5 Running OpenSSL benchmarking tests with cryptodev engine

Probe the cryptodev-module:

root@imx8qxpmek:~# modprobe cryptodev

[17044.896494] cryptodev: driver 1.10 loaded.
root@imx8qxpmek:~# openssl engine
(devcrypto) /dev/crypto engine
(dynamic) Dynamic engine loading support
root@imx8qxpmek:~#

Note:
Starting from OpenSSL 1.1.1, the cryptodev engine is invoked by OpenSSL by default if the corresponding
module has been inserted in the kernel. Thus to perform only SW benchmark test using OpenSSL, remove the
cryptodev module by running rmmod cryptodev.

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:

root@imx8qxpmek:~# openssl speed -engine devcrypto -multi 8 -elapsed -evp aes-128-cbc

Forked child 1
engine "devcrypto" set.
Forked child 2
engine "devcrypto" set.
...
Got: +F:22:aes-128-cbc:378616.72:1611328.00:5084501.33:13994666.67:10731793.98:16219060.40 from 6
Got: +H:16:64:256:1024:8192:16384 from 7
Got: +F:22:aes-128-cbc:120773.33:9344.00:3088298.67:13588480.00:31642965.33:16471967.79 from 7
OpenSSL 1.1.1b 26 Feb 2019
built on: Thu Nov 14 13:22:07 2019 UTC
options:bn(64,64) rc4(char) des(int) aes(partial) idea(int) blowfish(ptr)

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

103 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

compiler: aarch64-poky-linux-gcc --sysroot=recipe-sysroot -O2 -pipe -g -feliminate-unused-debug-

types -fmacro-prefix-map=
-fdebug-prefix-map= -fdebug-prefix-map= -fdebug-prefix-map= -DOPENSSL_USE_NODELETE -DOPENSSL_PIC -
DOPENSSL_CPUID_OBJ -DOPENSSL_BN_ASM_MONT
-DSHA1_ASM -DSHA256_ASM -DSHA512_ASM -DKECCAK1600_ASM -DVPAES_ASM -DECP_NISTZ256_ASM -DPOLY1305_ASM
-DNDEBUG
evp 2242.05k 9681.05k 35017.46k 106866.86k 127787.74k 130077.23k
root@imx8qxpmek:~#

Additional ciphers that could be benchmarked: aes-192-cbc, aes-256-cbc, aes-128-ecb, aes-192-ecb,

aes-256-ecb, aes-128-ctr, aes-192-ctr, aes-256-ctr, des-cbc, des-cbc, des-ede3-cbc.
Additional digests that could be benchmarked: sha1, sha224, sha256, sha384, sha512, md5.

10.4.6 Running OpenSSL benchmarking tests with AF_ALG engine

Execute the following commands:

Probe the af_alg:

root@imx8mmevk:~# rmmod cryptodev
root@imx8mmevk:~# modprobe af_alg
root@imx8mmevk:~# modprobe algif_hash
root@imx8mmevk:~# modprobe algif_skcipher
root@imx8mmevk:~# modprobe algif_rng
root@imx8mmevk:~# modprobe algif_aead

10.4.6.1 Running OpenSSL benchmarking tests for symmetric ciphering and digest
Execute the following command:

root@imx8mmevk:~# openssl speed -engine afalg -multi 8 -elapsed -evp aes-128-cbc

Forked child 0
Forked child 1
engine "afalg" set.
+DT:aes-128-cbc:3:16
engine "afalg" set.
engine "afalg" set.
engine "afalg" set.
...
Got: +H:16:64:256:1024:8192:16384 from 0
Got: +F:22:aes-128-cbc:333888.00:1359317.33:4248405.33:5720064.00:6160384.00:6176768.00 from 0
Got: +H:16:64:256:1024:8192:16384 from 1
Got: +F:22:aes-128-cbc:378336.00:1382826.67:5117269.33:5739178.67:6190421.33:6176768.00 from 1
...
OpenSSL 1.1.1k 25 Mar 2021
built on: Thu Mar 25 13:28:38 2021 UTC
options:bn(64,64) rc4(char) des(int) aes(partial) blowfish(ptr)
compiler: aarch64-poky-linux-gcc -mcpu=cortex-a53 -march=armv8-a+crc+crypto -fstack-protector-
strong -O2 -D_FORTIFY_SOURCE=2 -Wformat -Wformat-security -Werror=format-security --sysroot=recipe-
sysroot -O2 -pipe -g -feliminate-unused-debug-types -fmacro-prefix-map= -
fdebug-prefix-map= -fdebug-prefix-map= -fdebug-prefix-
map= -DOPENSSL_USE_NODELETE -DOPENSSL_PIC -DOPENSSL_CPUID_OBJ -DOPENSSL_BN_ASM_MONT -DSHA1_ASM -
DSHA256_ASM -DSHA512_ASM -DKECCAK1600_ASM -DVPAES_ASM -DECP_NISTZ256_ASM -DPOLY1305_ASM -DNDEBUG
evp 2682.45k 10842.73k 35957.50k 45915.48k 49722.71k 50135.04k

10.4.7 Running OpenSSL asymmetric tests with PKCS#11 based engine

Prerequisites:
1. For running the PKCS#11 OpenSSL Engine with our PKCS#11 Library, add the following into your global
OpenSSL configuration file (often in /etc/ssl/openssl.cnf).

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

104 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

This line must be placed at the top, before any sections are defined:
openssl_conf = openssl_init
Make sure there are no other openssl_conf = ... lines in the file.
This should be added to the bottom of the file:
[openssl_init]
engines=engine_section
[engine_section]
pkcs11 = pkcs11_section
[pkcs11_section]
engine_id = pkcs11
dynamic_path = /usr/lib/engines-3/pkcs11.so
MODULE_PATH = /usr/lib/libckteec.so.0
init = 0
The dynamic_path value is the PKCS#11 engine plug-in, and the MODULE_PATH value is the NXP
PKCS#11 library. The engine_id value is an arbitrary identifier for OpenSSL applications to select the
engine by the identifier.
2. Make sure tee-supplicant is running.
root@imx8mpevk:~# ps -aux | grep tee
root 661 0.0 0.0 76424 1432 ? Ssl May27 0:00 /usr/bin/
tee-supplicant
If it is not running, run the following command:
root@imx8mpevk:~# tee-supplicant &

10.4.7.1 Running p11tool to generate key (RSA or EC)

root@imx8mpevk:~# mkdir /etc/gnutls

root@imx8mpevk:~# echo load=`find /usr/lib -name libckteec.so.0` > /etc/gnutls/
pkcs11.conf
root@imx8mpevk:~# p11tool --list-tokens --provider=/usr/lib/libckteec.so.0
root@imx8mpevk:~# p11tool --initialize "<token url>" --label="<token label>" --
provider=/usr/lib/libckteec.so.0
Enter Security Officer's PIN:
root@imx8mpevk:~# p11tool --list-tokens --provider=/usr/lib/libckteec.so.0
root@imx8mpevk:~# p11tool --initialize-pin "<token url>" --provider=/usr/lib/
libckteec.so.0
Setting user's PIN...
Enter User's new PIN:
Token <toke label> with URL <token url> requires security officer PIN
Enter PIN:<Security Officer's PIN>

To generate an RSA key:

root@imx8mpevk:~# p11tool --login --generate-rsa --bits=2048 --label="RSA-

key-2048" --outfile="RSA-key-2048.pub" "<token url>" --set-pin="<user pin>" --
provider=/usr/lib/libckteec.so.0
root@imx8mpevk:~# p11tool --login --list-privkeys "<token url>" --set-pin="<user
pin>" --provider=/usr/lib/libckteec.so.0
Object 0:
URL: token url private
Type: Private key (RSA-2048)
Label: RSA-key-2048
Flags: CKA_PRIVATE; CKA_NEVER_EXTRACTABLE; CKA_SENSITIVE;

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

105 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

ID: bc:8e:f3:ca:95:d6:e7:ae:57:89:43:1f:67:a3:e5:d1:05:d8:5d:66

Or to generate an EC key:

root@imx8mpevk:~# p11tool --login --generate-ecc --curve=secp256r1 --label="ec-

key-256" --outfile="ec-key-256.pub" "<token url>" --set-pin="<user pin>" --
provider=/usr/lib/libckteec.so.0
root@imx8mpevk:~# p11tool --login --list-privkeys "<token url>" --set-pin="<user
pin>" --provider=/usr/lib/libckteec.so.0
Object 0:
URL: token url private
Type: Private key (EC/ECDSA-SECP256R1)
Label: ec-key-256
Flags: CKA_PRIVATE; CKA_NEVER_EXTRACTABLE; CKA_SENSITIVE;
ID: 9b:54:b4:c5:88:3f:19:44:cb:b2:40:04:46:fa:a0:48:19:eb:0e:70

10.4.7.2 Using OpenSSL from command line

To generate a certificate with its key in the PKCS #11 module, use the following commands. The first command
creates a self-signed Certificate for "NXP Semiconductor". The signing is done using the key specified by the
URL.

root@imx8mpevk:~# openssl req -engine pkcs11 -new -key "<token url private>" -
keyform engine -out req.pem -text -x509 -subj "/CN=NXP Semiconductor"
Engine "pkcs11" set.
Enter PKCS#11 token PIN for token label:<user pin>

The second command creates a self-signed certificate for the request. The private key used to sign the
certificate is the same private key used to create the request.

root@imx8mpevk:~# openssl x509 -engine pkcs11 -signkey "<token url private>" -

keyform engine -in req.pem -out cert.pem
Engine "pkcs11" set.
Enter PKCS#11 token PIN for token label:<user pin>
root@imx8mpevk:~# ls
cert.pem req.pem

10.4.7.3 Running OpenSSL test for RSA

root@imx8mpevk:~# echo "This is plain message 2021-01-18" > plain.text

root@imx8mpevk:~# openssl pkeyutl -engine pkcs11 -encrypt -in plain.text -out
encrypted.enc -inkey cert.pem -certin
root@imx8mpevk:~# openssl pkeyutl -engine pkcs11 -decrypt -in encrypted.enc -out
plain.dec -inkey "<token url private>" -keyform engine
Engine "pkcs11" set.
Enter PKCS#11 token PIN for token label:<user pin>
root@imx8mpevk:~# cat plain.text
This is plain message 2021-01-18
root@imx8mpevk:~# cat plain.dec
This is plain message 2021-01-18

10.4.7.4 Running OpenSSL test for EC

root@imx8mpevk:~# echo "This is plain message 2021-01-18" > plain.text

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

106 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

root@imx8mpevk:~# openssl pkeyutl -engine pkcs11 -sign -in plain.text -out

cert_ecc.sign -inkey "<token url private>" -keyform engine
Engine "pkcs11" set.
Enter PKCS#11 token PIN for token label:<user pin>
root@imx8mpevk:~# openssl pkeyutl -verify -in plain.text -sigfile cert_ecc.sign
-inkey ecc_cert.pem -certin
Signature Verified Successfully

10.4.7.5 Running the OPENSSL Kernel TLS test

Sever-side command:

openssl s_server -engine pkcs11 -key "<private token url>?pin-value=<user pin>"

-keyform engine -cert cert.pem -accept 443 -ssl_config ktls

For example:

openssl s_server -engine pkcs11 -key "pkcs11:model=OP-TEE

%20TA;manufacturer=Linaro;serial=0000000000000000;token=kshitiz_test;id=
%18%59%E1%2B%00%9B%13%4B%05%9E%2D%73%41%04%D1%8B%F8%24%03%ED;object=ecc-
key-256_using_p11;type=private?pin-value=1234" -keyform engine -cert cert.pem -
accept 443 -ssl_config ktls

Client-side command:

openssl s_client -quiet -connect <server_ip:port no> -tls1_2 -ssl_config ktls -

cipher <cipher suite>

For example:

openssl s_client -quiet -connect 10.232.132.7:443 -tls1_2 -ssl_config ktls -

cipher 'ECDHE-ECDSA-AES128-GCM-SHA256'

10.5 Disk encryption acceleration

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 for the platform that contains CAAM
hardware IP.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

107 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Linux
user space
Management tools File system (ext2, ext4)

Cryptsetup DM tools
Storage
/dev/mmcblk exports

af_alg LUKS /dev/sda /dev/lvmpool

Linux
Crypto-API kernel
RAID/LVM
Device mapper
drivers
CAAM
Symmetric Digest drivers Crypt target

asym mg
Peripherals
pcie drivers
caam_ jr
sata sdhc
JR

SATA PCI/NVMe
CAAM
USB SDHC
i.MX

SD aaa-052986

Figure 12. Disk Encryption use case

10.5.1 Enabling disk encryption support in kernel for the platform containing CAAM hardware
IP
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:

root@imx8mqevk:/# modprobe dm-mod

[ 266.982638] device-mapper: ioctl: 4.41.0-ioctl (2019-09-16) initialised: dm-
[email protected]
root@imx8mqevk:/# modprobe dm-crypt
root@imx8mqevk:/# dmsetup targets
crypt v1.19.0
striped v1.6.0
linear v1.4.0
error v1.5.0

If the disk encryption scenario is not enabled, some features in the kernel need to be enabled:

Kernel Configure Tree View Options Description

DM Crypt support enablement in
< Device Drivers ---> Linux kernel
[*] Multiple devices driver support (RAID and LVM) -
--> CONFIG_BLK_DEV_DM=y
<*> Device mapper support
[ ] Device mapper debugging support
< > Unstriped target (NEW)
<*> Crypt target support
<*> Multipath target

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

108 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Kernel Configure Tree View Options Description

[*] DM uevents

Enable user space crypto API's

< Cryptographic API ---> to allow simple cryptsetup
<*> User-space interface for hash algorithms benchmarks
<*> User-space interface for symmetric key cipher
algorithms
<*> User-space interface for AEAD cipher algorithms

Selecting this will register

Cryptographic API ---> algorithms supporting tagged
[*] Hardware crypto devices ---> key, generate black keys and
<*> CAAM/SNVS Security Violation Handler encapsulate them into black
(EXPERIMENTAL) blobs.
<*> Freescale CAAM-Multicore platform driver backend
[ ] Enable debug output in CAAM driver
<*> Freescale CAAM Job Ring driver backend --->
[*] Register tagged key cryptography implementations
with Crypto API

10.5.2 User space tools for disk encryption

All the required user space tools needed for DM-Crypt are already installed on the board when using Linux i.MX
BSP.
If the required user space tools are not installed in the build, add them by editing the conf/local.conf file
and appending:

CORE_IMAGE_EXTRA_INSTALL+="coreutils keyutils lvm2 e2fsprogs-mke2fs util-linux"

• 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 backed keys

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 secure key is to suppress the mechanism of encrypting the master
volume key with a key derived from a user-supplied passphrase.
DM-Crypt can be leveraged with:
• Trusted keys backed by CAAM
• CAAM's tagged key
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 encrypted, trusted, user, and logon.
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

109 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

The CAAM driver has associated a user-space application used to generate:

• A plain key and encapsulate it into a Red blob
• A tagged key and encapsulate it into a black blob

10.5.3.1 DM-Crypt with Trusted keys backed by CAAM

DM-Crypt fetches the trusted key which was generated through CAAM, from the kernel keyring, to take
advantage of CAAM state.
The key de-capsulated from Red-Blob is different for different CAAM states:
• If System is booted in secure boot with Chain-of-trust established, CAAM state is secure state.
• If system is booted in non-secure (or compromised) state, CAAM state is non-secure state.
Note:
Data that was written in secure state using the trusted key, is not read back from non-trusted or compromised
system.

10.5.3.1.1 Usage
The following steps shows how to perform a full disk encryption on i.MX devices using the DM-Crypt with
Trusted keys backed by CAAM method.
1. Insert the kernel module.
$>: modprobe trusted
2. Generate the trusted key:
$>: KEYNAME=dm_trust
$>: KEY="$(keyctl add trusted $KEYNAME 'new 32' @s)"
$>: keyctl pipe $KEY >~/$KEYNAME.blob
$>: keyctl list @s
Output:
$>: keyctl list @s
2 keys in keyring:
48178143: ----s-rv 0 0 user: invocation_id
143779047: --alswrv 0 0 trusted: dm_trust
3. Create a secure volume. It could be a physical partition. In this example, make use of an image file and
mount it later.
$>: DEV=/dev/loop0
$>: BLOCKS=20
$>: fallocate -l $((BLOCKS*512)) ~/loop0.img
$>: losetup -P $DEV ~/loop0.img
4. Create the mapping table "TABLE". Where:
• Algo is set in Kernel Crypto API format to use the plain key. Algo/cipher is set to cbc(aes)-plain.
• Key is set as the trusted key of length 32 and the name is exported as $KEYNAME.
$>: DEV=/dev/loop0
$>: ALGO=capi:cbc(aes)-plain
$>: KEYNAME=dm_trust
$>: BLOCKS=20
$>: TARGET=crypt

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

110 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

$>: TABLE="0 $BLOCKS $TARGET $ALGO :32:trusted:$KEYNAME 0 $DEV 0 1

allow_discards"
5. Use dmsetup to create a new device-mapper device named encrypted for example, and specify the
mapping table "TABLE" created above, as argument.
$>: echo $TABLE | dmsetup create encrypted
6. Load the device-mapper device named encrypted created in the previous step.
$>: echo $TABLE | dmsetup load encrypted
7. Create a secure volume.
$>: dd if=/dev/zero of=/dev/mapper/encrypted || true
8. Write to the volume.
$>: echo "It works. Congratulations" 1<> /dev/mapper/encrypted
9. Unmount the device.
$>: umount /mnt/encrypted/
10. Deactivate the device mapper device.
$>: dmsetup remove encrypted
Restart the board:
$>: reboot
11. In the next boot, insert the kernel module.
$>: modprobe trusted
Step 11: Load the trusted key:
$>: KEYNAME=dm_trust
$>: keyctl add trusted $KEYNAME "load $(cat ~/$KEYNAME.blob)" @s
$>: keyctl list @s
Output:
$>: keyctl list @s
2 keys in keyring:
48178143: ----s-rv 0 0 user: invocation_id
143779047: --alswrv 0 0 trusted: dm_trust
12. Create the mapping table "TABLE". Where:
• Algo is set in Kernel Crypto API format to use the plain key. Algo/cipher is set to cbc(aes)-plain.
• Key is set as the trusted key of length 32 and name is exported as $KEYNAME.
$>: DEV=/dev/loop0
$>: ALGO=capi:cbc(aes)-plain
$>: KEYNAME=dm_trust
$>: BLOCKS=20
$>: TARGET=crypt
$>: TABLE="0 $BLOCKS $TARGET $ALGO :32:trusted:$KEYNAME 0 $DEV 0 1
allow_discards"
13. Mount the encrypted device.
$>: losetup -P $DEV ~/loop0.img
14. Specify the mapping table "TABLE" to encrypt the volume using dmsetup.
$>: echo $TABLE | dmsetup create encrypted
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

111 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

$>: echo $TABLE | dmsetup load encrypted

15. Read from the device to verify if the content is same as it was written in the previous boot.
$>: hexdump -C /dev/mapper/encrypted

10.5.3.2 DM-Crypt with CAAM’s tagged key

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.
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.

caam-keygen application. Usability should be updated as below:-

$ 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>)
<text_type> can be -h or -p (default argument is -p)
-h generate a black key from the hex text that is provided in previous
argument
-p generate a black key from the plain text that is provided in previous
argument
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.
derive [-pass <pass_phrase>] [-md <digest>] [-S <salt>] <derived_key_name>
<pass_phrase> password value
<digest> Supported digest:-
sha1
sha224
sha256
sha384
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

112 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

sha512
Note:- Default algorithm is sha-256.
<salt> [Optional] The actual salt to use.
8 bytes salt value needs to be provided.
If salt value > 8 bytes, trim to 8 bytes.
If salt_value < 8 bytes, zero padding is added.
If no salt is provided, -nosalt option will be used.
<derived_key_name> Black key obtained after using PBKDF2
derivation function.

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.

User space Device-mapper table

parameters (target: crypt,
cipher, key ...)
Secure key blob

keyctl dmsetup

Linux kernel
ext2 ext3 ext4

keyring DM-crypt

Crypto-API Block devices

Linux
kernel drivers

aes-cbc aes-xts tk(cbc(aes)) pcie sata qspi

Hardware
PCIe SATA qSPI

CPU, CAAM
SD/MMC IFC

aaa-052987

Figure 13. 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.

10.5.3.3 Usage
The following are the steps to perform a full disk encryption on i.MX devices using the DM-Crypt with CAAM's
tagged key method.
1. After booting the device, make sure that cryptographic transformations using Tagged Key are registered.
root@imx8mqevk:~# grep -B1 -A2 tk- /proc/crypto|grep -v kernel
name : tk(ecb(aes))
driver : tk-ecb-aes-caam
priority : 3000
--
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

113 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

name : tk(cbc(aes))
driver : tk-cbc-aes-caam
priority : 3000
root@imx8mqevk:~#
And caam-keygen application is available:
root@imx8mmevk:~# cd /; find -name "caam-keygen"
./usr/bin/caam-keygen
./dev/caam-keygen
./sys/class/misc/caam-keygen
./sys/devices/virtual/misc/caam-keygen
For now, we only support AES algorithms. Therefore, the size of the key accepted for encryption/decryption
is 16, 24, and 32 bytes.
2. Make sure DM-Crypt is enabled.
root@imx8mqevk:~# dmsetup targets
crypt v1.19.0
striped v1.6.0
linear v1.4.0
error v1.5.0
If any of the above is missing, check Kernel configurations or see section Enable disk encryption support in
kernel.
3. Then, provide the device with its key, the black key, which could be created either from a defined plain key
or randomly.
Here is an example for black key encrypted with ECB, from a given plaintext of size 16 bytes:
root@imx8mqevk:~# ./caam-keygen create fromTextkey ecb -t 0123456789abcdef
The result is a Tagged Key and a Blob files written to filesystem (the default location is /data/caam). The
used key encryption scheme is ECB.
root@imx8mqevk:~# ls -la /data/caam/
total 16
drwxr-xr-x 2 root root 4096 Aug 25 15:38 .
drwxr-xr-x 3 root root 4096 Aug 25 15:38 ..
-rw-r--r-- 1 root root 36 Aug 25 15:38 fromTextkey
-rw-r--r-- 1 root root 96 Aug 25 15:38 fromTextkey.bb
Next, add the key in key retention service, using keyctl:
root@imx8mqevk:~# cat /data/caam/fromTextkey | keyctl padd logon logkey: @s
876928653
4. Create a secure volume. It could be a physical partition. In this example, make use of an image file and
mount it later.
root@imx8mqevk:~# dd if=/dev/zero of=encrypted.img bs=1M count=32
32+0 records in
32+0 records out
33554432 bytes (34 MB, 32 MiB) copied, 3.20227 s, 10.5 MB/s
root@imx8mqevk:~#
root@imx8mqevk:~# losetup /dev/loop0 encrypted.img
root@imx8mqevk:~#
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.
root@imx8mqevk:~# dmsetup -v create encrypted --table "0 $(blockdev --getsz /dev/loop0) crypt
capi:tk(cbc(aes))-plain :36:logon:logkey: 0 /dev/loop0 0 1 sector_size:512"
Name: encrypted
State: ACTIVE
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

114 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Read Ahead: 256

Tables present: LIVE
Open count: 0
Event number: 0
Major, minor: 253, 0
Number of targets: 1

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.
For more detailed options and descriptions, refer to https://gitlab.com/cryptsetup/cryptsetup/-/wikis/DMCrypt.
The created device appears in /dev/mapper:
root@imx8mqevk:~# dmsetup table --showkey encrypted
0 65536 crypt capi:tk(cbc(aes))-plain :36:logon:logkey: 0 7:0 0
6. Create a file system on the device.
root@imx8mqevk:~# mkfs.ext4 /dev/mapper/encrypted
mke2fs 1.45.3 (14-Jul-2019)
Creating filesystem with 32768 1k blocks and 8192 inodes
Filesystem UUID: 3ba01ad8-ba03-4389-a955-5136b3173c35
Superblock backups stored on blocks:
8193, 24577
Allocating group tables: done
Writing inode tables: done
Creating journal (4096 blocks): done
Writing superblocks and filesystem accounting information: done
7. Set up a mount point.
root@imx8mqevk:~# mkdir /mnt/encrypted
8. Mount the mapped device.
root@imx8mqevk:~# mount -t ext4 /dev/mapper/encrypted /mnt/encrypted/
[ 9409.936183] EXT4-fs (dm-0): mounted filesystem with ordered data mode.
Opts: (null)
[ 9409.943892] ext4 filesystem being mounted at /mnt/encrypted supports
timestamps until 2038 (0x7fffffff)
9. Write to device.
root@imx8mqevk:~# echo "This is an encrypt with black key (ECB from text 16 bytes key size) test
of full disk encryption on i.MX" > /mnt/encrypted/readme.txt

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

115 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

10. Unmount the device.

root@imx8mqevk:~# umount /mnt/encrypted/
11. Deactivate the device mapper device.
root@imx8mqevk:~# dmsetup remove encrypted
12. Restart the board.
root@imx8mqevk:~# reboot
...
root@imx8mqevk:~#
13. Import the key from blob and add it to key retention service.
root@imx8mqevk:~# ./caam-keygen import /data/caam/fromTextkey.bb importKey
root@imx8mqevk:~# cat /data/caam/importKey | keyctl padd logon logkey2: @s
605536287
root@imx8mqevk:~# ls -la /data/caam/
total 20
drwxr-xr-x 2 root root 4096 Aug 25 15:47 .
drwxr-xr-x 3 root root 4096 Aug 25 15:38 ..
-rw-r--r-- 1 root root 36 Aug 25 15:38 fromTextkey
-rw-r--r-- 1 root root 96 Aug 25 15:38 fromTextkey.bb
-rw-r--r-- 1 root root 36 Aug 25 15:47 importKey
root@imx8mqevk:~#
14. Mount the encrypted device.
root@imx8mqevk:~# losetup /dev/loop0 encrypted.img
root@imx8mqevk:~#
15. Specify the mapping table to encrypt the volume using dmsetup.
root@imx8mqevk:~# dmsetup -v create encrypted --table "0 $(blockdev --getsz /dev/loop0) crypt
capi:tk(cbc(aes))-plain :36:logon:logkey2: 0 /dev/loop0 0 1 sector_size:512"
Name: encrypted
State: ACTIVE
Read Ahead: 256
Tables present: LIVE
Open count: 0
Event number: 0
Major, minor: 253, 0
Number of targets: 1

16. Mount.
root@imx8mqevk:~# mount /dev/mapper/encrypted /mnt/encrypted/
[ 191.961828] EXT4-fs (dm-0): mounted filesystem with ordered data mode.
Opts: (null)
[ 191.969533] ext4 filesystem being mounted at /mnt/encrypted supports
timestamps until 2038 (0x7fffffff)
root@imx8mqevk:~
17. Read from the device.
root@imx8mqevk:~# cat /mnt/encrypted/readme.txt
This is an encrypt with black key (ECB from text 16 bytes key size) test of
full disk encryption on i.MX.
root@imx8mqevk:~#
18. Unmount the device and deactivate the device mapper device.
root@imx8mqevk:~# umount /mnt/encrypted/; dmsetup remove encrypted

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

116 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

10.5.4 DM-Crypt using DCP's Crypto Key

Perform the following steps to run DM-Crypt on the i.MX 6ULL platform:
1. Insert the trusted module for using the trusted keyring.
$ modprobe trusted;
2. Add the otp_crypto_key handle in the trusted keyring.
$ dd if=/dev/zero of=encrypted.img bs=4k count=4k
$ losetup /dev/loop0 encrypted.img;
3. Create a secure volume. It could be a physical partition. In this example, make use of an image file and
mount it later.
$ dd if=/dev/zero of=encrypted.img bs=4k count=4k
$ losetup /dev/loop0 encrypted.img;
4. Use dmsetup, and create a new device-mapper device named mydev, for example, and then specify a
mapping table. The table can be provided on stdin or as argument.
$ export TABLE="0 $(blockdev --getsz /dev/loop0) crypt capi:cbc-aes-dcp-
plain :16:trusted:<key name> 0 /dev/loop0 0 1 allow_discards"; $ echo $TABLE
| dmsetup create mydev
5. The created device appears in /dev/mapper.
$ echo $TABLE | dmsetup load mydev $ dmsetup table --showkey mydev
Example Output of above command:- 0 32768 crypt capi:cbc-aes-dcp-
plain :16:trusted:<key name> 0 7:0 0 1 allow_discards
6. Create a file system on the device.
$ mkfs.ext4 /dev/mapper/mydev
7. Set up a mount point.
$ mkdir /mnt/mydev
8. Mount the mapped device.
$ mount -t ext4 /dev/mapper/mydev /mnt/mydev/
Note: At this level, every data you write to /mnt/mydev is encrypted on the real block device /dev/
loop0.
9. Write to the device.
$ echo "This is a test of full disk encryption on i.MX" > /mnt/mydev/
readme.txt
10. Unmount the device.
$ umount /mnt/mydev/
11. Deactivate the device mapper device.
$ dmsetup remove mydev
12. Restart the device.
13. Insert the trusted keyring and load the otp_crypto_key handle in the key retention service.
modprobe trusted;export KEYNAME=<key name>; keyctl add trusted <key name>
"load $(hexdump -v -e '"" 1/1 "%02X"' /proc/device-tree/soc/bus@2200000/
crypto@2280000/otp_crypto_key) hw" @s;

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

117 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

14. Mount the encrypted device.

losetup /dev/loop0 encrypted.img
15. Specify the mapping table to encrypt the volume using dmsetup.
export TABLE="0 $(blockdev --getsz /dev/loop0) crypt capi:cbc-aes-dcp-
plain :16:trusted:<key name> 0 /dev/loop0 0 1 allow_discards"; echo $TABLE |
dmsetup create mydev
16. Mount.
mount /dev/mapper/mydev /mnt/mydev/
17. Read from the device.
cat /mnt/mydev/readme.txt
The system returns the following result:
This is a test of full disk encryption on i.MX

10.5.5 DM-Crypt usage on i.MX Platforms without CAAM hardware IP

The following figure shows the software stack that implements disk encryption on the i.MX platforms without
CAAM hardware IP.

Normal world Secure world

dm-setup keyctl
User space
Kernel space

K1
dm-crypt
K2 Key ring
encrypted keys Trusted ARM CE
HW RNG PTA
PTA
Block
Crypto-API
devices

SATA Ciphers HW RNG

ELE driver
Secure crypto
Linux Kernel drivers
driver
OP-TEE OS

Monitor mode
ATF

Hardware
RNG KDF
SATA OCRAM
Edgelock enclave

aaa-056243

Figure 14. Software stack that implements disk encryption on the i.MX platforms without CAAM hardware IP

This approach is implemented for protecting the DM-Crypt secrets and crypto operations, on the i.MX platforms
that do not have CAAM IP, such as i.MX 93.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

118 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

To protect the DM-Crypt operations like CAAM does, the following three key requirements are expected to be
fulfilled:
• Key does not leave the SoC:
In absence of the CAAM black key, the key(s) is/are always kept in Secure OCRAM giving security from
physical attacks.
• Crypto operations are done isolated boundary of CAAM:
In absence of CAAM, it is done in OP-TEE with Trusted Arm CE PTA, which is an isolated trusted execution
environment.
• Crypto Ops performance requirements like CAAM hardware accelerator:
In absence of CAAM, it can be satisfied by using the Arm Crypto Extension in Trusted Arm CE PTA.
Prerequisites:
• Ensure that a region of OCRAM is reserved. OP-TEE is using the OCRAM region:
– 0x20518000 - 0x2051C000 for saving keys.
• Compile the Kernel Image with the following KCONFIG:
– CONFIG_DM_CRYPT=y
– CONFIG_TRUSTED_KEYS=m
– CONFIG_TRUSTED_KEYS_TEE=y
• OP-TEE OS to be compiled to include Arm CE PTA:
– Currently, Runtime Random Number Generation from ELE is disabled because of some issues.
– To run this feature, disable CFG_WITH_SOFTWARE_PRNG.
– In all, flags that need to be enabled in OP-TEE:
– CFG_WITH_SOFTWARE_PRNG: To enable Runtime ELE RNG.
– Flags that need to be disabled in Linux OS:
– CONFIG_IMX_SEC_ENCLAVE: To disable the Linux Secure Enclave drive.
– CONFIG_IMX_ELE_TRNG: To disable Linux ELE TRNG support.
Usage:
• Insert the kernel module:
modprobe tee_crypto
modprobe trusted
modprobe dm_crypt
• Generate the key:
$:> export KEYNAME=dm_trust_plainkey
$:> KEY="$(keyctl add trusted $KEYNAME 'new 32' @s)"
$:> keyctl pipe $KEY >~/$KEYNAME.blob
$:> keyctl list @s
• Set the variables:
$:> export DEV=/dev/loop0
$:> export ALGO="capi:cbc-aes-tee-plain"
$:> export BLOCKS=$(blockdev --getsz /dev/loop0)
• Change the directory:
$:> cd /dev/shm/
• Run the following dd command:
$:> dd if=/dev/zero of=encrypted.img bs=1M count=512 && losetup /dev/loop0
encrypted.img

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

119 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• Set the DM-Setup TABLE variables:

$:> TABLE="0 $BLOCKS crypt $ALGO :32:trusted:$KEYNAME 0 $DEV 0 1
sector_size:512"
• Run the DM-Setup command:
$:> dmsetup -v create encrypted --table "$TABLE"
$:> mkfs.ext4 /dev/mapper/encrypted
$:> mount -t ext4 /dev/mapper/encrypted /mnt
• Change the directory:
$:> cd /mnt/
• Run the DD test:
$:> while [ 1 ] ; do dd if=/dev/random of=temp bs=4096 count=65536 conv=fsync;
done;

10.6 crypto_af_alg application support

10.6.1 Prerequisites
The caam-keygen application is needed to import the black key from the black blob. Make sure that the caam-
keygen application is already present at /usr/bin.

10.6.2 Building the kernel

10.6.2.1 Kernel configuration

• CONFIG_CRYPTO_USER_API
• CONFIG_CRYPTO_USER_API_HASH
• CONFIG_CRYPTO_USER_API_SKCIPHER
• CONFIG_CRYPTO_USER_API_RNG
• CONFIG_CRYPTO_USER_API_AEAD
Get a bootable image that includes the black key support and AF_ALG socket interface for the Linux kernel. Or
build the kernel from here: https://github.com/nxp-imx/linux-imx/.

10.6.2.2 Building a toolchain

Build a toolchain to cross compile the sources of the caam-crypt application. For details, see the i.MX Yocto
Project User's Guide (UG10164).

$ wget https://developer.arm.com/-/media/Files/downloads/gnu-a/8.2-2019.01/gcc-
arm-8.2-2019.01-x86_64-aarch64-elf.tar.xz
$ tar xf gcc-arm-8.2-2019.01-x86_64-aarch64-elf.tar.xz

10.6.2.3 Cross compiling the user space sources

Set up the environment for cross compilation using the toolchain previously prepared.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

120 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

1. In the toolchain folder, set up the environment.

$ export CROSS_COMPILE=<path to toolchain>/bin/aarch64-linux-gnu-
$ export CC=${CROSS_COMPILE}gcc
$ export LD=${CROSS_COMPILE}ld
2. Build the caam-crypt user space application. Go to the source folder and run:
$ make clean
$ make

10.6.3 Usage
After the device successfully boots with the previously generated image, caam-crypt can be used to decrypt an
encrypted data stored in a file.

$ ./caam-crypt
Application usage: caam-crypt [options]
Options:
<crypto_op> <algo> [-k <blob_name>] [-in <input_file>] [-out <output_file>] [-
iv <IV value>]
<crypto_op> can be enc or dec
enc for encryption.
dec for decryption.
<algo> can be AES-256-CBC
<blob_name> the absolute path of the file that contains the black blob
<input_file> the absolute path of the file that contains input data
In case of encryption, input file will contain plain data.
In case of decryption, input file will contain encrypted data.
<output_file> the absolute path of the file that contains output data
<IV value> 16 bytes IV value

10.6.4 Use case example

• For encryption:
$ caam-crypt enc AES-256-CBC -k myblob -in <plain_text_file> -out
<encrypted_file> -iv <16-byte IV value>
• For decryption:
$ caam-crypt dec AES-256-CBC -k myblob -in <encrypted_file> -out
<decrypted_file> -iv <16-byte IV value>

where:
• myblob: Generated black key blob. The caam-keygen application imports a black key from black blob. This
black key is used by CAAM for encryption/decryption.
• AES-256-CBC: Currently, the only supported symmetric algorithm used for encryption/decryption operation.
Note: Make sure that the algorithm used for encryption/decryption should be same.
• encrypted_file: Encrypted data stored in a file.
• plain_text_file: Plain text stored in a file (Padding is added for making data as multiples of block size).
• decrypted_file: Decrypted data stored in a file.
• iv: 16 bytes IV value.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

121 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Note: Customer should increase net.core.optmem_max using the command below:

# sysctl -w net.core.optmem_max=1048576

The default value is set to 20480. Increasing to 1048576 (1 MB) seems to fix the issue of decryption of file size
greater than 1 MB.

10.7 Kernel TLS offload

Linux kernel provides TLS connection offload infrastructure. Once a TCP connection is in ESTABLISHED state,
user space can enable the TLS Upper Layer Protocol (ULP) and install the cryptographic connection state. For
details regarding the user-facing interface, refer to the TLS documentation in Kernel TLS.

10.7.1 Prerequisites
Check OpenSSL version using the following command. It must be 3.0.0 or higher.

openssl version

10.7.2 Running Kernel TLS test

On server generate RSA 2048 key, certificate and run openssl s_server:

root@imx8mmevk:~# openssl req -new -newkey rsa:2048 -nodes -keyout rsa.key -out
rsa.csr
root@imx8mmevk:~# openssl x509 -req -sha256 -days 365 -in rsa.csr -signkey
rsa.key -out server.pem
root@imx8mmevk:~# openssl s_server -key rsa.key -cert server.pem -accept 443 -
ssl_config ktls
Using default temp DH parameters
ACCEPT

Run openssl s_client from another terminal:

root@imx8mmevk:~# openssl s_client -quiet -connect <server ip>:443 -tls1_2 -

ssl_config ktls -cipher 'ECDHE-RSA-AES256-GCM-SHA384'
Connecting to <server ip>
Can't use SSL_get_servername
...
Using Kernel TLS for sending
Using Kernel TLS for receiving
<write some message and enter>

Remove -quiet to see full client logs. With TLSv1.2, the Kernel TLS supports these ciphers:
• AES128-GCM-SHA256
• AES256-GCM-SHA384
• ECDHE-RSA-AES128-GCM-SHA256
• ECDHE-RSA-AES256-GCM-SHA384

10.8 IMA/EVM on i.MX SoCs

Integrity Measurement Architecture (IMA): is the Linux integrity subsystem used to detect if files have
been accidentally or maliciously altered. It appraises a file's measurement against a "good" value stored

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

122 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

as an extended attribute (security.ima) and enforces local file integrity checks. The extended attribute
(security.ima) of a file is the hash value (SHA-1, SHA-256, or SHA-512) of its content. IMA maintains a list
of hash values over all executables and other sensitive system files loaded at runtime into the system.
Extended Verification Module (EVM): protects a file’s extended attributes against integrity attacks. The
extended security attribute (security.evm) stores the HMAC value over other extended attributes associated
with the file such as security.selinux, security.SMACK64, and security.ima.
EVM depends on the kernel key retention system and requires an encrypted key named evm-key for the HMAC
operation. The key is loaded onto the root user keyring using keyctl utility. EVM is enabled by setting an enable
flag in securityfs/evm file.
In normal secure boot process, contents of root file system mounted over persistent storage device are not
validated by any mechanism and hence cannot be trusted. Any malicious changes in non-trusted rootfs
contents are undetected. IMA EVM is the Linux standard mechanism to verify the integrity of the rootfs. Integrity
checks over file attributes and its contents are performed by Linux IMA EVM module before its execution. IMA
EVM depends on encrypted key loaded on user’s keyring. Loading keys to root user keyring and enabling
EVM is typically done using initramfs image. The initramfs image is validated using secure boot process and
becomes the part of chain of trust. Initramfs switches control to main rootfs mounted over storage device, after
EVM is successfully enabled on the system.

10.8.1 EVM Key on user keyrings

The EVM security attribute depends on an encrypted key (named evm-key) loaded on the user keyring. The
encrypted key is derived by the kernel using the master key. The master key can be of the following types:
• User-Key
• Secure-Key
• Trusted-Key
Secure and trusted keys are derived using a hardware security engine for greater security while the security
of user-key depends on the user-defined mechanisms irrespective of the hardware. The secure-key is derived
using the Layerscape's SEC (aka CAAM). The trusted-key can be used on the platforms supporting TPM.
The encrypted key acts as an HMAC key, which is subsequently used to calculate the HMAC value
(security.evm) over other security attributes. This key is stored internally by the kernel and user can only see
its blob.

10.8.2 Modes of operation in IMA EVM

IMA/EVM is enabled in two modes:
• Fix mode
• Enforce mode
To enable a system with IMA EVM, both modes must be implemented in a sequence as described below:
1. The system needs to be booted in fix mode with ima_appraise=fix and evm=fix bootargs. After loading
the keys on the root keyring, the entire file system is labelled with security attributes. In fix mode, any file
with INTEGRITY_UNKNOWN is labelled with proper attribute values. This mode must be executed only once
while preparing system for field deployment.
2. After the fix mode execution is completed successfully, system needs to be booted IMA EVM in enforce
mode. Enforce mode is enabled by setting ima_appraise=enforce bootargs. In enforce mode, the files
are measured against their “good” values. In case there is a mismatch between calculated security attribute
value and stored value, access to that file is denied. While in field the system boot is done in enforce mode
only.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

123 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

10.8.3 Build Steps

Follow instructions mentioned in https://github.com/nxp-imx/meta-imx-integrity for building initramfs.
1. After building integrity-image-minimal-<board-name>-<build_no>.rootfs.tar.zst, extract
tar.zst.
2. Convert the extracted directory into CPIO file using the following command:
find . | cpio -H newc -o > ../<rootfs_name>.cpio
3. Gzip the built rootfs above using the following command:
gzip ../<rootfs_name>.cpio
4. Convert gzipped rootfs into Ramdisk file using the following command:
mkimage -A arm -O linux -T ramdisk -d <gzipped_rootfs> <Ramdisk_name>
5. Flash the kernel image, dtb file, and Ramdisk file on the i.MX board.
6. For fix mode, add the following bootargs to the current bootargs:
rootwait rw lsm=integrity rootflags=i_version ima_appraise=fix
ima_policy=appraise_tcb evm=fix initrd=<Ramdisk path>
7. For enforce mode, add the following bootargs to the current bootargs:
rootwait rw lsm=integrity rootflags=i_version ima_appraise_tcb
ima_appraise=enforce initrd=<Ramdisk path>"

10.8.4 Steps to verify IMA EVM feature

Perform the following checks to ensure that IMA EVM is successfully enabled in enforce mode.
1. Trusted keys and encrypted keys are enabled in kernel image. The Following kernel logs ensures that
secure key and encrypted key is successful registered.
[ 6.893635] Key type trusted registered
[ 6.905123] Key type encrypted registered
2. IMA EVM is enabled in the kernel image. The following kernel logs ensures IMA EVM is enabled.
[ 6.909218] ima: No TPM chip found, activating TPM-bypass!
[ 6.914738] ima: Allocated hash algorithm: sha1
[ 6.962804] evm: HMAC attrs: 0x1
3. System is up in enforce mode. The following logs from initramfs and kernel ensures enforce mode is
enabled.
[ 8.248060] EXT4-fs (mmcblk0p2): mounted filesystem with ordered data
mode. Opts: (null). Quota mode: none.
Loading blobs
[ 8.294368] evm: key initialized
4. EVM attributes over a file can be checked using getfattr utility.
root@imx8ulpevk:~# getfattr -d -m . /path/to/file
5. Security attribute are appraised successfully upon changing any file contents. The following commands
verify the appraise functionality.
root@imx8ulpevk:~# vim test_file // write contents "abc"
root@imx8ulpevk:~# getfattr -d -m . /path/to/test_file
root@imx8ulpevk:~# vim test_file // write contents "abcd"

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

124 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

root@imx8ulpevk:~# getfattr -d -m . /path/to/test_file

10.8.4.1 Testing hashing

Perform the following checks to ensure that IMA EVM is successfully enabled in enforce mode.
1. Trusted keys and encrypted keys are enabled in kernel image. The Following kernel logs ensures that
secure key and encrypted key is successful registered.
[ 6.893635] Key type trusted registered
[ 6.905123] Key type encrypted registered
2. IMA EVM is enabled in the kernel image. The following kernel logs ensures IMA EVM is enabled.
[ 6.909218] ima: No TPM chip found, activating TPM-bypass!
[ 6.914738] ima: Allocated hash algorithm: sha1
[ 6.962804] evm: HMAC attrs: 0x1
3. System is up in enforce mode. The following logs from initramfs and kernel ensures enforce mode is
enabled.
[ 8.248060] EXT4-fs (mmcblk0p2): mounted filesystem with ordered data
mode. Opts: (null). Quota mode: none.
Loading blobs
[ 8.294368] evm: key initialized
4. EVM attributes over a file can be checked using getfattr utility.
root@imx8ulpevk:~# getfattr -d -m . /path/to/file
5. Security attribute are appraised successfully upon changing any file contents. The following commands
verify the appraise functionality.
root@imx8ulpevk:~# vim test_file // write contents "abc"
root@imx8ulpevk:~# getfattr -d -m . /path/to/test_file
root@imx8ulpevk:~# vim test_file // write contents "abcd"
root@imx8ulpevk:~# getfattr -d -m . /path/to/test_file

10.8.4.2 Testing signing

Perform the following steps to test signing.
1. In enforce mode, link the user keyring with session keyring using the following command:
keyctl link @us @s
2. keyctl show is as follows:
root@imx8ulpevk:~# keyctl show @u
Keyring
903561555 --alswrv 0 65534 keyring: _uid.0
744420798 --alswrv 0 0 \_ trusted: kmk-trusted
779914984 --alswrv 0 0 \_ keyring: _ima
960482942 --alswrv 0 0 | \_ user: B0E0FBB7D52954F3
346128152 --alswrv 0 0 \_ encrypted: evm-key
3. Currently, all the .ko files are signed. Therefore, echoing to .ko files may cause permission denied error as
follows:
root@imx93evk:~# sudo echo "echo \"Hello from the root user\"" >> /lib/
modules/6.6.23-lts-next-06180-g4958e7a7a6dd/kernel/crypto/sm3.ko

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

125 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

-sh: /lib/modules/6.6.23-lts-next-06180-g4958e7a7a6dd/kernel/crypto/sm3.ko:
Permission denied

10.9 Security reference design

Note: All reference implementations are for development purposes only. Users should ensure that proper
security policy/guidelines are considered while handling security critical material.
The NXP Security Reference Design initiative aims to help customers enable/configure security features
through the Yocto build process. This framework is intended to ease the utilization and modification of security
features in the NXP processors.
The following sections describe the various enablement as part of the Security Reference Design framework.

10.9.1 Automated image signing for secure boot

Note: See meta layer documentation for the information on the supported devices and supported distro for this
release.
Secure boot is an industry standard to ensure that the device boots a trusted OEM software. NXP i.MX devices
support Root of Trust (RoT) through the secure boot method using the OEM trusted root keys. In general, the
secure boot mechanism involves signing and authentication of Bootloader and OS kernel image.
The signing method involves determining the part of the boot image that needs to be signed by analyzing the
build log and then using NXP Code Signing Tool. This method is usually found to be manual and error prone
causing signing/authentication failure, which may eventually lead to boot failures. To eliminate such errors, an
automated method is devised, which analyzes the input boot image and signs it using CST.
The NXP BSP release contains a meta-nxp-security-reference-design/meta-secure-boot Yocto
meta layer, which supports i.MX boot image signing automation. The usage of this meta layer is described in the
following sections.

10.9.1.1 NXP CST Signer

Note: See NXP CST Signer Tool for more documentation.
The NXP CST Signer Tool created by NXP, analyzes the input NXP BSP image to extract the offset and size of
the image(s) that need to be signed and prepares the corresponding Command Sequence File for CST to sign.
This tool is used in the meta layer to support image signing. However, it can also be used as a standalone tool.
The NXP CST Signer tool has the following properties:
• Supports signing both HAB (i.MX 6/7/8M Family) and AHAB (i.MX 8/8x/8ULP/9 Family) devices' images.
• Analyzes input image to sign IVT/FIT/Container format images.
• Extracts offsets and sizes from the input image and constructs CSFs.
• Default configuration files are present in the CST Signer repository, filled with the basic information related to
keys, certificate and flags to be used, both for HAB and AHAB devices.

10.9.1.2 Prerequisites for preparing a signed image

The prerequisites for preparing a signed image are as follows:
1. Download the NXP Code Signing Tool delivered by NXP, which is required for this automation to work.
2. Prepare the keys using CST.
By default, the NXP CST Signer Tool uses standard keys of type ECC P256-SHA256 for i.MX 8/8x/8ULP/9
Family and RSA 2048-SHA256 for i.MX 6/7/8M Family, to be available in the download location of CST.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

126 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Follow the CST User Guide available in the CST package to generate the keys, certificates, SRK table/
fuses and for more information.
Note: (Optional) Create and populate csf_hab4.cfg and/or csf_ahab.cfg with the preferred key type
at the CST location to use your preferred PKI tree. The default configuration files are located at the CST
Signer work directory in Yocto build.

10.9.1.3 Yocto setup for secure boot build

Note: See the i.MX Yocto Project User's Guide (UG10164) for how to set up and build Yocto project.
To set up the Yocto project for secure boot build, perform the following steps:
1. Set up Yocto Build Environment and Configuration.
repo init -u https://github.com/nxp-imx/imx-manifest -b imx-linux-scarthgap -
m imx-6.6.36-2.1.0_security-reference-design.xml
repo sync
DISTRO=<DISTRO> MACHINE=<MACHINE> source imx-setup-release.sh -b <build
directory>
2. Add the meta-secure-boot layer to the Yocto project.
bitbake-layers add-layer ../sources/meta-nxp-security-reference-design/meta-
secure-boot
3. Add CST in CST_PATH in local.conf.
Note: The absolute location of CST is required.
echo “CST_PATH = to cst package>\”” >> conf/local.conf

10.9.1.4 Generating a signed bootloader/kernel/WIC image in Yocto project

To generate a signed bootloader/kernel/WIC image in Yocto project, perform the following steps:
1. Build a signed WIC image.
bitbake core-image-minimal-secure-boot
2. Build a signed imx-boot bootloader (for i.MX 8M/8/8x/8ULP/9x).
bitbake imx-boot-signature
3. Build a signed U-Boot bootloader (for i.MX 6/7).
bitbake u-boot-signature
4. Build a signed Linux kernel image.
bitbake linux-imx-signature

10.9.1.5 Booting a signed image

The output of the Yocto build contains three signed artifacts:
• Signed Bootloader
• Signed Kernel image
• SDcard (WIC) image (Contains signed bootloader and signed kernel images only)
Either of the signed bootloader or kernel image can be downloaded to the target based on the development
scenario. The signed WIC image is provided to deploy the signed bootloader and kernel artifacts to the device.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

127 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• Secure boot in OPEN/OEM OPEN lifecycle

It is recommended to download the signed image in the OPEN lifecycle state and verify secure boot before
closing the part. For more information on how to program the SRK Fuses, verifying signature and closing
the part to enable secure boot, see the secure boot user guide, based on the SoC, at HAB4 UBoot Guide or
AHAB UBoot Guides.
• Secure boot in CLOSED/OEM CLOSED lifecycle
When the SoC is in CLOSED lifecycle state, the same pre-provisioned signed image must boot the device in a
closed security state ensuring a trusted image is running on the device.

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.

Table 88. On-board chips and external solutions for Bluetooth and Wi-Fi support
SoC On-board chip PCIe M.2 card uSD card or SDIO M.2 card
8QuadXPlus/8DXL - NXP PCIe 88W9098 (tested -
with Murata LBEE5ZZ1XL)
8QuadMax - NXP PCIe 88W9098 (tested -
with Murata LBEE5ZZ1XL)
8M Quad - NXP 88W8997 (tested with -
Murata LBEE5XV1YM)
On i.MX 8M Quad WEVK
board (use M.2 on the bottom
side):
NXP PCIe 88W9098 (tested
with Murata LBEE5ZZ1XL).
8M Nano NXP 88W8987 (tested with - NXP SDIO IW612 (tested
AzureWave AW-CM358SM) with Murata LBES5PL2EL)

8M Mini NXP 88W8987 (tested with - -

AzureWave AW-CM358SM)
7ULP - - NXP 88W8987 (tested with
Murata LBEE5QD1ZM)
7Dual - - NXP 88W8987 (tested with
Murata LBEE5QD1ZM)
6QuadPlus/Quad/Dual/Solo - - NXP 88W8987 (tested with
Murata LBEE5QD1ZM)
6SLL/6UltraLite/6ULL/6ULZ - - NXP 88W8987 (tested with
Murata LBEE5QD1ZM)
NXP SDIO IW416 (tested
with Murata LBEE5CJ1XK)
NXP SDIO 88W8801 (tested
with Murata LBWA0ZZ2DS)
NXP SDIO IW612 (tested
with Murata LBES5PL2EL)

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

128 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 88. On-board chips and external solutions for Bluetooth and Wi-Fi support...continued
SoC On-board chip PCIe M.2 card uSD card or SDIO M.2 card
8M Plus - NXP 88W8997 (tested with NXP SDIO 88W8997 (tested
AW-CM276MAPUR) with Murata LBEE5XV1YM)
NXP PCIe 88W9098 (tested NXP SDIO 88W9098 (tested
with Murata LBEE5ZZ1XL). with Murata LBEE5ZZ1XL)
8ULP - - NXP SDIO IW416 (tested
with Murata LBEE5CJ1XK)
i.MX 91 - - NXP SDIO IW612 (tested
with Murata LBES5PL2EL)
i.MX 93 - - NXP SDIO IW612 (tested
with Murata LBES5PL2EL)
i.MX 95 - On i.MX 95 19x19 EVK On i.MX 95 15x15 EVK
board: board:
NXP PCIe 88W9098 (tested NXP SDIO IW612 (tested
with U-Blox JODY-W3). with Murata LBES5PL2EL)

Note: All Murata LBEE5QD1ZM are tested on i.MX 6/i.MX 7 platforms along with the Murata M.2-to-usd
adapter.
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 88W9098-based modules
with PCIe and SDIO interfaces, NXP 88W8997-based modules with PCIe and SDIO interfaces, NXP IW416-
based modules with SDIO interface, NXP 88W8801-based modules with SDIO interface, and NXP IW612-
based modules with SDIO 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:
• For the following steps, execute these commands using connman
# For all the Wi-Fi modules:
modprobe moal mod_para=nxp/wifi_mod_para.conf
$connmanctl
connmanctl> enable wifi
connmanctl> scan wifi
connmanctl> services /* This should list of the network. For example
wifi_c0e4347f5053_4a62726f_managed_psk*/
connmanctl> agent on
connmanctl> connect wifi_c0e4347f5053_4a62726f_managed_psk /* Enter Passphrase
*/
Agent RequestInput wifi_c0e4347f5053_4a62726f_managed_psk
Passphrase = [ Type=psk, Requirement=mandatory ]
Passphrase?
connmanctl> quit

To run NXP Bluetooth with BlueZ stack, execute the following commands (it requires load Wi-Fi first to load
Bluetooth firmware):

modprobe btnxpuart
hciconfig hci0 up

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

129 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Run the following commands to connect the Bluetooth device for all chips:

$ bluetoothctl
[bluetooth]# default-agent
[bluetooth]# agent on
[bluetooth]# scan on
[bluetooth]# pair xx:xx:xx:xx:xx:xx
[BT dev]# connect xx:xx:xx:xx:xx:xx
[BT dev]# quit

Note:
MuRata has two kinds of 88W9098 1XL modules with no visual difference between them. The old version 1XL
sets the initial baud rate of Bluetooth firmware to 3 Mbps, and the new version 1XL sets it to 115200 bps.
BSP release supports the new 1XL module (115200 bps) by default. To use the old 1XL module (3 Mbps), add
the fw-init-baudrate = <3000000> property in Bluetooth device node of the dts file to make it work.
The i.MX 6 boards require board rework to support the Bluetooth/Wi-Fi enablement as well as running with the
Bluetooth/Wi-Fi device tree. 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.
Rework is also required to support NXP PCIe 88W9098 on i.MX 8M Plus.
• To run NXP PCIe 88W9098 on i.MX 8M Plus, perform the hardware rework as follows:
Change R452 to 0 ohm.

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.
• 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;

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

130 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

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.
• Users can use /sys/kernel/debug/tcpm/2-0050 to check the power delivery state, which is for
debugging purpose information.
The following describes the connectivity for USB type-C and power delivery connection on the i.MX 95 EVK
board.
• The Linux release includes USB type-C and PD stack, which are enabled by default. The specific power
parameters are passed in by DTS. The following takes i.MX 95 EVK as an example:
ptn5110: tcpc@50 {
compatible = "nxp,ptn5110";
reg = <0x50>;
interrupt-parent = <&gpio5>;
interrupts = <14 IRQ_TYPE_LEVEL_LOW>;
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_ptn5110>;

typec_con: connector {
compatible = "usb-c-connector";
label = "USB-C";
power-role = "dual";
data-role = "dual";
try-power-role = "sink";
source-pdos = <PDO_FIXED(5000, 3000, PDO_FIXED_USB_COMM)>;
sink-pdos = <PDO_FIXED(5000, 3000, PDO_FIXED_USB_COMM)
PDO_VAR(5000, 20000, 3000)>;
op-sink-microwatt = <15000000>;
self-powered;
};
};
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 95
EVK board only supports 5V supply.
• Users can use /sys/kernel/debug/usb/tcpm-6-0050/log to check the power delivery state, which is
for debugging purpose information.
• Booting only by type-C port power supply is not supported.

11.3 NXP Bluetooth/Wi-Fi information

The NXP Bluetooth/Wi-Fi information is as follows:
• SoC version: SDIO 88W8987, PCIe 88W8997, SDIO 88w8997, PCIe 88w9098, SDIO 88W9098, SDIO
IW416, SDIO 88W8801, SDIO IW612
• SDIO W8801 Firmware version: 14.92.36.p192
• SDIO-UART IW416 Firmware version: 16.92.21.p137.2
• PCIE-UART W9098 Firmware version: 17.92.1.p149.155
• SDIO-UART W8997 Firmware version: 16.92.21.p137.2
• PCIE-UART W8997 Firmware version: 16.92.21.p137.2
• SDIO-UART W8987 Firmware version: 16.92.21.p137.2

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

131 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• SDIO-UART W9098 Firmware version: 17.92.1.p149.155

• SDIO-UART IW612 Firmware version: 18.99.3.p15.8
• Wi-Fi/Bluetooth firmware version: for example, 16.92.10.p210
– 16: Major revision
– 92: Feature pack
– 10: Release version
– p210: Patch number
• For all Wi-Fi modules, Driver version: MXM6x17437.p30-MGPL
– 6X: Linux 6.x
– 17437: Release version
– p30: Patch number
– MGPL: General
Tested using iPerf3 version 3.16.

11.4 Hardware limitations

i.MX 8M Mini and Nano baseboards do not route SDIO, UART, and I2S to the M.2 Key-E connector, so it is not
possible to use SDIO/UART/I2S wireless cards on these boards.
VPCIe_3V3
J601

GND1 3V3_1
1 2
M2_USB_DP USB_D+ 3V3_2
TP601 3 Key E 4
M2_USB_DN USB_D- LED1
TP603 5 6
GND2 I2S_SCK M2_PCM_CLK
7 8 TP604
M2_SD_CLK SDIO_CLK I2S_WS M2_PCM_SYNC
TP605 9 10 TP606
M2_SD_CMD SDIO_CMD I2S_SD_IN M2_PCM_IN
TP607 11 1.8 V 12 TP608
M2_SD_DAT0 SDIO_DATA0 I2S_SD_OUT M2_PCM_OUT
TP609 13 14 TP610
M2_SD_DAT1 SDIO_DATA1 LED2
TP611 15 16
M2_SD_DAT2 SDIO_DATA2 1.8 V GND3
TP612 17 18
M2_SD_DAT3 SDIO_DATA3 UART_WAKE BT_DEV_WAKE
TP613 19 20 TP614
M2_SD_WAKE SDIO_WAKE UART_RXD M2_UART_RXD
TP615 21 22 TP616
M2_SD_nRST SDIO_RST
TP617 23

GND5 1.8 V UART_TXD M2_UART_TXD

33 32 TP618
C615
0402_CC PET_P0 UART_CTS M2_UART_RTS
7_PCIE_TXP 35 34 TP619
0.1 µF, 50 V 0402_CC PET_N0 UART_RTS M2_UART_CTS
37 36 TP620
C616 VEN_DEF1
7_PCIE_TXN GND4
39 38
0.1 µF, 50 V PER_P0 VEN_DEF2
VPCIe_3V3 7_PCIE_RXP 41 M.2/NGFF 40
PER_N0 VEN_DEF3
7_PCIE_RXN 43 42
GND6 COEX3
45 44 REF_CLK_32K GPIO1_IO00
REF_CLKP_CN REFCLK_P0 COEX_RXD
DNP 47 46
R627 REF_CLKN_CN REFCLK_N0 COEX_TXD
R626
10 kΩ
10 kΩ 49 48
5% GND7 SUSCLK SUSCLK_32K R629
5% 51 50 REF_CLK_32K_3V3 7
PCIe_nCLKREQ_DEV CLKREQ0 PERST0 0 Ω, 5 %
53 52 PCIe_nRST 7
A C PEWAKE0 W_DISABLE2
7 PCIe_nWAKE 55 54 TP621
D601 GND8 W_DISABLE1
57 56 PCIe_nDIS 7
NSR0320 PET_P1 I2C_DATA R630
M2_I2C_SDA
59 58 I2C2_SDA 7, 13
PET_N1 I2C_CLK M2_I2C_SCL 0 Ω, 5 %
61 1.8 V 60
GND9 ALERT R631
63 62 I2C2_SCL 7, 13
PER_P1 RESERVED 0Ω
65 64
12 contacts PER_N1 UIM_SWP/PERST1
67 66
GND10 UIM_PWR_SNK/CLKREQ1
69 68
REFCLK_P1 UIM_PWR_SRC/PEWAKE1
Key E slot 71 70
REFCLK_N1 3V3_3
12 pins 73 72
GND11 3V3_4
75 74
Key E
GND CON_MINI CARD_RA GND
MDT420E01001
aaa-052985

Figure 15. M.2 connector schematic

Workarounds

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

132 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Use an M.2 adapter to do a physical connection to bring SDIO, UART, and I2S signals to the J1003 connector.
With this, it is required to modify the device tree to enable these new modules.

11.5 Certification

11.5.1 WFA certification

The following table lists the WFA certification.

Table 89. WFA certification

STA Certification
STA 802.11n
STA 802.11ac
STA WPS2.0
STA PMF
STA WMM-PS
STA WPA3

For details, see Wi-Fi Alliance Derivative Certification (AN12976).

11.5.2 Bluetooth controller certification

Listing details: https://launchstudio.bluetooth.com/ListingDetails/115533

12 DDR Performance Monitor

12.1 Introduction
There are counters in some i.MX 8 DDR controllers, which are used to monitor DDR signals. Some signals can
help users monitor DDR transactions and calculate DDR bandwidth.

12.2 Frequently used events

The following events are frequently used to monitor DDR transactions for different platforms.
• i.MX 8QuadMax/8QuadXPlus/8M Quad/8M Mini/8M Nano: cycles, read-cycles, write-cycles
• i.MX 8M Plus: axid-read, axid-write
• i.MX 8DXL: cycles, read-cycles, write-cycles, axid-read, axid-write
Note:
• i.MX 8M Plus and 8DXL support AXI ID filtering.
• For i.MX 8M Plus, cycles, read-cycles, write-cycles cannot be used since there is a hardware bug that leads to
counter overflow.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

133 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

12.3 Showing supported events

Run the following commands to show the supported events:

# perf list pmu | grep imx8_ddr

imx8_ddr0/activate/ [Kernel PMU event]
imx8_ddr0/axid-read/ [Kernel PMU event]
imx8_ddr0/axid-write/ [Kernel PMU event]
imx8_ddr0/cycles/ [Kernel PMU event]
imx8_ddr0/hp-read-credit-cnt/ [Kernel PMU event]
imx8_ddr0/hp-read/ [Kernel PMU event]
imx8_ddr0/hp-req-nocredit/ [Kernel PMU event]
imx8_ddr0/hp-xact-credit/ [Kernel PMU event]
imx8_ddr0/load-mode/ [Kernel PMU event]
imx8_ddr0/lp-read-credit-cnt/ [Kernel PMU event]
imx8_ddr0/lp-req-nocredit/ [Kernel PMU event]
imx8_ddr0/lp-xact-credit/ [Kernel PMU event]
imx8_ddr0/perf-mwr/ [Kernel PMU event]
imx8_ddr0/precharge/ [Kernel PMU event]
imx8_ddr0/raw-hazard/ [Kernel PMU event]
imx8_ddr0/read-accesses/ [Kernel PMU event]
imx8_ddr0/read-activate/ [Kernel PMU event]
imx8_ddr0/read-command/ [Kernel PMU event]
imx8_ddr0/read-cycles/ [Kernel PMU event]
imx8_ddr0/read-modify-write-command/ [Kernel PMU event]
imx8_ddr0/read-queue-depth/ [Kernel PMU event]
imx8_ddr0/read-write-transition/ [Kernel PMU event]
imx8_ddr0/read/ [Kernel PMU event]
imx8_ddr0/refresh/ [Kernel PMU event]
imx8_ddr0/selfresh/ [Kernel PMU event]
imx8_ddr0/wr-xact-credit/ [Kernel PMU event]
imx8_ddr0/write-accesses/ [Kernel PMU event]
imx8_ddr0/write-command/ [Kernel PMU event]
imx8_ddr0/write-credit-cnt/ [Kernel PMU event]
imx8_ddr0/write-cycles/ [Kernel PMU event]
imx8_ddr0/write-queue-depth/ [Kernel PMU event]
imx8_ddr0/write/ [Kernel PMU event]

12.4 Examples for monitoring transactions

This section shows some examples to monitor DDR transactions.
• For i.MX 8QuadMax/8QuadXPlus/8M Quad/8M Mini/8M Nano:
# perf stat -a -I 1000 -e imx8_ddr0/cycles/,imx8_ddr0/read-cycles/,imx8_ddr0/
write-cycles/
• For i.MX 8M Plus:
– All masters:
# perf stat -a -I 1000 -e imx8_ddr0/axid-read,axi_mask=0xffff/,imx8_ddr0/
axid-write,axi_mask=0xffff/
– GPU 3D:
# perf stat -a -I 1000 -e imx8_ddr0/axid-read,axi_id=0x70/,imx8_ddr0/axid-
write,axi_id=0x70/

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

134 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

– LCDIF1:
# perf stat -a -I 1000 -e imx8_ddr0/axid-read,axi_id=0x68/,imx8_ddr0/axid-
write,axi_id=0x68/
• For i.MX 8DXL:
– All masters:
# perf stat -a -I 1000 -e imx8_ddr0/cycles/,imx8_ddr0/read-cycles/,imx8_ddr0/
write-cycles/
# perf stat -a -I 1000 -e imx8_ddr0/axid-read,axi_mask=0xffff/,imx8_ddr0/
axid-write,axi_mask=0xffff/
– USB 2.0:
# perf stat -a -I 1000 -e imx8_ddr0/axid-
read,axi_mask=0xb0,axi_id=0x40b/,imx8_ddr0/axid-
write,axi_mask=0xb0,axi_id=0x40b/
– USDHC0:
# perf stat -a -I 1000 -e imx8_ddr0/axid-read,axi_id=0x1b/,imx8_ddr0/axid-
write,axi_id=0x1b/

12.5 Performance metric

Try to use metric instead of event if event command line is too cumbersome to you. The following is the
example on i.MX 8QuadXPlus.

12.5.1 Showing supported metric

Run the following commands to show the supported metric:

# perf list metric

List of pre-defined events (to be used in -e):
Metrics:
imx8qxp_bandwidth_usage.lpddr4
[bandwidth usage for lpddr4 mek board. Unit: imx8_ddr ]
imx8qxp_ddr_read.all
[bytes all masters read from ddr based on read-cycles event. Unit:
imx8_ddr ]
imx8qxp_ddr_write.all
[bytes all masters wirte to ddr based on write-cycles event. Unit:
imx8_ddr ]

12.5.2 Monitoring transactions

Run the following commands to monitor transactions:

# perf stat -a -I 1000 -M imx8qxp_ddr_read.all,imx8qxp_ddr_write.all

# time counts unit events
1.001115250 28264 imx8_ddr0/read-cycles/ # 441.6 KB imx8qxp_ddr_read.all
1.001115250 11622 imx8_ddr0/write-cycles/ # 181.6 KB imx8qxp_ddr_write.all
2.002718000 14496 imx8_ddr0/read-cycles/ # 226.5 KB imx8qxp_ddr_read.all
2.002718000 4585 imx8_ddr0/write-cycles/ # 71.6 KB imx8qxp_ddr_write.all

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

135 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

12.6 DDR Performance usage summary

It is recommended to use metric to monitor DDR transactions, as it is more convenient. You can get DDR
bandwidth of all masters or specific master directly without doing extra calculations. Especially on platforms that
support AXI ID filtering, get rid of searching masters' ID.

13 One-Time Programmable Controller Driver Using NVMEM Subsystem

13.1 Introduction
The One-Time Programmable Controller driver is realized with the NVMEM Subsystem, which introduces
DT representation for consumer devices to get the data they require (MAC addresses, SoC/Revision ID, part
numbers, and so on) from the NVMEMs.

13.2 NVMEM provider OCOTP

Use struct nvmem_config to set the configuration of theNVMEM device OCOTP. This structure can define
callback to read/write the eFUSE data.
In the read/write function prototype:
• The 1st parameter is the private data of the OCOTP device, and it contains a pointer to the remapped
memory.
• The 2nd parameter is from the first data in property reg of a NVMEM consumer, and this offset represents the
OCOTP shadow register, to which a eFuse address is mapped.
• The 3rd parameter returns the read data when reading or passes the data to write when writing.
• The 4th parameter, which indicates how many bytes to read/write, is from the second data in property reg of a
NVMEM consumer.

typedef int (*nvmem_reg_read_t)(void *priv, unsigned int offset, void *val,

size_t bytes);
typedef int (*nvmem_reg_write_t)(void *priv, unsigned int offset, void *val,
size_t bytes);

13.3 NVMEM consumer

NVMEM consumers are the entities that make use of the NVMEM provider to read from and to NVMEM. In the
DTS file, the NVMEM consumer node needs to be written in the NVMEM provider node. The indispensable
property is reg. The first data represents the offset of the OCOTP shadow register, to which a eFuse address is
mapped. The second data indicates the number of bytes to read or write.
Take the MAC address on i.MX 8M Nano as an example:
The first data in reg is 0x90, 0x400 + 0x90 * 0x4 = 0x640, 0x640 is the first Fuse address of MAC_ADDR. 0x4
represents 4 bytes.

13.4 Examples to read/write the raw NVMEM file in user space

• i.MX 6/i.MX 7/i.MX 8M Mini/8M Nano/8M Plus/8M Quad
# hexdump /sys/bus/nvmem/devices/imx-ocotp0/nvmem

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

136 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• i.MX 8ULP
# hexdump /sys/bus/nvmem/devices/fsb_s400_fuse1/nvmem
• i.MX 8M Plus
f = open('/sys/devices/platform/30350000.ocotp-ctrl/imx-ocotp0/nvmem', 'br+')
f.seek(0x90) # MAC1_ADDR
f.write(pack('<L', 0x30445011))
f.close()

14 NXP eIQ Machine Learning

The NXP eIQ machine learning software development environment enables the use of machine learning
algorithms on i.MX family SoCs. The eIQ software for i.MX includes inference engines, optimized libraries for
hardware acceleration, and an ML security package.
The main eIQ toolkit is integrated in the Yocto BSP, contained in meta-imx/meta-ml layer. The following
inference engines are currently supported: TensorFlow Lite, ONNX Runtime, OpenCV, and PyTorch. See the
i.MX Machine Learning User's Guide (UG10166) for details about the eIQ software development environment.
In addition to the toolkit integrated in the Yocto BPS, there is an ML security package delivered separately. See
also Security for Machine Learning Package (AN12867).

15 Murata Wi-Fi/Bluetooth Solutions

Table 90. Murata Wi-Fi/Bluetooth solutions with NXP chipsets
NXP chipset Murata module (Part number) Embedded Artists M.2 EVB
88W8801 Type 2DS (LBWA0ZZ2DS)

EAR00386
IW416 Type 1XK (LBEE5CJ1XK)

EAR00385
88W8987 Type 1ZM (LBEE5QD1ZM)

EAR00364

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

137 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 90. Murata Wi-Fi/Bluetooth solutions with NXP chipsets...continued

NXP chipset Murata module (Part number) Embedded Artists M.2 EVB
88W8997 Type 1YM (LBEE5XV1YM)

EAR00370
88W9098 Type 1XL (LBEE5ZZ1XL)

EAR00387

Table 91. Murata Wi-Fi/Bluetooth solutions for NXP and Embedded Artists EVKs
Embedded Artists M.2
EVK Murata module Interconnect
Module Part #
NXP i.MX 8QuadMax Type 1YM (PCIe) M.2 EAR00370
Type 1XL (PCIe) M.2 EAR00387
NXP i.MX 8QuadXPlus Type 1YM (PCIe) M.2 EAR00370
Type 1XL (PCIe) M.2 EAR00387
NXP i.MX 8M Type 1XK uSD-M.2 EAR00385
Type 1ZM uSD-M.2 EAR00364
[1]
Type 1YM (SDIO ) uSD-M.2 EAR00370
[1]
Type 1XL (SDIO ) uSD-M.2 EAR00387
Type 1YM (PCIe) M.2 EAR00370
Type 1XL (PCIe) M.2 EAR00387
NXP i.MX 8DXL Type 1YM (PCIe) M.2 EAR00370
Type 1XL (PCIe) M.2 EAR00387
NXP i.MX 8M Plus Type 1XK M.2 EAR00385
Type 1ZM M.2 EAR00364
[1]
Type 1YM (SDIO ) M.2 EAR00370
[1]
Type 1XL (SDIO ) M.2 EAR00387
Type 1YM (PCIe) M.2 EAR00370
Type 1XL (PCIe) M.2 EAR00387
NXP i.MX 8M Mini LPDDR4 Type 1XK uSD-M.2 EAR00385

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

138 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Table 91. Murata Wi-Fi/Bluetooth solutions for NXP and Embedded Artists EVKs...continued
Embedded Artists M.2
EVK Murata module Interconnect
Module Part #
Type 1ZM uSD-M.2 EAR00364
[1]
Type 1YM (SDIO ) uSD-M.2 EAR00370
[1]
Type 1XL (SDIO ) uSD-M.2 EAR00387
Type 1YM (PCIe) M.2 EAR00370
Type 1XL (PCIe) M.2 EAR00387
NXP i.MX 8M Nano LPDDR4 Type 1XK uSD-M.2 EAR00385
Type 1ZM uSD-M.2 EAR00364
[1]
Type 1YM (SDIO ) uSD-M.2 EAR00370
[1]
Type 1XL (SDIO ) uSD-M.2 EAR00387
NXP i.MX 7Dual Type 1XK uSD-M.2 EAR00385
Type 1ZM uSD-M.2 EAR00364
[1]
Type 1YM (SDIO ) uSD-M.2 EAR00370
NXP i.MX 7ULP Type 1XK uSD-M.2 EAR00385
Type 1ZM uSD-M.2 EAR00364
NXP i.MX 6QuadPlus Type 1XK uSD-M.2 EAR00385
NXP i.MX 6Quad Type 1YM (SDIO )
[1]
uSD-M.2 EAR00370
NXP i.MX 6DL
NXP i.MX 6SLL Type 1XK uSD-M.2 EAR00385
NXP i.MX 6UL Type 1ZM uSD-M.2 EAR00364
NXP i.MX 6ULL/ULZ [1]
Type 1YM (SDIO ) uSD-M.2 EAR00370

[1] Default strapping option on Embedded Artists 1YM/1XL M.2 module is WLAN-PCIe. Refer to Embedded Artists datasheet on how to modify strapping on
M.2 module for WLAN-SDIO configuration.

16 NXP i.MX SystemReady IR

16.1 Preparing binaries

There are two methods to generate binaries, with Yocto build environment, or manual build.

16.1.1 Yocto
Arm SystemReady Certification bootloader can be generated in the latest NXP i.MX Yocto Linux BSP. The build
steps are as follows.
If you are building for your own board, update imx-mkimag/iMX8M/soc.mak CAPSULE_GUID to match your
board guid.
Do not use the default one in imx-mkimage/iMX8M/soc.mak; otherwise, the SR-IR-2.0 certification will fail.

16.1.1.1 Setting up the Yocto build environment

The Yocto Build Host must meet the following requirements:

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

139 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

• Run a supported Linux distribution. Ubuntu 20.04 or later is recommended.

• Some key tools:
– Git 1.8.3.1 or higher
– tar 1.28 or higher
– Python 3.6.0 or higher
– GCC 5.0 or higher

16.1.1.2 Building Host packages for Yocto build

You must install the essential host packages on your build host. The following command installs the host
packages based on an Ubuntu distribution:

$ sudo apt install gawk wget git diffstat unzip texinfo gcc build-essential
chrpath socat cpio python3 python3-pip python3-pexpect xz-utils debianutils
iputils-ping python3-git python3-jinja2 libegl1-mesa libsdl1.2-dev pylint3
xterm python3-subunit mesa-common-dev zstd liblz4-tool

See Yocto Project Quick Build

Note: For Arm SystemReady boot loader build, the special host package should be installed:

$ sudo apt install efitools

16.1.1.3 Setting up the NXP i.MX Yocto Linux BSP

Perform the following steps to set up the NXP i.MX Yocto Linux BSP.
1. Install the repo utility.
To use this manifest repo, install the repo tool first.
$ mkdir ~/bin
$ curl http://commondatastorage.googleapis.com/git-repo-downloads/repo > ~/
bin/repo
$ chmod a+x ~/bin/repo
$ PATH=${PATH}:~/bin
2. Download the Yocto Project BSP.
$ mkdir <release>
$ cd <release>
$ repo init -u https://source.codeaurora.org/external/imx/imx-manifest -b
imx-linux-kirkstone -m imx-5.15.52-2.1.0.xml
$ repo sync
3. Set up the BSP build folder.
$ MACHINE=<imx8m_board> DISTRO=fsl-imx-xwayland source ./imx-setup-release.sh
-b build

16.1.1.4 Building the STMM bootloader

The default BSP does not enable the STMM feature for SystemReady Certification.
To enable the feature, add stmm to MACHINE_FEATURES before building imx-boot.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

140 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

1. Add the following command line to conf/local.conf.

MACHINE_FEATURES:append = " stmm"
2. Build flash.bin, capsule1.bin, and CRT.*.
$ bitbake imx-boot
3. Find them under the build folder.
• imx-boot-<imx8m_board>-sd.bin-flash_evk_stmm_capsule (flash.bin) under tmp/
deploy/images/<imx8m_board>.
• capsule1.bin and CRT.* under tmp/deploy/images/<imx8m_board>/imx-boot-tools.
Note: Save the CRT.* files for future usage.

16.1.2 Manual build

Except EDK2, use NXP repos.

16.1.2.1 Building EDK2

$ git clone https://github.com/tianocore/edk2.git

$ git clone https://github.com/tianocore/edk2-platforms.git

Modify Platform/StandaloneMm/PlatformStandaloneMmPkg/
PlatformStandaloneMmRpmb.dsc
- gEfiMdeModulePkgTokenSpaceGuid.PcdMaxAuthVariableSize|0x2800
+ gEfiMdeModulePkgTokenSpaceGuid.PcdMaxAuthVariableSize|0x3800^M

$ cd edk2
$ git submodule init && git submodule update --init --recursive
$ cd ..
$ export WORKSPACE=$(pwd)
$ export PACKAGES_PATH=$WORKSPACE/edk2:$WORKSPACE/edk2-platforms
$ export ACTIVE_PLATFORM="Platform/StandaloneMm/PlatformStandaloneMmPkg/
PlatformStandaloneMmRpmb.dsc"
$ export GCC5_AARCH64_PREFIX= aarch64-poky-linux-
$ source edk2/edksetup.sh
$ make -C edk2/BaseTools
$ build -p $ACTIVE_PLATFORM -b RELEASE -a AARCH64 -t GCC5 -n `nproc`

Copy Build/MmStandaloneRpmb/RELEASE_GCC5/FV/BL32_AP_MM.fd to the OP-TEE OS source

directory.

16.1.2.2 Building OP-TEE OS

Apply this change to scripts/nxp_build.sh.

@@ -95,6 +95,18 @@ function build()

CROSS_COMPILE64="$CROSS_COMPILE64" \
CFG_TEE_CORE_LOG_LEVEL="$CFG_TEE_CORE_LOG_LEVEL" \
CFG_TEE_TA_LOG_LEVEL="$CFG_TEE_TA_LOG_LEVEL" \
+ CFG_STMM_PATH=BL32_AP_MM.fd \
+ CFG_RPMB_FS=y \
+ CFG_IMX_SNVS=n \
+ CFG_NXP_CAAM=n \
+ CFG_RPMB_WRITE_KEY=y \
+ CFG_RPMB_FS_DEV_ID=2 \
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

141 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

+ CFG_CORE_DYN_SHM=y CFG_RPMB_TESTKEY=y \
+ CFG_REE_FS=n \
+ CFG_SCTLR_ALIGNMENT_CHECK=n \
+ CFG_CORE_HEAP_SIZE=2097152 \
+ CFG_TEE_RAM_VA_SIZE=4194304 \
+ CFG_PREALLOC_RPC_CACHE=n \
CFG_WERROR=y \
PLATFORM="$plat" \
O="$O"/build."$plat" \

To i.MX 8M Quad EVK, the CFG_RPMB_FS_DEV_ID should be 0. For i.MX 8M Nano/Mini/Plus EVK, it is 2 as
follows:

@@ -95,6 +95,18 @@ function build()

CROSS_COMPILE64="$CROSS_COMPILE64" \
CFG_TEE_CORE_LOG_LEVEL="$CFG_TEE_CORE_LOG_LEVEL" \
CFG_TEE_TA_LOG_LEVEL="$CFG_TEE_TA_LOG_LEVEL" \
+ CFG_STMM_PATH=BL32_AP_MM.fd \
+ CFG_RPMB_FS=y \
+ CFG_IMX_SNVS=n \
+ CFG_NXP_CAAM=n \
+ CFG_RPMB_WRITE_KEY=y \
+ CFG_RPMB_FS_DEV_ID=2 \
+ CFG_CORE_DYN_SHM=y CFG_RPMB_TESTKEY=y \
+ CFG_REE_FS=n \
+ CFG_SCTLR_ALIGNMENT_CHECK=n \
+ CFG_CORE_HEAP_SIZE=2097152 \
+ CFG_TEE_RAM_VA_SIZE=4194304 \
+ CFG_PREALLOC_RPC_CACHE=n \
CFG_WERROR=y \
PLATFORM="$plat" \
O="$O"/build."$plat" \

Build cmd.
Do not source the Yocto toolchain setup file; otherwise, the build fails. Take the following script as a reference.

#!/bin/sh
export GCC_PATH=/home/Freenix/tools/fsl-imx-internal-xwayland/5.15-kirkstone/
export SDKTARGETSYSROOT=$GCC_PATH/sysroots/cortexa72-cortexa53-crypto-poky-linux
export PATH=$PATH:$GCC_PATH/sysroots/x86_64-pokysdk-linux/usr/bin/aarch64-poky-
linux/
export CC="aarch64-poky-linux-gcc -mcpu=cortex-a72.cortex-a53 -march=armv8-a
+crc+crypto -fstack-protector-strong -O2 -D_FORTIFY_SOURCE=2 -Wformat -Wformat-
security -Werror=format-security --sysroot=$SDKTARGETSYSROOT"
export CXX="aarch64-poky-linux-g++ -mcpu=cortex-a72.cortex-a53 -march=armv8-a
+crc+crypto -fstack-protector-strong -O2 -D_FORTIFY_SOURCE=2 -Wformat -Wformat-
security -Werror=format-security --sysroot=$SDKTARGETSYSROOT"
export CPP="aarch64-poky-linux-gcc -E -mcpu=cortex-a72.cortex-a53 -march=armv8-
a+crc+crypto -fstack-protector-strong -O2 -D_FORTIFY_SOURCE=2 -Wformat -
Wformat-security -Werror=format-security --sysroot=$SDKTARGETSYSROOT"
export AS="aarch64-poky-linux-as "
export LD="aarch64-poky-linux-ld --sysroot=$SDKTARGETSYSROOT"

./scripts/nxp_build.sh imx-mx8mnevk [For i.MX8MN-EVK]

./scripts/nxp_build.sh imx-mx8mmevk [For i.MX8MM-EVK]

./scripts/nxp_build.sh imx-mx8mpevk [For i.MX8MP-EVK]

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

142 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

./scripts/nxp_build.sh imx-mx8mqevk [For i.MX8MQ-EVK Remember to update

CFG_RPMB_FS_DEV_ID before run this]

Copy tee-raw.bin to imx-mkimage/iMX8M/ for the boards to be built with flash.bin/capsule1.bin

and rename it to tee.bin-stmm.

16.1.2.3 Building U-Boot

#!/bin/sh
. /home/Freenix/tools/fsl-imx-internal-xwayland/5.15-kirkstone/environment-
setup-cortexa72-cortexa53-crypto-poky-linux
make imx8mm_evk_defconfig O=imx8mmevk
make O=imx8mmevk -j4
make imx8mp_evk_defconfig O=imx8mpevk
make O=imx8mpevk -j4
make imx8mn_evk_defconfig O=imx8mnevk
make O=imx8mnevk -j4
make imx8mq_evk_defconfig O=imx8mqevk
make O=imx8mqevk -j4

16.1.2.4 Building ATF

make PLAT=imx8mp SPD=opteed bl31

make PLAT=imx8mn SPD=opteed bl31
make PLAT=imx8mm SPD=opteed bl31
make PLAT=imx8mq SPD=opteed bl31

16.1.2.5 Generating flash.bin and capsule1.bin

Generate the key:

make SOC=iMX8MP TEE=tee.bin-stmm capsule_key

There is no need to generate the capsule key every time. Save the CRT.* files to a backup folder. The files are
required when generating capsule1.bin. If these files are lost, regenerate flash.bin and capsule1.bin.
If your flash.bin and capsule1.bin use different CRT files, capsule update will fail.

Take i.MX 8M Plus as an example:

cp $U_BOOT/imx8mpevk/u-boot-nodtb.bin imx-mkimage/iMX8M
cp $U_BOOT/imx8mpevk/spl/u-boot-spl.bin imx-mkimage/iMX8M
cp $U_BOOT/imx8mpevk/arch/arm/dts/imx8mp-evk.dtb imx-mkimage/iMX8M

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

143 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

cp $ATF/build/imx8mp/release/bl31.bin imx-mkimage/iMX8M
cp $OPTEE/build.imx-mx8mpevk/core/tee-raw.bin imx-mkimage/iMX8M/tee.bin-stmm
Copy the ddr firmware to imx-mkimage/iMX8M/
make SOC=iMX8MP clean; make SOC=iMX8MP TEE=tee.bin-stmm flash_evk_stmm_capsule
[Save flash.bin, if you wanna generate capsule1.bin with other binaries, such
as updated u-boot/atf/tee, copy new binaries and rerun this command, and save
capsule1.bin]

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

144 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

16.2 Running capsule update

Perform the following steps to run the capsule update.
1. Set the board to serial download mode. Burn flash.bin to eMMC with uuu -b emmc flash.bin. The
flash.bin should be the STMM enabled ones.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

145 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

2. Download https://github.com/ARM-software/arm-systemready/blob/main/IR/prebuilt_images/v24.03_2.1.1/ir-
acs-live-image-generic-arm64.wic.xz.
3. Burn acs.img to the SD card and plug in the SD card to the board.

4. Set the board to boot from eMMC, and check the board silkscreen to set the boot switch correctly.
5. (Optional) Clear the eMMC key if necessary.
Note:
• Once you burn the key in Step 7, you do not need to clear the key again every time.
• The key is the OP-TEE test key for test purposes. For production, use your own key.
=>mw 0x60000000 0xc33eebd3;mw 0x60000004 0x9f4c336e;mw 0x60000008
0xc0e28c98;mw 0x6000000c 0x615459b8;mw 0x60000010 0x86cf2b0d;mw 0x60000014
0xf24d8464;mw 0x60000018 0xc6e656ab;mw 0x6000001c 0xe401b71b
=>mw.b 0x50000000 0 0x400000
=>mmc rpmb write 0x50000000 0 0x2000 0x60000000
For i.MX 93 11x11 EVK:
=>mw 0x80000000 0xc33eebd3;mw 0x80000004 0x9f4c336e;mw 0x80000008
0xc0e28c98;mw 0x8000000c 0x615459b8;mw 0x80000010 0x86cf2b0d;mw 0x80000014
0xf24d8464;mw 0x80000018 0xc6e656ab;mw 0x8000001c 0xe401b71b
=>mw.b 0x90000000 0 0x400000
=>mmc rpmb write 0x90000000 0 0x2000 0x80000000

6. Set dfu_alt_info for i.MX 8M Plus EVK.

=>env set dfu_alt_info "mmc 2=1 raw 0x0 0x2000 mmcpart 1"
=>saveenv
For i.MX 8M Quad EVK:
=>env set dfu_alt_info "mmc 0=1 raw 0x42 0x2000 mmcpart 1"
For i.MX 8M Mini EVK:
=>env set dfu_alt_info "mmc 2=1 raw 0x42 0x2000 mmcpart 1"
For i.MX 8M Nano/Plus EVK:
=>env set dfu_alt_info "mmc 2=1 raw 0x0 0x2000 mmcpart 1"
For i.MX 93 11x11 EVK:
=>env set dfu_alt_info "mmc 0=1 raw 0x0 0x2000 mmcpart 1"
=> saveenv
=> reset
7. Burn the key after the U-Boot is up.
=>fatload mmc 1:1 0x40480000 /security-interface-extension-keys/TestPK1.auth
=>setenv -e -nv -bs -rt -at -i 0x40480000:$filesize PK
=>fatload mmc 1:1 0x40480000 /security-interface-extension-keys/TestKEK1.auth
=>setenv -e -nv -bs -rt -at -i 0x40480000:$filesize KEK
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

146 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

=>fatload mmc 1:1 0x40480000 /security-interface-extension-keys/TestDB1.auth

=>setenv -e -nv -bs -rt -at -i 0x40480000:$filesize db
=>fatload mmc 1:1 0x40480000 /security-interface-extension-keys/TestDBX1.auth
=>setenv -e -nv -bs -rt -at -i 0x40480000:$filesize dbx
For i.MX 93 11x11 EVK:
=>fatload mmc 1:1 0x80480000 /security-interface-extension-keys/TestPK1.auth
=>setenv -e -nv -bs -rt -at -i 0x80480000:$filesize PK
=>fatload mmc 1:1 0x80480000 /security-interface-extension-keys/TestKEK1.auth
=>setenv -e -nv -bs -rt -at -i 0x80480000:$filesize KEK
=>fatload mmc 1:1 0x80480000 /security-interface-extension-keys/TestDB1.auth
=>setenv -e -nv -bs -rt -at -i 0x80480000:$filesize db
=>fatload mmc 1:1 0x80480000 /security-interface-extension-keys/TestDBX1.auth
=>setenv -e -nv -bs -rt -at -i 0x80480000:$filesize dbx

8. Store capsule1.bin to the eMMC or SD EFI system partition.

Note: It must be the EFI system partition; otherwise, capsule-update exits silently. The user can reuse
the SD card in Step 8. Format the first partition as the EFI system partition and create a directory: EFI/
UpdateCapsule/.
=>setenv serverip 10.193.108.171;dhcp $loadaddr capsule1.bin-imx8mpevk; [8MP]
=>setenv serverip 10.193.108.171;dhcp $loadaddr capsule1.bin-imx8mnevk; [8MN]
=>setenv serverip 10.193.108.171;dhcp $loadaddr capsule1.bin-imx8mmevk; [8MM]
=>setenv serverip 10.193.108.171;dhcp $loadaddr capsule1.bin-imx8mqevk; [8MQ]
If capsule1.bin is put in sd/emmc:
=>fatwrite mmc [X]:1 ${loadaddr} /EFI/UpdateCapsule/capsule1.bin 0x
${filesize}
[X] should be 2 for eMMC, and 1 for SD. Except i.MX 8M Quad and i.MX 93, 0 for eMMC, 1 for SD.
9. If [X] is set to 1 in Step 8:
=> efidebug boot add -b 0 Boot0000 mmc 1:1 capsule1.bin;efidebug boot next 0
If [X] is set to 2 in Step 8:
=> efidebug boot add -b 0 Boot0000 mmc 2:1 capsule1.bin;efidebug boot next 0
If [X] is set to 0 in Step 8, only for i.MX 8M Quad or i.MX 93 eMMC:
=> efidebug boot add -b 0 Boot0000 mmc 0:1 capsule1.bin;efidebug boot next 0
It depends on where capsule1.bin is stored.
• For i.MX 8M Mini/Nano/Plus EVK: eMMC is index 2, and SD is index 1.
• For i.MX 8M Quad or i.MX 93 EVK: eMMC is index 0, and SD is index 1.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

147 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

10. The following command can also be used to update and reset the board:
=>setenv -e -nv -bs -rt -v OsIndications =0x0000000000000004
=>efidebug capsule disk-update
When this command is executed:
u-boot=> efidebug capsule disk-update
Applying the capsule capsule1.bin succeeded.
Reboot after firmware update resetting.

16.3 Running the ACS Test

Note: flash.bin should be STMM enabled.
1. Download https://github.com/ARM-software/arm-systemready/blob/main/IR/prebuilt_images/v24.03_2.1.1/ir-
acs-live-image-generic-arm64.wic.xz.
2. Clear the key burned in Section 16.2.
U-Boot=>mw 0x60000000 0xc33eebd3;mw 0x60000004 0x9f4c336e;mw 0x60000008
0xc0e28c98;mw 0x6000000c 0x615459b8;mw 0x60000010 0x86cf2b0d;mw 0x60000014
0xf24d8464;mw 0x60000018 0xc6e656ab;mw 0x6000001c 0xe401b71b
=>mw.b 0x50000000 0 0x400000
=>mmc rpmb write 0x50000000 0 0x2000 0x60000000

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

148 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

3. Burn the ACS image to the SD card.

$sudo dd if=ir-acs-live-image-generic-arm64.wic of=/dev/sdb bs=128M conv=sync
status=progress
4. Boot the board from eMMC. It takes several hours to finish the test. Store the console log named as acs-
console.log.
5. After the test is complete, copy acs_results to the Host PC.
6. Check the SCT result.
a. Clone https://gitlab.arm.com/systemready/edk2-test-parser, branch: main.
b. Run the test as follows:
./parser.py --config EBBR.yaml ../acs_results/sct_results/Overall/
Summary.ekl ../acs_results/sct_results/Sequence/EBBR.seq
INFO ident_seq: Identified `../acs_results/sct_results/Sequence/EBBR.seq'
as "EBBR.seq from ACS-IR v22.10_2.0.0_BETA-1 .. v23.09_2.1.0".
INFO apply_rules: Updated 69 tests out of 10793 after applying 157 rules
INFO print_summary: 0 dropped, 0 failure, 55 ignored, 1 known acs
limitation, 13 known u-boot limitations, 10724 pass, 0 warning
7. Run the dt-schema check.
a. Install the dt-schema tools from https://github.com/devicetree-org/dt-schema.
b. Download the linux-next tree.
c. Get the latest SystemReady scripts from https://gitlab.arm.com/systemready/systemready-scripts.
d. Copy BsaDevTree.dtb out from Step 6 acs_results/uefi/BsaDevTree.dtb.
e. Run the following scripts:
dt-validate -s Documentation/devicetree/bindings -m BsaDevTree.dtb |& tee
log
./systemready-scripts/compatibles Documentation/devicetree/bindings >
compatibles.txt
systemready-scripts/dt-parser.py --compatibles compatibles.txt --print log
f. The result is as follows, without error:
Results as below, no error:
INFO parse: 31 entries
INFO dedupe_entries: De-duplicated 0 entries
INFO apply_rules: Updated 10 entries with rules

Summary
-------
21 dt-validate warning
3 dt-validate warning missing property
7 dt-validate warning naming

Non-ignored entries
-------------------

16.4 Distribution installation

Install from the SDCard (ISO) to eMMC. After the installation is complete, eject your SD card, and then switch
off/on the power on the board.
For i.MX 8M Quad EVK or i.MX 93 11x11 EVK:

setenv video_off yes; saveenv; reset;

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

149 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

16.4.1 Fedora
Perform the following steps:
1. Download the Fedora 36 Workstation or newer version.
• https://getfedora.org/en/iot/download/
• https://download.fedoraproject.org/pub/alt/iot/36/IoT/aarch64/iso/Fedora-IoT-ostree-
aarch64-36-20220618.0.iso
2. Burn the ISO to an SD card, and then boot the board from eMMC.
When the Fedora installer starts up, select what needs to be configured. After all configurations are
complete, enter b to start the installation.
When Fedora reports “Failed to set new efi boot target. This is most likely a kernel or firmware bug”,
respond with “yes” and continue. Ignore it, and it will not block the installation and boot.
3. After the installation is complete, press Enter to auto reboot.

16.4.2 OpenSuse
Perform the following steps:
1. Download openSUSE Leap 15.4 or newer version: https://get.opensuse.org/leap/15.4/#download/
openSUSE-Leap-15.4-DVD-aarch64-Build243.2-Media.iso
2. For i.MX 8M Quad EVK, use https://download.opensuse.org/ports/aarch64/tumbleweed/iso/openSUSE-
Tumbleweed-DVD-aarch64-Snapshot20220719-Media.iso?mirrorlist=
openSUSE-Leap 15.4 is not able to recognize eMMC on i.MX 8M Quad EVK, so use Tumbleweed.
3. Burn it to the SD card, and boot it from eMMC.
4. After the install is complete, remove the installation media.
5. Switch off/on the power on the board

16.5 Result check

Perform the following steps to check the result:
1. Use the following template to store the result:
https://gitlab.arm.com/systemready/systemready-ir-template
2. Run the SystemReady scripts:
a. Clone [email protected]:systemready/systemready-scripts.git
b. Run check-sr-results.py.
For more information, see README.md.

17 Note About the Source Code in the Document

Example code shown in this document has the following copyright and BSD-3-Clause license:
Copyright 2024 NXP Redistribution and use in source and binary forms, with or without modification, are
permitted provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions and the
following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the
following disclaimer in the documentation and/or other materials provided with the distribution.
3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or
promote products derived from this software without specific prior written permission.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

150 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY
EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT
SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
DAMAGE.

18 Revision History
This table provides the revision history.

Revision history
Document ID Release date Description
UG10163 v.LF6.6.36_2.1.0 30 September Upgraded to the 6.6.36 kernel.
2024
IMXLUG_6.6.23_2.0.0 26 July 2024 Updated a command line in Section 10.9.1.3.
IMXLUG_6.6.23_2.0.0 28 June 2024 Upgraded to the 6.6.23 kernel, U-Boot v2024.04, TF-A
v2.10, OP-TEE 4.2.0, Yocto 5.0 Scarthgap, and added the
i.MX 91 as Alpha quality, i.MX 95 as Beta quality.
IMXLUG v.LF6.6.3_1.0.0 29 March 2024 Upgraded to the 6.6.3 kernel, removed the i.MX 91P, and
added the i.MX 95 as Alpha Quality.
IMXLUG v.LF6.1.55_2.2.0 12/2023 Updated a link in Section 10.9.1.
IMXLUG v.LF6.1.55_2.2.0 12/2023 Upgraded to the 6.1.55 kernel.
IMXLUG v.LF6.1.36_2.1.0 09/2023 Upgraded to the 6.1.36 kernel.
IMXLUG v.LF6.1.22_2.0.0 08/2023 Changed “0x9F800” to "0x9F8000" in Section Section 4.3.
IMXLUG v.LF6.1.22_2.0.0 07/2023 Updated a description in Section Section 4.3.4.
IMXLUG v.LF6.1.22_2.0.0 06/2023 Upgraded to the 6.1.22 kernel.
IMXLUG v.LF6.1.1_1.0.00 03/2023 Upgraded to the 6.1.1 kernel.
IMXLUG v.LF5.15.71_2.2.0 12/2022 Upgraded to the 5.15.71 kernel.
IMXLUG v.LF5.15.52_2.1.0 09/2022 Upgraded to the 5.15.52 kernel, and added the i.MX 93.
IMXLUG v.LF5.15.32_2.0.0 06/2022 Upgraded to the 5.15.32 kernel, U-Boot 2022.04, and
Kirkstone Yocto.
IMXLUG v.LF5.15.5_1.0.0 03/2022 Upgraded to the 5.15.5 kernel, Honister Yocto, and Qt6.
IMXLUG v.LF5.10.72_2.2.0 12/2021 Upgraded the kernel to 5.10.72 and updated the BSP.
IMXLUG v.LF5.10.52_2.1.0 10/2021 Added an appendix for Murata Wi-Fi/Bluetooth solutions.
IMXLUG v.LF5.10.52_2.1.0 09/2021 Updated for i.MX 8ULP Alpha and the kernel upgraded to
5.10.52.
IMXLUG v.LF5.10.35_2.0.0 06/2021 Upgraded Yocto Project to Hardknott and the kernel
upgraded to 5.10.35.
IMXLUG v.LF5.10.9_1.0.0 03/2021 Upgraded Yocto Project to Gatesgarth and the kernel
upgraded to 5.10.9.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

151 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Revision history...continued
Document ID Release date Description
IMXLUG v.L5.4.70_2.3.0 01/2021 Updated the command lines in Section "Running the Arm
Cortex-M4 image".
IMXLUG v.L5.4.70_2.3.0 12/2020 i.MX 5.4 consolidated GA for release i.MX boards including i.
MX 8M Plus and i.MX 8DXL.
IMXLUG v.L5.4.47_2.2.0 09/2020 i.MX 5.4 Beta2 release for i.MX 8M Plus, Beta for 8DXL, and
consolidated GA for released i.MX boards.
IMXLUG v.L5.4.24_2.1.0 06/2020 i.MX 5.4 Beta release for i.MX 8M Plus, Alpha2 for 8DXL,
and consolidated GA for released i.MX boards.
IMXLUG v.L5.4.3_2.0.0 04/2020 i.MX 5.4 Alpha release for i.MX 8M Plus and 8DXL EVK
boards.
IMXLUG v.LF5.4.3_1.0.0 03/2020 i.MX 5.4 Kernel and Yocto Project Upgrades.
IMXLUG v.L4.19.35_1.1.0 10/2019 i.MX 4.19 Kernel and Yocto Project Upgrades.
IMXLUG v.L4.19.35_1.0.0 07/2019 i.MX 4.19 Beta Kernel and Yocto Project Upgrades.
IMXLUG v.L4.14.98_2.0.0_ga 04/2019 i.MX 4.14 Kernel upgrade and board updates.
IMXLUG v.L4.14.78_1.0.0_ga 01/2019 i.MX 6, i.MX 7, i.MX 8 family GA release.
IMXLUG v.L4.14.62_1.0.0_beta 11/2018 i.MX 4.14 Kernel Upgrade, Yocto Project Sumo upgrade.
IMXLUG v.L4.9.123_2.3.0_8mm 09/2018 i.MX 8M Mini GA release.
IMXLUG v.L4.9.88_2.2.0_8qxp-beta2 07/2018 i.MX 8QuadXPlus Beta2 release.
IMXLUG v.L4.9.88_2.1.0_8mm-alpha 06/2018 i.MX 8M Mini Alpha release.
IMXLUG v.L4.9.88_2.0.0-ga 05/2018 i.MX 7ULP and i.MX 8M Quad GA release.
IMXLUG v.L4.9.51_imx8mq-ga 03/2018 Added i.MX 8M Quad GA.
IMXLUG v.L4.9.51_8qm-beta2/8qxp- 02/2018 Added i.MX 8QuadMax Beta2 and i.MX 8QuadXPlus Beta.
beta
IMXLUG v.L4.9.51_imx8mq-beta 12/2017 Added i.MX 8M Quad.
IMXLUG v.L4.9.51_imx8qm-beta1 12/2017 Added i.MX 8QuadMax.
IMXLUG v.L4.9.51_imx8qxp-alpha 11/2017 Initial release.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

152 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Legal information
Definitions Terms and conditions of commercial sale — NXP Semiconductors
products are sold subject to the general terms and conditions of commercial
sale, as published at https://www.nxp.com/profile/terms, unless otherwise
Draft — A draft status on a document indicates that the content is still agreed in a valid written individual agreement. In case an individual
under internal review and subject to formal approval, which may result agreement is concluded only the terms and conditions of the respective
in modifications or additions. NXP Semiconductors does not give any agreement shall apply. NXP Semiconductors hereby expressly objects to
representations or warranties as to the accuracy or completeness of applying the customer’s general terms and conditions with regard to the
information included in a draft version of a document and shall have no purchase of NXP Semiconductors products by customer.
liability for the consequences of use of such information.
Export control — This document as well as the item(s) described herein
may be subject to export control regulations. Export might require a prior
Disclaimers authorization from competent authorities.

Limited warranty and liability — Information in this document is believed Suitability for use in non-automotive qualified products — Unless
to be accurate and reliable. However, NXP Semiconductors does not give this document expressly states that this specific NXP Semiconductors
any representations or warranties, expressed or implied, as to the accuracy product is automotive qualified, the product is not suitable for automotive
or completeness of such information and shall have no liability for the use. It is neither qualified nor tested in accordance with automotive testing
consequences of use of such information. NXP Semiconductors takes no or application requirements. NXP Semiconductors accepts no liability for
responsibility for the content in this document if provided by an information inclusion and/or use of non-automotive qualified products in automotive
source outside of NXP Semiconductors. equipment or applications.
In no event shall NXP Semiconductors be liable for any indirect, incidental, In the event that customer uses the product for design-in and use in
punitive, special or consequential damages (including - without limitation - automotive applications to automotive specifications and standards,
lost profits, lost savings, business interruption, costs related to the removal customer (a) shall use the product without NXP Semiconductors’ warranty
or replacement of any products or rework charges) whether or not such of the product for such automotive applications, use and specifications, and
damages are based on tort (including negligence), warranty, breach of (b) whenever customer uses the product for automotive applications beyond
contract or any other legal theory. NXP Semiconductors’ specifications such use shall be solely at customer’s
own risk, and (c) customer fully indemnifies NXP Semiconductors for any
Notwithstanding any damages that customer might incur for any reason
liability, damages or failed product claims resulting from customer design and
whatsoever, NXP Semiconductors’ aggregate and cumulative liability
use of the product for automotive applications beyond NXP Semiconductors’
towards customer for the products described herein shall be limited in
standard warranty and NXP Semiconductors’ product specifications.
accordance with the Terms and conditions of commercial sale of NXP
Semiconductors.
HTML publications — An HTML version, if available, of this document is
provided as a courtesy. Definitive information is contained in the applicable
Right to make changes — NXP Semiconductors reserves the right to
document in PDF format. If there is a discrepancy between the HTML
make changes to information published in this document, including without
document and the PDF document, the PDF document has priority.
limitation specifications and product descriptions, at any time and without
notice. This document supersedes and replaces all information supplied prior
Translations — A non-English (translated) version of a document, including
to the publication hereof.
the legal information in that document, is for reference only. The English
version shall prevail in case of any discrepancy between the translated and
Suitability for use — NXP Semiconductors products are not designed,
English versions.
authorized or warranted to be suitable for use in life support, life-critical or
safety-critical systems or equipment, nor in applications where failure or
Security — Customer understands that all NXP products may be subject to
malfunction of an NXP Semiconductors product can reasonably be expected
unidentified vulnerabilities or may support established security standards or
to result in personal injury, death or severe property or environmental
specifications with known limitations. Customer is responsible for the design
damage. NXP Semiconductors and its suppliers accept no liability for
and operation of its applications and products throughout their lifecycles
inclusion and/or use of NXP Semiconductors products in such equipment or
to reduce the effect of these vulnerabilities on customer’s applications
applications and therefore such inclusion and/or use is at the customer’s own
and products. Customer’s responsibility also extends to other open and/or
risk.
proprietary technologies supported by NXP products for use in customer’s
applications. NXP accepts no liability for any vulnerability. Customer should
Applications — Applications that are described herein for any of these
regularly check security updates from NXP and follow up appropriately.
products are for illustrative purposes only. NXP Semiconductors makes no
representation or warranty that such applications will be suitable for the Customer shall select products with security features that best meet rules,
specified use without further testing or modification. regulations, and standards of the intended application and make the
ultimate design decisions regarding its products and is solely responsible
Customers are responsible for the design and operation of their
for compliance with all legal, regulatory, and security related requirements
applications and products using NXP Semiconductors products, and NXP
concerning its products, regardless of any information or support that may be
Semiconductors accepts no liability for any assistance with applications or
provided by NXP.
customer product design. It is customer’s sole responsibility to determine
whether the NXP Semiconductors product is suitable and fit for the NXP has a Product Security Incident Response Team (PSIRT) (reachable
customer’s applications and products planned, as well as for the planned at [email protected]) that manages the investigation, reporting, and solution
application and use of customer’s third party customer(s). Customers should release to security vulnerabilities of NXP products.
provide appropriate design and operating safeguards to minimize the risks
associated with their applications and products. NXP B.V. — NXP B.V. is not an operating company and it does not distribute
NXP Semiconductors does not accept any liability related to any default, or sell products.
damage, costs or problem which is based on any weakness or default
in the customer’s applications or products, or the application or use by
customer’s third party customer(s). Customer is responsible for doing all Trademarks
necessary testing for the customer’s applications and products using NXP
Semiconductors products in order to avoid a default of the applications
Notice: All referenced brands, product names, service names, and
and the products or of the application or use by customer’s third party
trademarks are the property of their respective owners.
customer(s). NXP does not accept any liability in this respect.
NXP — wordmark and logo are trademarks of NXP B.V.

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

153 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

Contents
1 Overview ........................................................... 2 4.7 Running Linux OS on the target ...................... 42
1.1 Audience ............................................................ 2 4.7.1 Running the image from NAND ....................... 44
1.2 Conventions ....................................................... 2 4.7.2 Running Linux OS from Parallel NOR ............. 45
1.3 Supported hardware SoCs and boards ............. 2 4.7.3 Running the Linux OS image from QuadSPI ... 45
1.4 References .........................................................3 4.7.4 Running the Arm Cortex-M4/7/33 image ......... 45
2 Introduction ...................................................... 5 4.7.5 Linux OS login ................................................. 48
3 Basic Terminal Setup ...................................... 5 4.7.6 Running Linux OS from MMC/SD ....................48
4 Booting Linux OS ............................................ 5 4.7.7 Running the Linux image from NFS ................ 49
4.1 Software overview ............................................. 6 4.8 Arm SystemReady-IR ...................................... 50
4.1.1 Bootloader ..........................................................7 4.8.1 Arm SystemReady-IR ACS compliance test .... 50
4.1.2 Linux kernel image and device tree ...................7 4.8.2 Capsule update ............................................... 50
4.1.3 Root file system .................................................8 4.8.3 Linux distro installation .................................... 50
4.2 Universal update utility ...................................... 8 5 Enabling Solo Emulation .............................. 51
4.2.1 Downloading UUU ............................................. 8 6 Power Management ....................................... 51
4.2.2 Using UUU .........................................................8 6.1 Suspend and resume ...................................... 51
4.3 Preparing an SD/MMC card to boot .................. 9 6.2 CPU frequency scaling .................................... 52
4.3.1 Preparing the card ...........................................10 6.3 Bus frequency scaling ..................................... 52
4.3.2 Copying the full SD card image .......................10 7 Multimedia ...................................................... 54
4.3.3 Partitioning the SD/MMC card ......................... 11 7.1 i.MX multimedia packages ...............................54
4.3.4 Copying a bootloader image ............................11 7.2 Building limited access packages .................... 54
4.3.5 Copying the kernel image and DTB file ........... 12 7.3 Multimedia use cases ...................................... 54
4.3.6 Copying the root file system (rootfs) ................ 13 7.3.1 Playback use cases .........................................54
4.4 Downloading images ....................................... 13 7.3.1.1 Audio-only playback .........................................55
4.4.1 Downloading images using U-Boot ..................13 7.3.1.2 Video-only playback .........................................55
4.4.1.1 Flashing an Arm Cortex-M4 image on 7.3.1.3 Audio/Video file playback .................................55
QuadSPI .......................................................... 14 7.3.1.4 Multichannel audio playback ............................56
4.4.1.2 Downloading an image to MMC/SD .................15 7.3.1.5 Other methods for playback ............................ 56
4.4.1.3 Using eMMC ....................................................16 7.3.1.6 Video playback to multiple displays ................. 56
4.4.1.4 Flashing U-Boot on SPI-NOR from U-Boot ...... 19 7.3.2 Audio encoding ................................................57
4.4.1.5 Flashing U-Boot on Parallel NOR from U- 7.3.3 Video encoding ................................................ 57
Boot ..................................................................20 7.3.4 Transcoding ..................................................... 59
4.4.2 Using an i.MX board as the host server to 7.3.5 Audio recording ............................................... 59
create a rootfs ................................................. 21 7.3.6 Video recording ................................................60
4.5 How to boot the i.MX boards ...........................23 7.3.7 Audio/Video recording ..................................... 61
4.5.1 Booting from an SD card in slot SD1 ...............24 7.3.8 Camera preview .............................................. 61
4.5.2 Booting from an SD card in slot SD2 ...............24 7.3.9 Libcamera ........................................................ 62
4.5.3 Booting from an SD card in slot SD3 ...............26 7.3.9.1 Overview .......................................................... 62
4.5.4 Booting from an SD card in slot SD4 ...............26 7.3.9.2 Neo ISP Image Processing Algorithm ............. 63
4.5.5 Booting from eMMC ........................................ 27 7.3.9.3 Camera modules ............................................. 65
4.5.6 Booting from SATA .......................................... 29 7.3.9.4 Libcamera configuration .................................. 66
4.5.7 Booting from NAND ......................................... 29 7.3.9.5 GStreamer pipelines ........................................ 67
4.5.8 Booting from SPI-NOR .................................... 30 7.3.9.6 cam test application .........................................68
4.5.9 Booting from EIM (Parallel) NOR .................... 30 7.3.9.7 Limitations ........................................................70
4.5.10 Booting from QuadSPI or FlexSPI ................... 30 7.3.10 Recording the TV-in source .............................70
4.5.11 Serial download mode for the 7.3.11 Web camera .................................................... 70
Manufacturing Tool .......................................... 32 7.3.12 HTTP streaming .............................................. 71
4.5.12 How to build U-Boot and Kernel in 7.3.13 HTTP live streaming ........................................ 71
standalone environment .................................. 34 7.3.14 MPEG-DASH streaming .................................. 71
4.5.13 How to build imx-boot image by using imx- 7.3.15 Real Time Streaming Protocol (RTSP)
mkimage .......................................................... 37 playback ...........................................................71
4.6 Flash memory maps ........................................ 41 7.3.16 RTP/UDP MPEGTS streaming ........................ 72
4.6.1 MMC/SD/SATA memory map .......................... 41 7.3.17 RTSP streaming server ................................... 73
4.6.2 NAND flash memory map ................................41 7.3.18 Video conversion ............................................. 74
4.6.3 Parallel NOR flash memory map ..................... 41 7.3.19 Video composition ........................................... 77
4.6.4 SPI-NOR flash memory map ........................... 42 7.4 PipeWire input/output settings ......................... 78
4.6.5 QuadSPI flash memory map ........................... 42 7.4.1 PipeWire settings .............................................78
UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

154 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

7.5 Installing gstreamer1.0-libav into rootfs ........... 79 10.4.7.5 Running the OPENSSL Kernel TLS test ........ 107
8 Audio ...............................................................79 10.5 Disk encryption acceleration ..........................107
8.1 DSP support .................................................... 80 10.5.1 Enabling disk encryption support in kernel
8.1.1 HiFi 4 DSP framework .....................................80 for the platform containing CAAM hardware
8.1.2 Sound Open Firmware .................................... 80 IP ....................................................................108
8.2 HDMI eARC support ........................................80 10.5.2 User space tools for disk encryption ..............109
8.3 Low power voice UI .........................................81 10.5.3 DM-Crypt using CAAM backed keys ............. 109
8.3.1 Introduction ...................................................... 81 10.5.3.1 DM-Crypt with Trusted keys backed by
8.3.2 Standard voice solution ................................... 82 CAAM .............................................................110
8.3.3 Audio Front End (AFE) .................................... 83 10.5.3.2 DM-Crypt with CAAM’s tagged key ............... 112
8.3.4 Linux drivers .................................................... 87 10.5.3.3 Usage .............................................................113
8.3.5 Cortex-M Image ............................................... 87 10.5.4 DM-Crypt using DCP's Crypto Key ................ 117
8.3.5.1 Application name ............................................. 87 10.5.5 DM-Crypt usage on i.MX Platforms without
8.3.5.2 Board setup ..................................................... 87 CAAM hardware IP ........................................118
8.3.5.3 Execution ......................................................... 89 10.6 crypto_af_alg application support .................. 120
8.3.6 Power consumption notes ............................... 89 10.6.1 Prerequisites .................................................. 120
8.3.7 Using Linux remoteproc (i.MX 93 and i.MX 10.6.2 Building the kernel ......................................... 120
95) ....................................................................89 10.6.2.1 Kernel configuration .......................................120
8.4 Conversa Integration ....................................... 90 10.6.2.2 Building a toolchain ....................................... 120
9 Graphics ......................................................... 91 10.6.2.3 Cross compiling the user space sources ....... 120
9.1 imx-gpu-sdk ..................................................... 92 10.6.3 Usage .............................................................121
9.2 G2D-imx-samples ............................................ 92 10.6.4 Use case example ......................................... 121
9.3 viv_samples ..................................................... 92 10.7 Kernel TLS offload .........................................122
9.4 Qt 6 ..................................................................93 10.7.1 Prerequisites .................................................. 122
10 Security ...........................................................93 10.7.2 Running Kernel TLS test ............................... 122
10.1 CAAM kernel driver ......................................... 93 10.8 IMA/EVM on i.MX SoCs ................................ 122
10.1.1 Introduction ...................................................... 93 10.8.1 EVM Key on user keyrings ............................ 123
10.1.2 Source files ......................................................94 10.8.2 Modes of operation in IMA EVM ....................123
10.1.3 Module loading ................................................ 95 10.8.3 Build Steps .................................................... 124
10.1.4 Kernel configuration .........................................95 10.8.4 Steps to verify IMA EVM feature ................... 124
10.1.5 How to test the drivers .................................... 96 10.8.4.1 Testing hashing ..............................................125
10.2 Crypto algorithms support ............................... 98 10.8.4.2 Testing signing ............................................... 125
10.3 CAAM Job Ring backend driver 10.9 Security reference design .............................. 126
specifications ................................................... 99 10.9.1 Automated image signing for secure boot ......126
10.3.1 Verifying driver operation and correctness .....100 10.9.1.1 NXP CST Signer ........................................... 126
10.3.2 Incrementing IRQs in /proc/interrupts ............ 100 10.9.1.2 Prerequisites for preparing a signed image ... 126
10.3.3 Verifying the 'self test' fields say 'passed' 10.9.1.3 Yocto setup for secure boot build .................. 127
in /proc/crypto ................................................ 100 10.9.1.4 Generating a signed bootloader/kernel/WIC
10.4 OpenSSL offload ........................................... 101 image in Yocto project ................................... 127
10.4.1 OpenSSL software architecture ..................... 101 10.9.1.5 Booting a signed image .................................127
10.4.2 OpenSSL's ENGINE interface ....................... 102 11 Connectivity ................................................. 128
10.4.3 NXP solution for OpenSSL hardware 11.1 Connectivity for Bluetooth wireless
offloading ....................................................... 102 technology and Wi-Fi .....................................128
10.4.4 Deploying OpenSSL into rootfs ..................... 103 11.2 Connectivity for USB type-C ..........................130
10.4.5 Running OpenSSL benchmarking tests with 11.3 NXP Bluetooth/Wi-Fi information ................... 131
cryptodev engine ........................................... 103 11.4 Hardware limitations ...................................... 132
10.4.5.1 Running OpenSSL benchmarking tests for 11.5 Certification .................................................... 133
symmetric ciphering and digest ..................... 103 11.5.1 WFA certification ............................................133
10.4.6 Running OpenSSL benchmarking tests with 11.5.2 Bluetooth controller certification .....................133
AF_ALG engine ............................................. 104 12 DDR Performance Monitor ..........................133
10.4.6.1 Running OpenSSL benchmarking tests for 12.1 Introduction .................................................... 133
symmetric ciphering and digest ..................... 104 12.2 Frequently used events ................................. 133
10.4.7 Running OpenSSL asymmetric tests with 12.3 Showing supported events ............................ 134
PKCS#11 based engine ................................ 104 12.4 Examples for monitoring transactions ............134
10.4.7.1 Running p11tool to generate key (RSA or 12.5 Performance metric ....................................... 135
EC) .................................................................105 12.5.1 Showing supported metric ............................. 135
10.4.7.2 Using OpenSSL from command line ............. 106 12.5.2 Monitoring transactions ..................................135
10.4.7.3 Running OpenSSL test for RSA .................... 106 12.6 DDR Performance usage summary ............... 136
10.4.7.4 Running OpenSSL test for EC .......................106

UG10163 All information provided in this document is subject to legal disclaimers. © 2024 NXP B.V. All rights reserved.

User guide Rev. LF6.6.36_2.1.0 — 30 September 2024 Document feedback

155 / 156
NXP Semiconductors
UG10163
i.MX Linux User's Guide

13 One-Time Programmable Controller

Driver Using NVMEM Subsystem ............... 136
13.1 Introduction .................................................... 136
13.2 NVMEM provider OCOTP ..............................136
13.3 NVMEM consumer .........................................136
13.4 Examples to read/write the raw NVMEM file
in user space ................................................. 136
14 NXP eIQ Machine Learning ......................... 137
15 Murata Wi-Fi/Bluetooth Solutions .............. 137
16 NXP i.MX SystemReady IR ..........................139
16.1 Preparing binaries ..........................................139
16.1.1 Yocto .............................................................. 139
16.1.1.1 Setting up the Yocto build environment ..........139
16.1.1.2 Building Host packages for Yocto build ..........140
16.1.1.3 Setting up the NXP i.MX Yocto Linux BSP .....140
16.1.1.4 Building the STMM bootloader ...................... 140
16.1.2 Manual build .................................................. 141
16.1.2.1 Building EDK2 ............................................... 141
16.1.2.2 Building OP-TEE OS ..................................... 141
16.1.2.3 Building U-Boot ..............................................143
16.1.2.4 Building ATF .................................................. 143
16.1.2.5 Generating flash.bin and capsule1.bin ...........143
16.2 Running capsule update ................................145
16.3 Running the ACS Test ................................... 148
16.4 Distribution installation ...................................149
16.4.1 Fedora ............................................................150
16.4.2 OpenSuse ...................................................... 150
16.5 Result check .................................................. 150
17 Note About the Source Code in the
Document ..................................................... 150
18 Revision History .......................................... 151
Legal information .........................................153

Please be aware that important notices concerning this document and the product(s)
described herein, have been included in section 'Legal information'.

© 2024 NXP B.V. All rights reserved.

For more information, please visit: https://www.nxp.com Document feedback
Date of release: 30 September 2024
Document identifier: UG10163