The DENX U-Boot and Linux Guide (DULG) for m28

1. Abstract

This is the DENX U-Boot and Linux Guide to Embedded PowerPC, ARM and MIPS Systems.

The document describes how to configure, build and use the firmware Das U-Boot (typically abbreviated as just "U-Boot") and the operating system Linux for Embedded PowerPC, ARM and MIPS Systems.

The focus of this version of the document is on m28 boards.

This document was generated at 10 Dec 2012 - 06:19.

2. Introduction

This document describes how to use the firmware U-Boot and the operating system Linux in Embedded Power Architecture®, ARM and MIPS Systems.

There are many steps along the way, and it is nearly impossible to cover them all in depth, but we will try to provide all necessary information to get an embedded system running from scratch. This includes all the tools you will probably need to configure, build and run U-Boot and Linux.

First, we describe how to install the Cross Development Tools Embedded Linux Development Kit which you probably need - at least when you use a standard x86 PC running Linux or a Sun Solaris 2.6 system as build environment.

Then we describe what needs to be done to connect to the serial console port of your target: you will have to configure a terminal emulation program like cu or kermit.

In most cases you will want to load images into your target using ethernet; for this purpose you need TFTP and DHCP / BOOTP servers. A short description of their configuration is given.

A description follows of what needs to be done to configure and build the U-Boot for a specific board, and how to install it and get it working on that board.

The configuration, building and installing of Linux in an embedded configuration is the next step. We use SELF, our Simple Embedded Linux Framework, to demonstrate how to set up both a development system (with the root filesystem mounted over NFS) and an embedded target configuration (running from a ramdisk image based on busybox).

This document does not describe what needs to be done to port U-Boot or Linux to a new hardware platform. Instead, it is silently assumed that your board is already supported by U-Boot and Linux.

The focus of this document is on m28 boards.

2.1. Copyright

You have the freedom to distribute copies of this document in any format or to create a derivative work of it and distribute it provided that you:

• Distribute this document or the derivative work at no charge at all. It is not permitted to sell this document or the derivative work or to include it into any package or distribution that is not freely available to everybody.

• Send your derivative work (in the most suitable format such as sgml) to the author.

• License the derivative work with this same license or use GPL. Include a copyright notice and at least a pointer to the license used.

• Give due credit to previous authors and major contributors.

It is requested that corrections and/or comments be forwarded to the author.

If you are considering to create a derived work other than a translation, it is requested that you discuss your plans with the author.

2.2. Disclaimer

Use the information in this document at your own risk. DENX disavows any potential liability for the contents of this document. Use of the concepts, examples, and/or other content of this document is entirely at your own risk. All copyrights are owned by their owners, unless specifically noted otherwise. Use of a term in this document should not be regarded as affecting the validity of any trademark or service mark. Naming of particular products or brands should not be seen as endorsements.

2.3. Availability

The latest version of this document is available in a number of formats:

• HTML http://www.denx.de/wiki/publish/DULG/DULG-m28.html
• plain ASCII text http://www.denx.de/wiki/publish/DULG/DULG-m28.txt
• PostScript European A4 format http://www.denx.de/wiki/publish/DULG/DULG-m28.ps
• PDF European A4 format http://www.denx.de/wiki/publish/DULG/DULG-m28.pdf

2.4. Credits

2.5. Translations

2.6. Feedback

2.7. Conventions

Descriptions	Appearance
Warnings
Hint
Notes	Note.
Information requiring special attention	Warning
File Names	file.extension
Directory Names	directory
Commands to be typed	a command
Applications Names	another application
Prompt of users command under bash shell	bash$
Prompt of root users command under bash shell	bash#
Prompt of users command under tcsh shell	tcsh$
Environment Variables	VARIABLE
Emphasized word	word
Code Example	ls -l

3. Embedded Linux Development Kit

The Embedded Linux Development Kit (ELDK) includes the GNU cross development tools, such as the compilers, binutils, gdb, etc., and a number of pre-built target tools and libraries necessary to provide some functionality on the target system.

It is provided for free with full source code, including all patches, extensions, programs and scripts used to build the tools.

Some versions of ELDK (4.1) are available in two versions, which use Glibc resp. uClibc as the main C library for the target packages.

Packaging and installation is based on the RPM package manager.

3.1. ELDK Availability

The ELDK is available

• on DVD-ROM from DENX Computer Systems
• for download on the following server:
  
  

  FTP	HTTP
  ftp://ftp.denx.de/pub/eldk/	http://www.denx.de/ftp/pub/eldk/

• for download on the following mirrors:
  
  

  FTP	HTTP
  ftp://ftp-stud.hs-esslingen.de/pub/Mirrors/eldk/	http://ftp-stud.hs-esslingen.de/pub/Mirrors/eldk/
  ftp://mirror.switch.ch/mirror/eldk/	http://mirror.switch.ch/ftp/mirror/eldk/
  not available	http://mira.sunsite.utk.edu/eldk/
  ftp://ftp.sunet.se/pub/Linux/distributions/eldk/	http://ftp.sunet.se/pub/Linux/distributions/eldk/

3.2. ELDK Getting Help

Community support for the ELDK is available through the ELDK Mailing List.
Previous postings to this mailing list are available from the ELDK archives.

Commercial support is also available; please feel free to contact DENX Software Engineering GmbH.

3.3. Supported Host Systems

The ELDK can be installed onto and operate with the following operating systems:

• Fedora Core 4, 5, 6 Fedora 7, 8, 9, 10, 11, 12
• Red Hat Linux 7.3, 8.0, 9
• SuSE Linux 8.x, 9.0, 9.1, 9.2, 9.3, 10.0
• OpenSUSE 10.2, 10.3 (32 Bit); OpenSUSE 11.0 (32 and 64 Bit)
• Debian 3.0 (Woody), 3.1 (Sarge) and 4.0 (Etch)
• Ubuntu 4.10, 5.04, 6.10, 8.04, 9.04, 9.10, 10.04
• FreeBSD 5.0

Users also reported successful installation and use of the ELDK on the following host systems:

• Suse Linux 7.2, 7.3
• Mandrake 8.2
• Slackware 8.1beta2
• Gentoo Linux 2006.1

Note: It may be necessary, and is usually recommended, to install the latest available software updates on your host system. For example, on Fedora systems, you can use yum or apt-get to keep your systems current.

3.4. Supported Target Architectures

The ELDK for ARM systems supports processors complying with the ARM architecture version 2 to 6. This includes ARM7, ARM9, XScale, AT91RM9200, i.MX31, S3C6400 and other ARM based systems.

The version of 4.2 and higher of ELDK has two ARM targets in distribution - one with the soft-float math support, and another one with the Vector Floating Point math support. Both targets comply with ARM Embedded Application Binary Interface (EABI).

The ELDK ARM target architectures are:

• arm-linux = soft-float math
• armVFP-linux = Vector Floating Point (VFP) math

There is also an ELDK for PowerPC and MIPS systems.

3.5. Installation

3.5.1. Product Packaging

Stable versions of the ELDK are distributed in the form of an ISO image, which can be either burned onto a DVD or mounted directly, using the loopback Linux device driver (Linux host only).

Development versions of the ELDK are available as directory trees so it is easy to update individual packages; instructions for download of these trees and creation of ISO images from it is described in section 3.5.2. Downloading the ELDK.

The ELDK contains an installation utility and a number of RPM packages, which are installed onto the hard disk of the cross development host by the installation procedure. The RPM packages can be logically divided into two parts:

• Embedded Linux Development Tools (ELDT)
• Target components

The first part contains the cross development tools that are executed on the host system. Most notably, these are the GNU cross compiler, binutils, and gdb. For a full list of the provided ELDT packages, refer to section 3.9.1. List of ELDT Packages below.

The target components are pre-built tools and libraries which are executed on the target system. The ELDK includes necessary target components to provide a minimal working NFS-based environment for the target system. For a list of the target packages included in the ELDK, refer to section 3.9.2. List of Target Packages below.

The ELDK contains several independent sets of the target packages, one for each supported target architecture CPU family. Each set has been built using compiler code generation and optimization options specific to the respective target CPU family.

3.5.2. Downloading the ELDK

You can either download the ready-to-burn ISO-images from one of the mirror sites (see 3.1. ELDK Availability), or you can download the individual files of the ELDK from the development directory tree and either use these directly for installation or create an ISO image that can be burned on DVD-ROM.

Change to a directory with sufficient free disk space; for the ARM version of the ELDK you need about 510 MiB, or twice as much (1.1 GiB) if you also want to create an ISO image in this directory.

To download the ISO image from the arm-linux-x86/iso directory of one of the mirror sites you can use standard tools like wget or ncftpget, for example:

bash$ wget ftp://ftp.sunet.se/pub/Linux/distributions/eldk/4.2/arm-linux-x86/iso/arm-2008-11-24.iso

If you want to download the whole ELDK directory tree instead you can - for example - use the ncftp FTP client:

bash$ ncftp ftp.sunet.se
...
ncftp / > cd /pub/Linux/distributions/eldk/4.2
ncftp /pub/Linux/distributions/eldk/4.2 > bin
ncftp /pub/Linux/distributions/eldk/4.2 > get -R arm-linux-x86/distribution
...
ncftp /pub/Linux/distributions/eldk/4.2 > bye

If you don't find the ncftp tool on your system you can download the NcFTP client from http://www.ncftp.com/download/

There are a few executable files (binaries and scripts) in the ELDK tree. Make sure they have the execute permissions set in your local copy:

bash$ for file in \
>       tools/bin/rpm \
>       tools/usr/lib/rpm/rpmd \
>       install \
>       ELDK_MAKEDEV \
>       ELDK_FIXOWNER
> do
> chmod +x arm-linux-x86/distribution/$file
> done

Now create an ISO image from the directory tree:

bash$ mkisofs \
> -A "ELDK-4.2 -- Target: ARM -- Host: x86 Linux" \
> -publisher "(C) `date "+%Y"` DENX Software Engineering,   www.denx.de" \
> -p "`id -nu`@`hostname` -- `date`" \
> -V arm-linux-x86 \
> -l -J -R -o eldk-arm-linux-x86.iso arm-linux-x86/distribution

This will create an ISO image eldk-arm-linux-x86.iso in your local directory that can be burned on DVD or mounted using the loopback device and used for installation as described above. Of course you can use the local copy of the directory tree directly for the installation, too.

Please refer to section 3.10.2. Setting Up ELDK Build Environment for instructions on obtaining the build environment needed to re-build the ELDK from scratch.

3.5.3. Initial Installation

The initial installation is performed using the install utility located in the root of the ELDK ISO image directory tree. The install utility has the following syntax:

$ ./install [-d <dir>] [<cpu_family1>] [<cpu_family2>] ...

Note: Make sure that the "exec" option to the mount command is in effect when mounting the ELDK ISO image. Otherwise the install program cannot be executed. On some distributions, it may be necessary to modify the /etc/fstab file, adding the "exec" mount option to the cdrom entry - it may also be the case that other existing mount options, such as "user" prevent a particular configuration from mounting the ELDK DVD with appropriate "exec" permission. In such cases, consult your distribution documentation or mount the DVD explicitly using a command such as "sudo mount -o exec /dev/cdrom /mnt/cdrom" (sudo allows regular users to run certain privileged commands but may not be configured - run the previous command as root without "sudo" in the case that "sudo" has not been setup for use on your particular GNU/Linux system).

You can install the ELDK to any empty directory you wish, the only requirement being that you have to have write and execute permissions on the directory. The installation process does not require superuser privileges.

Depending on the parameters the install utility is invoked with, it installs one or more sets of target components. The ELDT packages are installed in any case.

Refer to section 3.6. Working with ELDK for a sample usage of the ELDK.

Note: If you intend to use the installation as a root filesystem exported over NFS, then you now have to finish the configuration of the ELDK following the instructions in 3.7. Mounting Target Components via NFS.

Note: Installation of the Glibc- and uClibc-based ELDK versions into one directory is not yet supported.

Note: Installation of the 32-bit and 64-bit ELDK versions into one directory is not yet supported.

3.5.4. Installation and Removal of Individual Packages

The ELDK has an RPM-based structure. This means that on the ISO image, individual components of the ELDK are in the form of RPM packages, and after installation, the ELDK maintains its own database which contains information about installed packages. The RPM database is kept local to the specific ELDK installation, which allows you to have multiple independent ELDK installations on your host system. (That is, you can install several instances of ELDK under different directories and work with them independently). Also, this provides for easy installation and management of individual ELDK packages.

To list the installed ELDK RPM packages, use the following command:

bash$ ${CROSS_COMPILE}rpm -qa

To remove an ELDK package, use the following command:

bash$ ${CROSS_COMPILE}rpm -e <package_name>

To install a package, use the following command:

bash$ ${CROSS_COMPILE}rpm -i <package_file_name>

To update a package, use the following command:

bash$ ${CROSS_COMPILE}rpm -U <package_file_name>

For the above commands to work correctly, it is crucial that the correct rpm binary gets invoked. In case of multiple ELDK installations and RedHat-based host system, there may well be several rpm tools installed on the host system.

You must make sure, either by using an explicit path or by having set an appropriate PATH environment variable, that when you invoke rpm to install/remove components of a ELDK installation, it is the ELDK's rpm utility that gets actually invoked. The rpm utility is located in the bin subdirectory relative to the ELDK root installation directory.

To avoid confusion with the host OS (RedHat) rpm utility, the ELDK creates symlinks to its rpm binary with the names such that it could be invoked using the ${CROSS_COMPILE}rpm notation, for all supported $CROSS_COMPILE values.

The standard (host OS) rpm utility allows various macros and configuration parameters to specified in user-specific ~/.rpmrc and ~/.rpmmacros files. The ELDK rpm tool also has this capability, but the names of the user-specific configuration files are ~/.eldk_rpmrc and ~/.eldk_rpmmacros, respectively.

3.5.5. Removal of the Entire Installation

To remove the entire ELDK installation, use the following command while in the ELDK root directory:

bash$ rm -rf <dir>

where <dir> specifies the root directory of the ELDK to be removed.

3.6. Working with ELDK

After the initial installation is complete, all you have to do to start working with the ELDK is to set and export the CROSS_COMPILE environment variable. Optionally, you may wish to add the bin and usr/bin directories of your ELDK installation to the value of your PATH environment variable. For instance, a sample ELDK installation and usage scenario looks as follows:

• Create a new directory where the ELDK is to be installed, say:

  bash$ mkdir /opt/eldk
  

• Mount a CD or an ISO image with the distribution:

  bash$ mount /dev/cdrom /mnt/cdrom
  

• Run the installation utility included on the distribution to install into that specified directory:

  bash$ /mnt/cdrom/install -d /opt/eldk
  

• After the installation utility completes, export the CROSS_COMPILE variable:

     bash$ export CROSS_COMPILE=arm-linux-gnueabi-
     

• The trailing '-' character in the CROSS_COMPILE variable value is optional and has no effect on the cross tools behavior. However, it is required when building Linux kernel and U-Boot images.

• Add the directories /opt/eldk/usr/bin and /opt/eldk/bin to PATH:

  bash$ PATH=$PATH:/opt/eldk/usr/bin:/opt/eldk/bin
  

• Compile a file:

  bash$ ${CROSS_COMPILE}gcc -o hello_world hello_world.c
  

  You can also call the cross tools using the generic prefix arm-linux- for example:

  bash$ arm-linux-gcc -o hello_world hello_world.c
  

• or, equivalently:

  bash /opt/eldk/usr/arm-linux/bin/gcc -o hello_world hello_world.c                                                                               
  

The value of the CROSS_COMPILE variable must correspond to the target CPU family you want the cross tools to work for. Refer to the table below for the supported CROSS_COMPILE variable values:

3.6.A Table of possible values for $CROSS_COMPILE

3.6.1. Switching Between Multiple Installations

No special actions are required from the user to switch between multiple ELDK installations on the same host system. Which ELDK installation is used is determined entirely by the filesystem location of the binary that is being invoked. This approach can be illustrated using the following example.

Assume the directory /work/denx_tools/usr/bin, where the arm-linux-gcc compiler binary has been installed, is a part of the PATH environment variable. The user types the command as follows:

$ arm-linux-gcc -c myfile.c

To load the correct include files, find the correct libraries, spec files, etc., the compiler needs to know the ELDK root directory. The compiler determines this information by analyzing the shell command it was invoked with ( arm-linux-gcc - without specifying the explicit path in this example) and, if needed, the value of the PATH environment variable. Thus, the compiler knows that it has been executed from the /work/denx_tools/usr/bin directory.

Then, it knows that the compiler is installed in the usr/bin subdirectory of the root installation directory, so the ELDK, the compiler is a part of, has been installed in the subdirectories of the /work/denx_tools directory. This means that the target include files are in /work/denx_tools/<target_cpu_variant>/usr/include, and so on.

3.7. Mounting Target Components via NFS

The target components of the ELDK can be mounted via NFS as the root file system for your target machine. For instance, for an AT91-based target, and assuming the ELDK has been installed into the /opt/eldk directory, you can use the following directory as the NFS-based root file system:

/opt/eldk/arm

Before the NFS-mounted root file system can work, you must create necessary device nodes in the <ELDK_root>/<target_cpu_variant>/dev directory. This process requires superuser privileges and thus cannot be done by the installation procedure (which typically runs as non-root). To facilitate creation of the device nodes, the ELDK provides a script named ELDK_MAKEDEV, which is located in the root of the ELDK distribution ISO image. The script acccepts the following optional arguments:

# /mnt/cdrom/ELDK_MAKEDEV -d /opt/eldk

Some of the target utilities included in the ELDK, such as mount and su, have the SUID bit set. This means that when run, they will have privileges of the file owner of these utilities. That is, normally, they will have the privileges of the user who installed the ELDK on the host system. However, for these utilities to work properly, they must have superuser privileges. This means that if the ELDK was not installed by the superuser, the file owner of the target ELDK utilities that have the SUID bit set must be changed to root before a target component may be mounted as the root file system. The ELDK distribution image contains an ELDK_FIXOWNER script, which you can use to change file owners of all the appropriate files of the ELDK installation to root. The script accepts the same arguments as the ELDK_MAKEDEV script above. Please note that you must have superuser privileges to run this script. For instance, if you have installed the ELDK in the /opt/eldk directory, you can use the following commands:

# cd /opt/eldk
# /mnt/cdrom/ELDK_FIXOWNER

Please note, that in the case that the installation directory, where the new ELDK distribution is being installed, is already populated with other ELDK distributions, the execution of the ELDK_FIXOWNER script without arguments will make the script work with all installed ELDK target architecture directories. This could take some time. To save the time, please use the -a argument to specify the appropriate target architecture. For instance:

# cd /opt/eldk
# /mnt/cdrom/ELDK_FIXOWNER -a arm

3.8. Rebuilding ELDK Components

3.8.1. ELDK Source Distribution

The ELDK is distributed with the full sources of all the components, so you may rebuild any ELDK package. The sources are provided in the form of SRPM packages, distributed as a separate ISO image.

To rebuild a target or ELDT package, you must first install the appropriate source RPM package from the ISO image into the ELDK environment. This can be done using the following command:

$ ${CROSS_COMPILE}rpm -i /mnt/cdrom/SRPMS/<source_rpm_file_name>.src.rpm

After an ELDK source RPM is installed using the above command, its spec file and sources can be found in the subdirectories of the <ELDK_root>/usr/src/denx subdirectory.

The sections that follow provide detailed instructions on rebuilding ELDT and target components of the ELDK.

3.8.2. Rebuilding Target Packages

All the target packages can be rebuilt from the provided source RPM packages. At first you have to install the Source RPM itself:

bash$ ${CROSS_COMPILE}rpm -iv <package_name>.src.rpm

Then you can rebuild the binary target RPM using the following command from the ELDK environment:

bash$ ${CROSS_COMPILE}rpmbuild -ba <package_name>.spec

In order for the rebuilding process to work correctly, the following conditions must be true:

• The $CROSS_COMPILE environment variable must be set as appropriate for the target CPU family.

• The <ELDK_root>/usr/arm-linux/bin directory must be in PATH before the /usr/bin directory. This is to make sure that the command gcc results in the fact that the ELDK cross compiler is invoked, rather than the host gcc.

The newly built package can then be installed just as easily:

bash$ ${CROSS_COMPILE}rpm -i <package_name>.rpm

3.8.3. Rebuilding ELDT Packages

All the ELDT packages allow for rebuilding from the provided source RPM packages using the following command from the ELDK environment:

$ unset CROSS_COMPILE
$ <ELDK_root>/usr/bin/rpmbuild -ba <package_name.spec>

In order for the rebuilding process to work correctly, make sure all of the following is true:

• The $CROSS_COMPILE environment variable must NOT be set.

• Do NOT use the $CROSS_COMPILE command prefix.

• The <ELDK_root>/usr/arm-linux/bin directory must NOT be in PATH. This is to make sure that the command gcc causes invokation of the host gcc, rather than the ELDK cross compiler.

Note that the newly built package should be installed with the "global" rpm, not with the arch specific one:

bash$ <ELDK_root>/bin/rpm -i <package_name>.rpm

3.9. ELDK Packages

3.9.1. List of ELDT Packages

Note: The crosstool 0.43 ELDT package provides the following packages: gcc 4.2.2, gcc-c++ 4.2.2, gcc-java 4.2.2, cpp 4.2.2 and binutils 2.17.90. For more information about the crosstool package please refer to http://kegel.com/crosstool.

3.9.2. List of Target Packages

Note 1: Not all packages will be installed automatically; for example the boa and thttpd web servers are mutually exclusive - you will have to remove one package before you can (manually) install the other one.

Note 2: The crosstool 0.43 target package provides the following packages: glibc 2.6, glibc-common 2.6, glibc-devel 2.6, libstdc++ 4.2.2, libgcj 4.2.2, libgcj-devel 4.2.2 and libstdc++-devel 4.2.2. For more information about the crosstool package please refer to http://kegel.com/crosstool

Note 3: The Xenomai and gcc-java packages are unavailable in ARM ELDK version.

3.10. Rebuilding the ELDK from Scratch

In this section, you will find instructions on how to build the ELDK from scratch, using the pristine package sources available on the Internet, and patches, spec files, and build scripts provided on the ELDK source CD-ROM.

3.10.1. ELDK Build Process Overview

The ELDK uses the Fedora 7 Linux distribution as source code reference. Any modifications to Fedora's sources the ELDK has introduced are in the form of patches applied by the RPM tool while building the packages. Also, the ELDK uses modified spec files for its RPM packages. So, the sources of almost every ELDK package consist of the following parts:

• Fedora pristine sources or
• ELDK source tarball,
• ELDK patches,
• ELDK spec file.

The Fedora pristine sources may be obtained from the Internet, see http://download.fedora.redhat.com/pub/fedora/linux/core.

The ELDK patches and spec files are available on the ELDK source CD-ROM and from the DENX GIT repositories. Also, for convenience, the pristine Fedora sources are available here, too.

Please use the following commands to check out a copy of one of the modules:

git-clone git://www.denx.de/git/eldk/module

Then you may switch to a specific release of the ELDK using the "git-checkout" command; for example, to get the files for ELDK release 4.1, please do the following from the module directory:

git-checkout ELDK_4_2

It must be noted that some of the packages which are included in the ELDK are not included in Fedora. Examples of such packages are appWeb, microwindows, and wu-ftpd. For these packages tarballs are provided in the DENX GIT repository.

To facilitate building of the ELDK, a build infrastructure has been developed. The infrastructure is composed of the following components:

• ELDK_BUILD script
• build.sh script
• cpkgs.lst file
• tpkgs.lst file
• SRPMS.lst file
• tarballs.lst file

The ELDK_BUILD script is the main script of the ELDK build procedure. It is the tool that you would normally use to build the ELDK from scratch. In the simplest case, the script may be invoked without arguments, and it will perform all necessary steps to build the ELDK in a fully automated way. You may pass the following optional arguments to the ELDK_BUILD script:

Warning: The ELDK build scripts rely on standard behaviour of the RPM tool. Make sure you don't use non-standard settings in your personal ~/.rpmmacros file that might cause conflicts.

build.sh is a supplementary script that is called by ELDK_BUILD to accomplish certain steps of the build. Refer to section 3.10.3. build.sh Usage below for more details.

The cpkgs.lst and tpkgs.lst files are read by build.sh and must contain lines describing sub-steps of the eldt and trg build procedure steps. Essentially, the files contain the list of the ELDT and target packages to be included in the ELDK. The SRPMS.lst file contains the list of the Fedora source RPM packages used during the ELDK build. The tarballs.lst file contains the list of source tarballs of the packages that are included in the ELDK but are not present in Fedora 7.

For the ELDK_BUILD script to work correctly, it must be invoked from a certain build environment created on the host system. The build environment can be either checked out from the DENX GIT repository (see section 3.10.2. Setting Up ELDK Build Environment below for details) or copied from the ELDK build environment CD-ROM.

To be more specific, the following diagram outlines the build environment needed for correct operation of the ELDK_BUILD script:

<some_directory>/
                 build/cross_rpms/<package_name>/SPECS/...
                                                 SOURCES/...
                       target_rpms/<package_name>/SPECS/...
                                                  SOURCES/...
                       install/install.c
                               Makefile
                       misc/ELDK_MAKEDEV
                            ELDK_FIXOWNER
                            README.html
                       cpkgs.lst
                       tpkgs.lst
                       build.sh
                                                                                
                       ELDK_BUILD
                                                                                
                       SRPMS.lst
                                                                                
                       tarballs.lst

                 tarballs/....
                                                                                
                 SRPMS/....
                 SRPMS-updates/....

In subdirectories of the cross_rpms and target_rpms directories, the sources and RPM spec files of, respectively, the ELDT and target packages are stored. The install subdirectory contains the sources of the installation utility which will be built and placed in the root of the ISO image. tarballs directory contains the source tarballs of the packages that are included in the ELDK but are not present in Fedora 7.

The SRPMS and SRPMS-updates directories may contain the source RPM packages of Fedora 7. The ELDK_BUILD script looks for a package in the SRPMS directory and then, if the package is not found, in the SRPMS-updates directory. If some (or all) of the Fedora SRPMs needed for the build are missing in the directories, the ELDK_BUILD script will download the source RPMs automatically from the Internet.

The ELDK build environment CD-ROM provides a ready-to-use ELDK build environment. Please refer to section 3.10.2. Setting Up ELDK Build Environment below for detailed instructions on setting up the build environment.

The ELDK_BUILD script examines the contents of the ELDK_PREFIX environment variable to determine the root directory of the ELDK build environment. If the variable is not set when the script is invoked, it is assumed that the root directory of the ELDK build environment is /opt/eldk. To build the ELDK in the example directory layout given above, you must set and export the ELDK_PREFIX variable <some_directory> prior to invoking ELDK_BUILD.

After all the build steps are complete, the following subdirectories are created in the ELDK build environment:

build/<build_name>/work/             - full ELDK environment
build/<build_name>/logs/             - build procedure log files
build/<build_name>/results/b_cdrom/  - binary cdrom tree, ready for mkisofs
                   results/s_cdrom/  - source cdrom tree, ready for mkisofs
                   results/d_cdrom/  - debuginfo cdrom tree, ready for mkisofs

On Linux hosts, the binary and source ISO images are created automatically by the ELDK_BUILD script and placed in the results directory. On Solaris hosts, creating the ISO images is a manual step. Use the contents of the b_cdrom and s_cdrom directories for the contents of the ISO images.

3.10.2. Setting Up ELDK Build Environment

For your convenience, the ELDK build environment CD-ROM provides full ELDK build environment. All you need to do is copy the contents of the CD-ROM to an empty directory on your host system. Assuming the ELDK build environment CD-ROM is mounted at /mnt/cdrom, and the empty directory where you want to create the build environment is named /opt/eldk, use the following commands to create the build environment:

bash$ cd /opt/eldk
bash$ cp -r /mnt/cdrom/* .

These commands will create the directory structure as described in section 3.10.1. ELDK Build Process Overview above. All necessary scripts and ELDK specific source files will be placed in the build subdirectory, and the required tarballs can be found in the tarballs subdirectory. In the SRPMS subdirectory, you will find all the Fedora 7 SRPMS needed to build the ELDK.

Alternatively, you can obtain the ELDK build environment from the DENX GIT repository. Two modules are provided for check out: build and tarballs. The first one contains the files for the build subdirectory in the build environment, and the second one contains source tarballs of the packages that are included in the ELDK but are not present in Fedora 7. To create the ELDK build environment from the DENX GIT repository, use the following commands (the example below assumes that the root directory of the build environment is /opt/eldk):

bash$ cd /opt/eldk
bash$ git-clone git://www.denx.de/git/eldk/build
bash$ git-clone git://www.denx.de/git/eldk/tarballs
bash$ git-clone git://www.denx.de/git/eldk/SRPMS

Note: To allow to install the ELDK on as many as possible Linux distributions (including old systems), we use a Red Hat 7.3 host system for building. Also, Fedora Core 5 is known to work as a build environment. Other, especially more recent Linux distributions, will most likely have problems. We therefor provide a Red Hat 7.3 based root file system image than can run in some virtualization environment (like qemu etc.). Here is an application note with detailed instructions: http://www.denx.de/wiki/DULG/AN2009_02_EldkReleaseBuildEnvironment

3.10.3. build.sh Usage

The whole build procedure is logically divided into six steps, and the build.sh must be told which of the build steps to perform. The build steps are defined as follows:

• rpm - build RPM
• eldt - build ELDT packages
• seldt - save ELDT SRPM packages to create a source ISO image later on
• trg - build target packages
• biso - prepare the file tree to create the binary ISO image
• siso - prepare the file tree to create the source ISO image
• diso - prepare the file tree to create the debuginfo ISO image

Further, the eldt and trg build steps are devided into sub-steps, as defined in the cpkgs.lst and tpkgs.lst

files (see below for details). You may specify which sub-steps of the build step are to be performed.

The formal syntax for the usage of build.sh is as follows:

bash$ ./build.sh [-a <arch>] [-n <name>] [-p <prefix>] [-r <result>] \
                 [-w <work>] <step_name> [<sub_step_number>]

Please note that you must never use build.sh to build the ELDK from scratch. For build.sh to work correctly, the script must be invoked from the build environment after a successful build using the ELDK_BUILD script. A possible scenario of build.sh usage is such that you have a build environment with results of a build performed using the ELDK_BUILD script and want to re-build certain ELDT and target packages, for instance, because you have updated sources of a package or added a new package to the build.

When building the target packages (during the trg buildstep), build.sh examines the contents of the TARGET_CPU_FAMILY_LIST environment variable, which may contain a list indicating which target CPU variants the packages must be built for. Possible CPU variants are arm. For example, the command below rebuilds the target RPM listed in the tpckgs.lst file under the number of 47 (see section 3.10.4. Format of the cpkgs.lst and tpkgs.lst Files for description of the tpckgs.lst and cpkgs.lst files), for the arm CPU:

bash$ TARGET_CPU_FAMILY_LIST="arm" \
> /opt/eldk/build.sh -a arm \
>                    -n 2007-01-21 \
>                    -p /opt/eldk/build/arm-2007-01-21 \
>                    -r /opt/eldk/build/arm-2007-01-21/results \
>                    -w /opt/eldk/build/arm-2007-01-21/work \
>                    trg 47 47

Note: If you are going to invoke build.sh to re-build a package that has already been built in the build environment by the ELDK_BUILD script, then you must first manually uninstall the package from ELDK installation created by the build procedure under the work directory of the build environment.

Note: It is recommended that you use the build.sh script only at the final stage of adding/updating a package to the ELDK. For debugging purposes, it is much more convenient and efficient to build both ELDT and target packages using a working ELDK installation, as described in the sections 3.8.2. Rebuilding Target Packages and 3.8.3. Rebuilding ELDT Packages above.

3.10.4. Format of the cpkgs.lst and tpkgs.lst Files

Each line of these files has the following format:

<sub_step_number> <package_name> <spec_file_name> \
  <binary_package_name> <package_version>

The ELDK source CD-ROM contains the cpkgs.lst and tpkgs.lst files used to build this version of the ELDK distribution. Use them as reference if you want to include any additional packages into the ELDK, or remove unneeded packages.

To add a package to the ELDK you must add a line to either the cpkgs.lst file, if you are adding a ELDT package, or to the tpkgs.lst file, if it is a target package. Keep in mind that the relative positions of packages in the cpkgs.lst and tpkgs.lst files (the sub-step numbers) are very important. The build procedure builds the packages sequentially as defined in the *.lst files and installs the packages in the "work" environment as they are built. This implies that if a package depends on other packages, those packages must be specified earlier (with smaller sub-step numbers) in the *.lst files.

Note: For cpkgs.lst, the package_version may be replaced by the special keyword "RHAUX". Such packages are used as auxiliary when building ELDK 4.2 on non-Fedora hosts. These packages will be built and used during the build process, but will not be put into the ELDK 4.2 distribution ISO images.

4. System Setup

Some tools are needed to install and configure U-Boot and Linux on the target system. Also, especially during development, you will want to be able to interact with the target system. This section describes how to configure your host system for this purpose.

4.1. Serial Console Access

To use U-Boot and Linux as a development system and to make full use of all their capabilities you will need access to a serial console port on your target system. Later, U-Boot and Linux can be configured to allow for automatic execution without any user interaction.

There are several ways to access the serial console port on your target system, such as using a terminal server, but the most common way is to attach it to a serial port on your host. Additionally, you will need a terminal emulation program on your host system, such as cu or kermit.

4.2. Configuring the "cu" command

The cu command is part of the UUCP package and can be used to act as a dial-in terminal. It can also do simple file transfers, which can be used in U-Boot for image download.

On RedHat systems you can check if the UUCP package is installed as follows:

$ rpm -q uucp

If necessary, install the UUCP package from your distribution media.

To configure cu for use with U-Boot and Linux please make sure that the following entries are present in the UUCP configuration files; depending on your target configuration the serial port and/or the console baudrate may be different from the values used in this example: (/dev/ttyS0, 115200 bps, 8N1):

• /etc/uucp/sys:

#
# /dev/ttyS0 at 115200 bps:
#
system          S0@115200
port            serial0_115200
time            any
     

• /etc/uucp/port:

#
# /dev/ttyS0 at 115200 bps:
#
port            serial0_115200
type            direct
device          /dev/ttyS0
speed           115200
hardflow        false
     

You can then connect to the serial line using the command

$ cu S0@115200
Connected.

To disconnect, type the escape character '~' followed by '.' at the beginning of a line.

See also: cu(1), info uucp.

4.3. Configuring the "kermit" command

The name kermit stands for a whole family of communications software for serial and network connections. The fact that it is available for most computers and operating systems makes it especially well suited for our purposes.

kermit executes the commands in its initialization file, .kermrc, in your home directory before it executes any other commands, so this can be easily used to customize its behaviour using appropriate initialization commands. The following settings are recommended for use with U-Boot and Linux:

• ~/.kermrc:

set line /dev/ttyS0
set speed 115200
set carrier-watch off
set handshake none
set flow-control none
robust
set file type bin
set file name lit
set rec pack 1000
set send pack 1000
set window 5

This example assumes that you use the first serial port of your host system (/dev/ttyS0) at a baudrate of 115200 to connect to the target's serial console port.

You can then connect to the serial line:

$ kermit -c
Connecting to /dev/ttyS0, speed 115200.
The escape character is Ctrl-\ (ASCII 28, FS)
Type the escape character followed by C to get back,
or followed by ? to see other options.
----------------------------------------------------

Due to licensing conditions you will often find two kermit packages in your GNU/Linux distribution. In this case you will want to install the ckermit package. The gkermit package is only a command line tool implementing the kermit transfer protocol.

If you cannot find kermit on the distribution media for your Linux host system, you can download it from the kermit project home page: http://www.columbia.edu/kermit/

4.4. Using the "minicom" program

minicom is another popular serial communication program. Unfortunately, many users have reported problems using it with U-Boot and Linux, especially when trying to use it for serial image download. It's use is therefore discouraged.

4.5. Permission Denied Problems

The terminal emulation program must have write access to the serial port and to any locking files that are used to prevent concurrent access from other applications. Depending on the used Linux distribution you may have to make sure that:

• the serial device belongs to the same group as the cu command, and that the permissions of cu have the setgid bit set

• the kermit belongs to the same group as cu and has the setgid bit set

• the /var/lock directory belongs to the same group as the cu command, and that the write permissions for the group are set

4.6. Configuration of a TFTP Server

The fastest way to use U-Boot to load a Linux kernel or an application image is file transfer over Ethernet. For this purpose, U-Boot implements the TFTP protocol (see the tftpboot command in U-Boot).

To enable TFTP support on your host system you must make sure that the TFTP daemon program /usr/sbin/in.tftpd is installed. On RedHat systems you can verify this by running:

$ rpm -q tftp-server

If necessary, install the TFTP daemon program from your distribution media.

Most Linux distributions disable the TFTP service by default. To enable it for example on RedHat systems, edit the file /etc/xinetd.d/tftp and remove the line

disable = yes

# default: off
# description: The tftp server serves files using the trivial file transfer
#       protocol.  The tftp protocol is often used to boot diskless
#       workstations, download configuration files to network-aware printers,
#       and to start the installation process for some operating systems.
service tftp
{
        socket_type             = dgram
        protocol                = udp
        wait                    = yes
        user                    = root
        server                  = /usr/sbin/in.tftpd
        server_args             = -s /tftpboot
#       disable                 = yes
        per_source              = 11
        cps                     = 100 2
}

Also, make sure that the /tftpboot directory exists and is world-readable (permissions at least "dr-xr-xr-x").

4.7. Configuration of a BOOTP / DHCP Server

BOOTP resp. DHCP can be used to automatically pass configuration information to the target. The only thing the target must "know" about itself is its own Ethernet hardware (MAC) address. The following command can be used to check if DHCP support is available on your host system:

$ rpm -q dhcp

If necessary, install the DHCP package from your distribution media.

Then you have to create the DHCP configuration file /etc/dhcpd.conf that matches your network setup. The following example gives you an idea what to do:

subnet 192.168.0.0 netmask 255.255.0.0 {
        option routers          192.168.1.1;
        option subnet-mask      255.255.0.0;

        option domain-name      "local.net";
        option domain-name-servers ns.local.net;

        host trgt {     hardware ethernet       00:30:BF:01:02:D0;
                        fixed-address           192.168.20.38;
                        option root-path        "/opt/eldk-5.2/armv5te/rootfs";
                        option host-name        "m28";
                        next-server             192.168.1.1;
                        filename                "/tftpboot/duts/m28/uImage";
        }
}

With this configuration, the DHCP server will reply to a request from the target with the ethernet address 00:30:BF:01:02:D0 with the following information:

• The target is located in the subnet 192.168.0.0 which uses the netmask 255.255.0.0.

• The target has the hostname m28 and the IP address 192.168.20.38.

• The host with the IP address 192.168.1.1 will provide the boot image for the target and provide NFS server function in cases when the target mounts it's root filesystem over NFS.
  The host listed with the next-server option can be different from the host that is running the DHCP server.

• The host provides the file /tftpboot/duts/m28/uImage as boot image for the target.

• The target can mount the directory /opt/eldk-5.2/armv5te/rootfs on the NFS server as root filesystem.

4.8. Configuring a NFS Server

For a development environment it is very convenient when the host and the target can share the same files over the network. The easiest way for such a setup is when the host provides NFS server functionality and exports a directory that can be mounted from the target as the root filesystem.

Assuming NFS server functionality is already provided by your host, the only configuration that needs to be added is an entry for your target root directory to your /etc/exports file, for instance like this:

/opt/eldk-5.2/armv5te/rootfs       192.168.0.0/255.255.0.0(rw,no_root_squash,sync)

This line exports the /opt/eldk-5.2/armv5te/rootfs directory with read and write permissions to all hosts on the 192.168.0.0 subnet.

After modifying the /etc/exports file you must make sure the NFS system is notified about the change, for instance by issuing the command:

# /sbin/service nfs restart

5. Das U-Boot

5.1. Current Versions

Das U-Boot (or just "U-Boot" for short) is Open Source Firmware for Embedded Power Architecture®, ARM, MIPS, x86 and other processors. The U-Boot project is hosted by DENX, where you can also find the project home page: http://www.denx.de/wiki/U-Boot/

The current version of the U-Boot source code can be retrieved from the DENX "git" repository.

You can browse the "git" repositories at http://git.denx.de/

The trees can be accessed through the git, HTTP, and rsync protocols. For example you can use one of the following commands to create a local clone of one of the source trees:

git clone git://git.denx.de/u-boot.git u-boot/
git clone http://git.denx.de/u-boot.git u-boot/
git clone rsync://git.denx.de/u-boot.git u-boot/

For details please see here.

Official releases of U-Boot are also available through FTP. Compressed tar archives can downloaded from the directory ftp://ftp.denx.de/pub/u-boot/.

5.2. Unpacking the Source Code

If you use GIT to get a copy of the U-Boot sources, here an example how you get the sources with git:

Note: Included topic DULGData_m28.UBootGetSource does not exist yet

If you used GIT to get a copy of the U-Boot sources, then you can skip this next step since you already have an unpacked directory tree. If you downloaded a compressed tarball from the DENX FTP server, you can unpack it as follows:

$ cd /opt/eldk/usr/src
$ wget ftp://ftp.denx.de/pub/u-boot/2012.07.tar.bz2
$ rm -f u-boot
$ bunzip2 < 2012.07.tar.bz2 | tar xf -
$ ln -s 2012.07 u-boot
$ cd u-boot

5.3. Configuration

After changing to the directory with the U-Boot source code you should make sure that there are no build results from any previous configurations left:

$ make distclean

The following (model) command configures U-Boot for the m28 board:

$ make m28_config

[marex@pollux]$

And finally we can compile the tools and U-Boot itself:

$ make all

By default the build is performed locally and the objects are saved in the source directory. One of the two methods can be used to change this behaviour and build U-Boot to some external directory:

1. Add O= to the make command line invocations:

make O=/tmp/build distclean
make O=/tmp/build m28_config
make O=/tmp/build all

Note that if the 'O=output/dir' option is used then it must be used for all invocations of make.

2. Set environment variable BUILD_DIR to point to the desired location:

export BUILD_DIR=/tmp/build
make distclean
make m28_config
make all

Note that the command line "O=" setting overrides the BUILD_DIR environment variable.

5.4. Installation

5.4.1. Before You Begin

5.4.1.1. Installation Requirements

The following section assumes that flash memory is used as the storage device for the firmware on your board. If this is not the case, the following instructions will not work - you will probably have to replace the storage device (probably ROM or EPROM) on such systems to install or update U-Boot.

5.4.1.2. Board Identification Data

All m28 boards use a serial number for identification purposes. Also, all boards have at least one ethernet (MAC) address assigned. You may lose your warranty on the board if this data gets lost. Before installing U-Boot or otherwise changing the software configuration of a board (like erasing some flash memory) you should make sure that you have all necessary information about such data.

5.4.2. Installation Using a BDM/JTAG Debugger

A fast and simple way to write new data to flash memory is via the use of a debugger or flash programmer with a BDM or JTAG interface. In cases where there is no running firmware at all (for instance on new hardware), this is usually the only way to install any software at all.

We use (and highly recommend) the BDI2000/BDI3000 by Abatron .

Other BDM / JTAG debuggers may work too, but how to use them is beyond the scope of this document. Please see the documentation for the tool you want to use.

Before you can use the BDI2000 you have to configure it. A configuration file that can be used with m28 boards is included in section 13.2. BDI2000 Configuration file

To install a new U-Boot image on your m28 board using a BDI2000, proceed as follows:

Note: Included topic DULGData_m28.InstallUBootUsingBDI2000 does not exist yet

5.4.3. Installation using U-Boot

If U-Boot is already installed and running on your board, you can use these instructions to download another U-Boot image to replace the current one.

Warning: Before you can install the new image, you have to erase the current one. If anything goes wrong your board will be dead. It is strongly recommended that:

• you have a backup of the old, working U-Boot image
• you know how to install an image on a virgin system

Proceed as follows:

=> printenv load update
load=echo $update_sd_firmware ; run update_sd_firmware
update=echo done
=> setenv u-boot /tftpboot/duts/m28/u-boot.bin
=> run load update
if mmc rescan ; then if tftp ${update_sd_firmware_filename} ; then setexpr fw_sz ${filesize} / 0x200 ; setexpr fw_sz ${fw_sz} + 1 ; mmc write ${loadaddr} 0x800 ${fw_sz} ; fi ; fi
Using FEC0 device
TFTP from server 192.168.1.1; our IP address is 192.168.20.33
Filename 'duts/m28/u-boot.mx28.sd'.
Load address: 0x42000000
Loading: ###############################
done
Bytes transferred = 446080 (6ce80 hex)

MMC write: dev # 0, block # 2048, count 872 ... 872 blocks write: OK
done
=> reset
resetting ...


U-Boot 2012.07-00471-ge8925d7-dirty (Oct 01 2012 - 18:20:02)

CPU:   Freescale i.MX28 rev1.2 at 454 MHz
BOOT:  SSP SD/MMC #0, 3V3
I2C:   ready
DRAM:  256 MiB
NAND:  256 MiB
MMC:   MXS MMC: 0
In:    serial
Out:   serial
Err:   serial
Net:   FEC0 [PRIME], FEC1
Hit any key to stop autoboot:  0
=> version

U-Boot 2012.07-00471-ge8925d7-dirty (Oct 01 2012 - 18:20:02)
arm-linux-gnueabi-gcc (Debian 4.7.2-2) 4.7.2
GNU ld (GNU Binutils for Debian) 2.22
=>

5.5. Tool Installation

U-Boot uses a special image format when loading the Linux kernel or ramdisk or other images. This image contains (among other things) information about the time of creation, operating system, compression type, image type, image name and CRC32 checksums.

The tool mkimage is used to create such images or to display the information they contain. When using the ELDK, the mkimage command is already included with the other ELDK tools.

If you don't use the ELDK then you should install mkimage in some directory that is in your command search PATH, for instance:

$ cp tools/mkimage /usr/local/bin/

5.6. Initialization

To initialize the U-Boot firmware running on your m28 board, you have to connect a terminal to the board's serial console port.

The default configuration of the console port on the m28 board uses a baudrate of 115200/8N1 (115200 bps, 8 Bit per character, no parity, 1 stop bit, no handshake).

If you are running Linux on your host system we recommend either kermit or cu as terminal emulation programs. Do not use minicom, since this has caused problems for many users, especially for software download over the serial port.

For the configuration of your terminal program see section 4.1. Serial Console Access

Make sure that both hardware and software flow control are disabled.

5.7. Initial Steps

In the default configuration, U-Boot operates in an interactive mode which provides a simple command line-oriented user interface using a serial console on port ttyAMA0.

In the simplest case, this means that U-Boot shows a prompt (default: =>) when it is ready to receive user input. You then type a command, and press enter. U-Boot will try to run the required action(s), and then prompt for another command.

To see a list of the available U-Boot commands, you can type help (or simply ?). This will print a list of all commands that are available in your current configuration. [Please note that U-Boot provides a lot of configuration options; not all options are available for all processors and boards, and some options might be simply not selected for your configuration.]

=>
=> hel

With the command help <command> you can get additional information about most commands:

=> help tftpboot
tftpboot - boot image via network using TFTP protocol

Usage:
tftpboot [loadAddress] [[hostIPaddr:]bootfilename]
=> help setenv printenv
setenv - set environment variables

Usage:
setenv name value ...
    - set environment variable 'name' to 'value ...'
setenv name
    - delete environment variable 'name'
printenv - print environment variables

Usage:
printenv
    - print values of all environment variables
printenv name ...
    - print value of environment variable 'name'
=>

Most commands can be abbreviated as long as the string remains unambiguous:

=> help printe tftp
printenv - print environment variables

Usage:
printenv
    - print values of all environment variables
printenv name ...
    - print value of environment variable 'name'
tftpboot - boot image via network using TFTP protocol

Usage:
tftpboot [loadAddress] [[hostIPaddr:]bootfilename]
=>

5.8. The First Power-On

Note: If you bought your m28 board with U-Boot already installed, you can skip this section since the manufacturer probably has already performed these steps.

Connect the port labeled ttyAMA0 on your m28 board to the designated serial port of your host, start the terminal program, and connect the power supply of your m28 board. You should see messages like this:

=>
=> reset
resetting ...


U-Boot 2012.07-00471-ge8925d7-dirty (Oct 01 2012 - 18:20:02)

CPU:   Freescale i.MX28 rev1.2 at 454 MHz
BOOT:  SSP SD/MMC #0, 3V3
I2C:   ready
DRAM:  256 MiB
NAND:  256 MiB
MMC:   MXS MMC: 0
In:    serial
Out:   serial
Err:   serial
Net:   FEC0 [PRIME], FEC1
Hit any key to stop autoboot:  0
=>

You can interrupt the "Count-Down" by pressing any key. If you don't you will probably see some (harmless) error messages because the system has not been initialized yet.

In some cases you may see a message

*** Warning - bad CRC, using default environment

This is harmless and will go away as soon as you have initialized and saved the environment variables.

At first you have to enter the serial number and the ethernet address of your board. Pay special attention here since these parameters are write protected and cannot be changed once saved (usually this is done by the manufacturer of the board). To enter the data you have to use the U-Boot command setenv, followed by the variable name and the data, all separated by white space (blank and/or TAB characters). Use the variable name serial# for the board ID and/or serial number, and ethaddr for the ethernet address, for instance:

=> setenv serial# DUTS
=> setenv ethaddr !!!!!!FILL_THIS!!!!!!
Can't overwrite "ethaddr"
=>

Use the printenv command to verify that you have entered the correct values:

=> printenv serial# ethaddr
serial#=DUTS
ethaddr=C0:E5:4E:02:00:00
=>

Please double check that the printed values are correct! You will not be able to correct any errors later! If there is something wrong, reset the board and restart from the beginning; otherwise you can store the parameters permanently using the saveenv command:

=> saveenv
Saving Environment to NAND...
Erasing NAND...
Erasing at 0x300000 --  25% complete.Erasing at 0x320000 --  50% complete.Erasing at 0x340000 --  75% complete.Erasing at 0x360000 -- 100% complete.
Writing to NAND... done
=>

5.9. U-Boot Command Line Interface

The following section describes the most important commands available in U-Boot. Please note that U-Boot is highly configurable, so not all of these commands may be available in the configuration of U-Boot installed on your hardware, or additional commands may exist. You can use the help command to print a list of all available commands for your configuration.

For most commands, you do not need to type in the full command name; instead it is sufficient to type a few characters. For instance, help can be abbreviated as h.

The behaviour of some commands depends on the configuration of U-Boot and on the definition of some variables in your U-Boot environment.

Almost all U-Boot commands expect numbers to be entered in hexadecimal input format. (Exception: for historical reasons, the sleep command takes it's argument in decimal input format.)

Be careful not to use edit keys besides 'Backspace', as hidden characters in things like environment variables can be very difficult to find.

5.9.1. Information Commands

5.9.1.1. bdinfo - print Board Info structure

=> help bdinfo
bdinfo - print Board Info structure

Usage:
bdinfo
=>

The bdinfo command (short: bdi) prints the information that U-Boot passes about the board such as memory addresses and sizes, clock frequencies, MAC address, etc. This information is mainly needed to be passed to the Linux kernel.

=> bdi
arch_number = 0x00000E1D
boot_params = 0x40000100
DRAM bank   = 0x00000000
-> start    = 0x40000000
-> size     = 0x10000000
ethaddr     = C0:E5:4E:02:00:00
ip_addr     = 192.168.20.33
baudrate    = 115200 bps
TLB addr    = 0x4FFF0000
relocaddr   = 0x4FF48000
reloc off   = 0x0FF47F00
irq_sp      = 0x4FB43F68
sp start    = 0x4FB43F58
FB base     = 0x00000000
=>

5.9.1.2. coninfo - print console devices and informations

=> help conin
coninfo - print console devices and information

Usage:
coninfo
=>

The coninfo command (short: conin) displays information about the available console I/O devices.

=> conin
List of available devices:
serial   80000003 SIO stdin stdout stderr
=>

The output contains the device name, flags, and the current usage. For example, the output

serial   80000003 SIO stdin stdout stderr

means that the serial device is a system device (flag 'S') which provides input (flag 'I') and output (flag 'O') functionality and is currently assigned to the 3 standard I/O streams stdin, stdout and stderr.

5.9.1.3. flinfo - print FLASH memory information

Note: Included topic DULGData_m28.UBootFlinfoHelp does not exist yet

The command flinfo (short: fli) can be used to get information about the available flash memory (see Flash Memory Commands below).

Note: Included topic DULGData_m28.UBootFlinfo does not exist yet

5.9.1.4. iminfo - print header information for application image

=> help iminfo
iminfo - print header information for application image

Usage:
iminfo addr [addr ...]
    - print header information for application image starting at
      address 'addr' in memory; this includes verification of the
      image contents (magic number, header and payload checksums)
=>

iminfo (short: imi) is used to print the header information for images like Linux kernels or ramdisks. It prints (among other information) the image name, type and size and verifies that the CRC32 checksums stored within the image are OK.

=> tftp ${ram_ws} ${bootfile}
Using FEC0 device
TFTP from server 192.168.1.1; our IP address is 192.168.20.33
Filename '/tftpboot/duts/m28/uImage'.
Load address: 0x42000000
Loading: #################################################################
         #################################################################
         ####################
done
Bytes transferred = 2191008 (216ea0 hex)
=> imi ${ram_ws}

## Checking Image at 42000000 ...
   Legacy image found
   Image Name:   Linux-3.6.0-next-20121001-00014-
   Created:      2012-10-02  13:23:40 UTC
   Image Type:   ARM Linux Kernel Image (uncompressed)
   Data Size:    2190944 Bytes = 2.1 MiB
   Load Address: 40008000
   Entry Point:  40008000
   Verifying Checksum ... OK
=>

Like with many other commands, the exact operation of this command can be controlled by the settings of some U-Boot environment variables (here: the verify variable). See below for details.

5.9.1.5. help - print online help

=> help help
help - print command description/usage

Usage:
help
        - print brief description of all commands
help command ...
        - print detailed usage of 'command'
=>

The help command (short: h or ?) prints online help. Without any arguments, it prints a list of all U-Boot commands that are available in your configuration of U-Boot. You can get detailed information for a specific command by typing its name as argument to the help command:

Note: Included topic DULGData_m28.UBootProtectHelp does not exist yet

5.9.2. Memory Commands

5.9.2.1. base - print or set address offset

=> help base
base - print or set address offset

Usage:
base
    - print address offset for memory commands
base off
    - set address offset for memory commands to 'off'
=>

You can use the base command (short: ba) to print or set a "base address" that is used as address offset for all memory commands; the default value of the base address is 0, so all addresses you enter are used unmodified. However, when you repeatedly have to access a certain memory region (like the internal memory of some embedded Power Architecture® processors) it can be very convenient to set the base address to the start of this area and then use only the offsets:

=> base
Base Address: 0x00000000
=> md 0 0xc
00000000: ea00000e ea00000e ea00000e ea00000e    ................
00000010: ea00000e ea00000e ea00000e ea00000e    ................
00000020: 4ff48000 4ff48004 4ff48008 4ff4800c    ...O...O...O...O
=> base 0x42000000
Base Address: 0x42000000
=> md 0 0xc
42000000: 00000001 00000001 00000001 00000001    ................
42000010: 00000000 00000000 00000001 00000801    ................
42000020: 00000367 00000000 00000000 00000000    g...............
=>

5.9.2.2. crc32 - checksum calculation

The crc32 command (short: crc) can be used to caculate a CRC32 checksum over a range of memory:

=>  crc 0x42000004 0x3FC
CRC32 for 42000004 ... 420003ff ==> 1ec64665
=>

When used with 3 arguments, the command stores the calculated checksum at the given address:

=>  crc 0x42000004 0x3FC 0x42000000
CRC32 for 42000004 ... 420003ff ==> 1ec64665
=> md 0x42000000 4
42000000: 1ec64665 00000001 00000001 00000001    eF..............
=>

As you can see, the CRC32 checksum was not only printed, but also stored at address 0x100000.

5.9.2.3. cmp - memory compare

=> help cmp
cmp - memory compare

Usage:
cmp [.b, .w, .l] addr1 addr2 count
=>

With the cmp command you can test of the contents of two memory areas is identical or not. The command will either test the whole area as specified by the 3rd (length) argument, or stop at the first difference.

=> cmp 0x42000000 0x43000000 0x400
Total of 1024 word(s) were the same
=> md 0x42000000 0xc
42000000: 56190527 8daa7c80 dcea6a50 606e2100    '..V.|..Pj...!n`
42000010: 00800040 00800040 9de1110f 00020205    @...@...........
42000020: 756e694c 2e332d78 2d302e36 7478656e    Linux-3.6.0-next
=> md 0x43000000 0xc
43000000: 56190527 8daa7c80 dcea6a50 606e2100    '..V.|..Pj...!n`
43000010: 00800040 00800040 9de1110f 00020205    @...@...........
43000020: 756e694c 2e332d78 2d302e36 7478656e    Linux-3.6.0-next
=>

Like most memory commands the cmp can access the memory in different sizes: as 32 bit (long word), 16 bit (word) or 8 bit (byte) data. If invoked just as cmp the default size (32 bit or long words) is used; the same can be selected explicitely by typing cmp.l instead. If you want to access memory as 16 bit or word data, you can use the variant cmp.w instead; and to access memory as 8 bit or byte data please use cmp.b.

Please note that the count argument specifies the number of data items to process, i. e. the number of long words or words or bytes to compare.

=> cmp.l 0x42000000 0x43000000 0x400
Total of 1024 word(s) were the same
=> cmp.w 0x42000000 0x43000000 0x800
Total of 2048 halfword(s) were the same
=> cmp.b 0x42000000 0x43000000 0x1000
Total of 4096 byte(s) were the same
=>

5.9.2.4. cp - memory copy

=> help cp
cp - memory copy

Usage:
cp [.b, .w, .l] source target count
=> help cp
cp - memory copy

Usage:
cp [.b, .w, .l] source target count
=>

The cp is used to copy memory areas.

=> cp 0x42000000 0x43000000 0x10000
=>

The cp understands the type extensions .l, .w and .b :

=> cp.l 0x43000000 0x42000000 0x10000
=> cp.w 0x43000000 0x42000000 0x20000
=> cp.b 0x43000000 0x42000000 0x40000
=>

5.9.2.5. md - memory display

=> help md
md - memory display

Usage:
md [.b, .w, .l] address [# of objects]
=>

The md can be used to display memory contents both as hexadecimal and ASCII data.

=> md 0x42000000
42000000: 56190527 8daa7c80 dcea6a50 606e2100    '..V.|..Pj...!n`
42000010: 00800040 00800040 9de1110f 00020205    @...@...........
42000020: 756e694c 2e332d78 2d302e36 7478656e    Linux-3.6.0-next
=>
42000030: 3130322d 30303132 30302d31 2d343130    -20121001-00014-
42000040: e1a00000 e1a00000 e1a00000 e1a00000    ................
42000050: e1a00000 e1a00000 e1a00000 e1a00000    ................
=>

This command, too, can be used with the type extensions .l, .w and .b :

=>
=> md.w 0x42000000
42000000: 0527 5619 7c80 8daa 6a50 dcea 2100 606e    '..V.|..Pj...!n`
42000010: 0040 0080 0040 0080    @...@...
=> d.b 0x4200000

The last displayed memory address and the value of the count argument are remembered, so when you enter md again without arguments it will automatically continue at the next address, and use the same count again.

=>
=> md.b 0x42000000 0x20
42000000: 27 05 19 56 80 7c aa 8d 50 6a ea dc 00 21 6e 60    '..V.|..Pj...!n`
42000010: 40 00 80 00 40 00 80 00 0f 11 e1 9d 05 02 02 00    @...@...........
=> d.w 0x42000000
Unknown command 'd.w' - try 'help'
=> md 0x4200000

5.9.2.6. mm - memory modify (auto-incrementing)

=>
=> help

The mm is a method to interactively modify memory contents. It will display the address and current contents and then prompt for user input. If you enter a legal hexadecimal number, this new value will be written to the address. Then the next address will be prompted. If you don't enter any value and just press ENTER, then the contents of this address will remain unchanged. The command stops as soon as you enter any data that is not a hex number (like .):

=>
=>
=> mm 0x42000000
42000000: 56190527 ? 0
42000004: 8daa7c80 ? 0xaabbccdd
42000008: dcea6a50 ? 0x01234567
4200000c: 606e2100 ? .
=> md 0x42000000 0x10
42000000: 00000000 aabbccdd 01234567 606e2100    ........gE#..!n`
42000010: 00800040 00800040 9de1110f 00020205    @...@...........
42000020: 756e694c 2e332d78 2d302e36 7478656e    Linux-3.6.0-next
42000030: 3130322d 30303132 30302d31 2d343130    -20121001-00014-
=>

Again this command can be used with the type extensions .l, .w and .b :

=>
=> mm.w 0x42000000
42000000: 0000 ? 0x0101
42000002: 0000 ? 0x0202
42000004: ccdd ? 0x4321
42000006: aabb ? 0x8765
42000008: 4567 ? .
=> md 0x42000000 0x10
42000000: 02020101 87654321 01234567 606e2100    ....!Ce.gE#..!n`
42000010: 00800040 00800040 9de1110f 00020205    @...@...........
42000020: 756e694c 2e332d78 2d302e36 7478656e    Linux-3.6.0-next
42000030: 3130322d 30303132 30302d31 2d343130    -20121001-00014-
=>

=>
=> mm.b 0x42000000
42000000: 01 ? 0x48
42000001: 01 ? 0x65
42000002: 02 ? 0x6c
42000003: 02 ? 0x6c
42000004: 21 ? 0x6f
42000005: 43 ? 0x20
42000006: 65 ? 0x20
42000007: 87 ? 0x20
42000008: 67 ? .
=> md 0x42000000 0x10
42000000: 6c6c6548 2020206f 01234567 606e2100    Hello   gE#..!n`
42000010: 00800040 00800040 9de1110f 00020205    @...@...........
42000020: 756e694c 2e332d78 2d302e36 7478656e    Linux-3.6.0-next
42000030: 3130322d 30303132 30302d31 2d343130    -20121001-00014-
=>

5.9.2.7. mtest - simple RAM test

=> help mtest
mtest - simple RAM read/write test

Usage:
mtest [start [end [pattern [iterations]]]]
=>

The mtest provides a simple memory test.

=>
=> mtest 0x42000000 0x42100000
Pattern 00000000  Writing...  Reading...Pattern FFFFFFFF  Writing...  Reading...Pattern 00000001  Writing...  Reading...Pattern FFFFFFFE  Writing...  Reading...Pattern 00000002  Writing...  Reading...Pattern FFFFFFFD  Writing...  Reading...Pattern 00000003  Writing...  Reading...Pattern FFFFFFFC  Writing...  Reading...Pattern 00000004  Writing...  Reading...Pattern FFFFFFFB  Writing...  Reading...Pattern 00000005  Writing...  Reading...Pattern FFFFFFFA  Writing...  Reading...Pattern 00000006  Writing...  Reading...Pattern FFFFFFF9  Writing...  Reading...Pattern 00000007  Writing...  Reading...Pattern FFFFFFF8  Writing...  Reading...Pattern 00000008  Writing...  Reading...Pattern FFFFFFF7  Writing...  Reading...Pattern 00000009  Writing...  Reading...Pattern FFFFFFF6  Writing...  Reading...Pattern 0000000A  Writing...  Reading...Pattern FFFFFFF5  Writing...  Reading...Pattern 0000000B  Writing...  Reading...Pattern FFFFFFF4  Writing...  Reading...Pattern 0000000C  Writing...  Reading...Pattern FFFFFFF3  Writing...  Reading...Pattern 0000000D  Writing...  Reading...Pattern FFFFFFF2  Writing...  Reading...Pattern 0000000E  Writing...  Reading...Pattern FFFFFFF1  Writing...  Reading...Pattern 0000000F  Writing...  Reading...Pattern FFFFFFF0  Writing...  Reading...Pattern 00000010  Writing...  Reading...Pattern FFFFFFEF  Writing...  Reading...Pattern 00000011  Writing...  Reading...Pattern FFFFFFEE  Writing...  Reading...
=>

This tests writes to memory, thus modifying the memory contents. It will fail when applied to ROM or flash memory.

This command may crash the system when the tested memory range includes areas that are needed for the operation of the U-Boot firnware (like exception vector code, or U-Boot's internal program code, stack or heap memory areas).

5.9.2.8. mw - memory write (fill)

=> help mw
mw - memory write (fill)

Usage:
mw [.b, .w, .l] address value [count]
=>

The mw is a way to initialize (fill) memory with some value. When called without a count argument, the value will be written only to the specified address. When used with a count, then a whole memory areas will be initialized with this value:

=> md 0x42000000 0x10
42000000: ffffffee ffffffed ffffffec ffffffeb    ................
42000010: ffffffea ffffffe9 ffffffe8 ffffffe7    ................
42000020: ffffffe6 ffffffe5 ffffffe4 ffffffe3    ................
42000030: ffffffe2 ffffffe1 ffffffe0 ffffffdf    ................
=> mw 0x42000000 0xaabbccdd
=> md 0x42000000 0x10
42000000: aabbccdd ffffffed ffffffec ffffffeb    ................
42000010: ffffffea ffffffe9 ffffffe8 ffffffe7    ................
42000020: ffffffe6 ffffffe5 ffffffe4 ffffffe3    ................
42000030: ffffffe2 ffffffe1 ffffffe0 ffffffdf    ................
=> mw 0x42000000 0 6
=> md 0x42000000 0x10
42000000: 00000000 00000000 00000000 00000000    ................
42000010: 00000000 00000000 ffffffe8 ffffffe7    ................
42000020: ffffffe6 ffffffe5 ffffffe4 ffffffe3    ................
42000030: ffffffe2 ffffffe1 ffffffe0 ffffffdf    ................
=>

This is another command that accepts the type extensions .l, .w and .b :

=>  mw.w 0x42000004 0x1155 6
=> md 0x42000000 0x10
42000000: 00000000 11551155 11551155 11551155    ....U.U.U.U.U.U.
42000010: 00000000 00000000 ffffffe8 ffffffe7    ................
42000020: ffffffe6 ffffffe5 ffffffe4 ffffffe3    ................
42000030: ffffffe2 ffffffe1 ffffffe0 ffffffdf    ................
=>  mw.b 0x42000007 0xff 7
=> md 0x42000000 0x10
42000000: 00000000 ff551155 ffffffff 1155ffff    ....U.U.......U.
42000010: 00000000 00000000 ffffffe8 ffffffe7    ................
42000020: ffffffe6 ffffffe5 ffffffe4 ffffffe3    ................
42000030: ffffffe2 ffffffe1 ffffffe0 ffffffdf    ................
=>

5.9.2.9. nm - memory modify (constant address)

=> help nm
nm - memory modify (constant address)

Usage:
nm [.b, .w, .l] address
=>

The nm command (non-incrementing memory modify) can be used to interactively write different data several times to the same address. This can be useful for instance to access and modify device registers:

=>
=> nm.b 0x42000000
42000000: 00 ? 0x48
42000000: 48 ? 0x65
42000000: 65 ? 0x6c
42000000: 6c ? 0x6c
42000000: 6c ? 0x6f
42000000: 6f ? .
=> md 0x42000000 8
42000000: 0000006f ff551155 ffffffff 1155ffff    o...U.U.......U.
42000010: 00000000 00000000 ffffffe8 ffffffe7    ................
=>

The nm command too accepts the type extensions .l, .w and .b.

5.9.2.10. loop - infinite loop on address range

=> help loop
loop - infinite loop on address range

Usage:
loop [.b, .w, .l] address number_of_objects
=>

The loop command reads in a tight loop from a range of memory. This is intended as a special form of a memory test, since this command tries to read the memory as fast as possible.

This command will never terminate. There is no way to stop it but to reset the board!

=> loop 100000 8

5.9.3. Flash Memory Commands

5.9.3.1. cp - memory copy

=> help cp
cp - memory copy

Usage:
cp [.b, .w, .l] source target count
=> help cp
cp - memory copy

Usage:
cp [.b, .w, .l] source target count
=>

The cp command "knows" about flash memory areas and will automatically invoke the necessary flash programming algorithm when the target area is in flash memory.

Note: Included topic DULGData_m28.UBootCpFlash does not exist yet

Writing to flash memory may fail when the target area has not been erased (see erase below), or if it is write-protected (see protect below).

Note: Included topic DULGData_m28.UBootCpFlashErrors does not exist yet

Remember that the count argument specifies the number of items to copy. If you have a "length" instead (= byte count) you should use cp.b or you will have to calculate the correct number of items.

5.9.3.2. flinfo - print FLASH memory information

The command flinfo (short: fli) can be used to get information about the available flash memory. The number of flash banks is printed with information about the size and organization into flash "sectors" or erase units. For all sectors the start addresses are printed; write-protected sectors are marked as read-only (RO). Some configurations of U-Boot also mark empty sectors with an (E).

Note: Included topic DULGData_m28.UBootFlinfo does not exist yet

5.9.3.3. erase - erase FLASH memory

Note: Included topic DULGData_m28.UBootEraseHelp does not exist yet

The erase command (short: era) is used to erase the contents of one or more sectors of the flash memory. It is one of the more complex commands; the help output shows this.

Probably the most frequent usage of this command is to pass the start and end addresses of the area to be erased:

Note: Included topic DULGData_m28.UBootEraseStartEnd does not exist yet

Note that both the start and end addresses for this command must point exactly at the start resp. end addresses of flash sectors. Otherwise the command will not be executed.

Another way to select certain areas of the flash memory for the erase command uses the notation of flash banks and sectors:

Technically speaking, a bank is an area of memory implemented by one or more memory chips that are connected to the same chip select signal of the CPU, and a flash sector or erase unit is the smallest area that can be erased in one operation.

For practical purposes it is sufficient to remember that with flash memory a bank is something that eventually may be erased as a whole in a single operation. This may be more efficient (faster) than erasing the same area sector by sector.

[It depends on the actual type of flash chips used on the board if such a fast bank erase algorithm exists, and on the implementation of the flash device driver if is actually used.]

In U-Boot, flash banks are numbered starting with 1, while flash sectors start with 0.

To erase the same flash area as specified using start and end addresses in the example above you could also type:

Note: Included topic DULGData_m28.UBootEraseSectors does not exist yet

To erase a whole bank of flash memory you can use a command like this one:

Note: Included topic DULGData_m28.UBootEraseBank does not exist yet

Note that a warning message is printed because some write protected sectors exist in this flash bank which were not erased.

With the command:

Note: Included topic DULGData_m28.UBootEraseAll does not exist yet

the whole flash memory (except for the write-protected sectors) can be erased.

5.9.3.4. protect - enable or disable FLASH write protection

Note: Included topic DULGData_m28.UBootProtectHelp does not exist yet

The protect command is another complex one. It is used to set certain parts of the flash memory to read-only mode or to make them writable again. Flash memory that is "protected" (= read-only) cannot be written (with the cp command) or erased (with the erase command). Protected areas are marked as (RO) (for "read-only") in the output of the flinfo command:

Note: Included topic DULGData_m28.UBootProtect does not exist yet

The actual level of protection depends on the flash chips used on your hardware, and on the implementation of the flash device driver for this board. In most cases U-Boot provides just a simple software-protection, i. e. it prevents you from erasing or overwriting important stuff by accident (like the U-Boot code itself or U-Boot's environment variables), but it cannot prevent you from circumventing these restrictions - a nasty user who is loading and running his own flash driver code cannot and will not be stopped by this mechanism. Also, in most cases this protection is only effective while running U-Boot, i. e. any operating system will not know about "protected" flash areas and will happily erase these if requested to do so.

5.9.3.5. mtdparts - define a Linux compatible MTD partition scheme

U-Boot implements two different approaches to define a MTD partition scheme that can be shared easily with the linux kernel.

The first one is to define a single, static partition in your board config file, for example:

#undef CONFIG_JFFS2_CMDLINE
#define CONFIG_JFFS2_DEV               "nor0"
#define CONFIG_JFFS2_PART_SIZE         0xFFFFFFFF     /* use whole device */
#define CONFIG_JFFS2_PART_SIZE         0x00100000     /* use 1MB */
#define CONFIG_JFFS2_PART_OFFSET       0x00000000

#define CONFIG_CMD_MTDPARTS
#define MTDIDS_DEFAULT      "nor1=zuma-1,nor2=zuma-2"
#define MTDPARTS_DEFAULT   "mtdparts=zuma-1:-(jffs2),zuma-2:-(user)"

The mtdparts command offers an easy to use and powerful interface to define the contents of the environment variable of the same name that can be passed as boot argument to the Linux kernel:

=> help mtdparts
mtdparts - define flash/nand partitions

Usage:
mtdparts
    - list partition table
mtdparts delall
    - delete all partitions
mtdparts del part-id
    - delete partition (e.g. part-id = nand0,1)
mtdparts add <mtd-dev> <size>[@<offset>] [<name>] [ro]
    - add partition
mtdparts default
    - reset partition table to defaults

-----

this command uses three environment variables:

'partition' - keeps current partition identifier

partition  := <part-id>
<part-id>  := <dev-id>,part_num

'mtdids' - linux kernel mtd device id <-> u-boot device id mapping

mtdids=<idmap>[,<idmap>,...]

<idmap>    := <dev-id>=<mtd-id>
<dev-id>   := 'nand'|'nor'|'onenand'<dev-num>
<dev-num>  := mtd device number, 0...
<mtd-id>   := unique device tag used by linux kernel to find mtd device (mtd->name)

'mtdparts' - partition list

mtdparts=mtdparts=<mtd-def>[;<mtd-def>...]

<mtd-def>  := <mtd-id>:<part-def>[,<part-def>...]
<mtd-id>   := unique device tag used by linux kernel to find mtd device (mtd->name)
<part-def> := <size>[@<offset>][<name>][<ro-flag>]
<size>     := standard linux memsize OR '-' to denote all remaining space
<offset>   := partition start offset within the device
<name>     := '(' NAME ')'
<ro-flag>  := when set to 'ro' makes partition read-only (not used, passed to kernel)
=>

For example, on the m28 target system the mtdparts command display this information:

=> mtdparts

device nand0 <gpmi-nand>, # parts = 5
 #: name                size            offset          mask_flags
 0: bootloader          0x00300000      0x00000000      1
 1: environment         0x00080000      0x00300000      0
 2: redundant-environment0x00080000     0x00380000      0
 3: kernel              0x00400000      0x00400000      0
 4: filesystem          0x0f800000      0x00800000      0

active partition: nand0,0 - (bootloader) 0x00300000 @ 0x00000000

defaults:
mtdids  : nand0=gpmi-nand
mtdparts: mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel),-(filesystem)
=>

The partition table printed here obviously differs from the default value for the mtdparts variable printed in the last line. To verify this, we can check the current content of this variable:

=> print mtdparts
mtdparts=mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel),-(filesystem)
=>

and we can see that it exactly matches the partition table printed above.

Then we delete the last 2 partitions ...

=> print mtdparts
mtdparts=mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel),-(filesystem)
=> mtdparts del kernel
=> mtdparts del filesystem
=> mtdparts

device nand0 <gpmi-nand>, # parts = 3
 #: name                size            offset          mask_flags
 0: bootloader          0x00300000      0x00000000      1
 1: environment         0x00080000      0x00300000      0
 2: redundant-environment0x00080000     0x00380000      0

active partition: nand0,0 - (bootloader) 0x00300000 @ 0x00000000

defaults:
mtdids  : nand0=gpmi-nand
mtdparts: mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel),-(filesystem)
=>

... and combine the free space into a singe big partition:

=> print mtdparts
mtdparts=mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel)
=> mtdparts add nand0 - filesystem
=> mtdparts

device nand0 <gpmi-nand>, # parts = 5
 #: name                size            offset          mask_flags
 0: bootloader          0x00300000      0x00000000      1
 1: environment         0x00080000      0x00300000      0
 2: redundant-environment0x00080000     0x00380000      0
 3: kernel              0x00400000      0x00400000      0
 4: filesystem          0x0f800000      0x00800000      0

active partition: nand0,0 - (bootloader) 0x00300000 @ 0x00000000

defaults:
mtdids  : nand0=gpmi-nand
mtdparts: mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel),-(filesystem)
=>

Now let's switch back to the default settings:

=> mtdparts default
=> mtdparts

device nand0 <gpmi-nand>, # parts = 5
 #: name                size            offset          mask_flags
 0: bootloader          0x00300000      0x00000000      1
 1: environment         0x00080000      0x00300000      0
 2: redundant-environment0x00080000     0x00380000      0
 3: kernel              0x00400000      0x00400000      0
 4: filesystem          0x0f800000      0x00800000      0

active partition: nand0,0 - (bootloader) 0x00300000 @ 0x00000000

defaults:
mtdids  : nand0=gpmi-nand
mtdparts: mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel),-(filesystem)
=>

5.9.3.6. UBI Usage in U-Boot

As in Linux, UBI access in U-Boot refers to MTD partitions, either through their partition number (like "nand0,7") or partition name (like "userfs"). So let's first check the partitions on the device:

=> mtdparts

device nand0 <gpmi-nand>, # parts = 5
 #: name                size            offset          mask_flags
 0: bootloader          0x00300000      0x00000000      1
 1: environment         0x00080000      0x00300000      0
 2: redundant-environment0x00080000     0x00380000      0
 3: kernel              0x00400000      0x00400000      0
 4: filesystem          0x0f800000      0x00800000      0

active partition: nand0,0 - (bootloader) 0x00300000 @ 0x00000000

defaults:
mtdids  : nand0=gpmi-nand
mtdparts: mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel),-(filesystem)
=>

U-Boot provides the following command line interface to UBI:

=> help ubi
ubi - ubi commands

Usage:
ubi part [part] [offset]
 - Show or set current partition (with optional VID header offset)
ubi info [l[ayout]] - Display volume and ubi layout information
ubi create[vol] volume [size] [type] - create volume name with size
ubi write[vol] address volume size - Write volume from address with size
ubi read[vol] address volume [size] - Read volume to address with size
ubi remove[vol] volume - Remove volume
[Legends]
 volume: character name
 size: specified in bytes
 type: s[tatic] or d[ynamic] (default=dynamic)
=>

To make a UBI device available to U-Boot it needs to be attached. This is done using the "ubi part" command. If an UBI device exists on the specified MTD partition it will be attached, otherwise a new UBI device will be created.

* WARNING * "ubi part" will, without any warning, overwrite any existing data and create a new UBI device if you run it on a partition that does not contain an UBI device yet.

Let's attach the "userfs" partition to UBI:

=> ubi part filesystem 2048
Creating 1 MTD partitions on "nand0":
0x000000800000-0x000010000000 : "mtd=4"
UBI: attaching mtd1 to ubi0
UBI: physical eraseblock size:   131072 bytes (128 KiB)
UBI: logical eraseblock size:    126976 bytes
UBI: smallest flash I/O unit:    2048
UBI: VID header offset:          2048 (aligned 2048)
UBI: data offset:                4096
UBI: empty MTD device detected
UBI: create volume table (copy #1)
UBI: create volume table (copy #2)
UBI: attached mtd1 to ubi0
UBI: MTD device name:            "mtd=4"
UBI: MTD device size:            248 MiB
UBI: number of good PEBs:        1984
UBI: number of bad PEBs:         0
UBI: max. allowed volumes:       128
UBI: wear-leveling threshold:    4096
UBI: number of internal volumes: 1
UBI: number of user volumes:     0
UBI: available PEBs:             1961
UBI: total number of reserved PEBs: 23
UBI: number of PEBs reserved for bad PEB handling: 19
UBI: max/mean erase counter: 1/0
=>

Now that the UBI device is attached, this device can be accessed using the following commands:

ubi info        Display volume and ubi layout information
ubi createvol   Create UBI volume on UBI device
ubi removevol   Remove UBI volume from UBI device
ubi read        Read data from UBI volume to memory
ubi write       Write data from memory to UBI volume

For example display volume and ubi layout information with ubi info:

=> ubi info l
UBI: volume information dump:
UBI: vol_id          2147479551
UBI: reserved_pebs   2
UBI: alignment       1
UBI: data_pad        0
UBI: vol_type        3
UBI: name_len        13
UBI: usable_leb_size 126976
UBI: used_ebs        2
UBI: used_bytes      253952
UBI: last_eb_bytes   2
UBI: corrupted       0
UBI: upd_marker      0
UBI: name            layout volume

=>

There is another set of commands to access UBIFS file systems:

ubifsmount - mount UBIFS volume
ubifsls    - list files in a directory
ubifsload  - load file from an UBIFS filesystem

First, we have to mount the UBIFS file system:

=> ubifsmount filesystem
UBIFS: mounted UBI device 0, volume 0, name "filesystem"
UBIFS: mounted read-only
UBIFS: file system size:   247603200 bytes (241800 KiB, 236 MiB, 1950 LEBs)
UBIFS: journal size:       9023488 bytes (8812 KiB, 8 MiB, 72 LEBs)
UBIFS: media format:       w4/r0 (latest is w4/r0)
UBIFS: default compressor: LZO
UBIFS: reserved for root:  0 bytes (0 KiB)
=>

after successfully mounted we can access it:

=> ubifsls
               30  Tue Oct 02 14:57:09 2012  date_of_creation
               30  Tue Oct 02 14:57:11 2012  date_of_modification
               16  Tue Oct 02 14:57:08 2012  README
=> ubifsload 0x42000000 date_of_creation
Loading file 'date_of_creation' to addr 0x42000000 with size 30 (0x0000001e)...
Done
=> md 0x42000000
42000000: 20657554 2074634f 31203220 37353a36    Tue Oct  2 16:57
42000010: 2039303a 54534543 31303220 00000a32    :09 CEST 2012...
42000020: 00000800 0001f000 0000000e 000007c0    ................
42000030: 00800000 00000000 00000005 00000002    ................
42000040: 00000001 00000001 00000008 00000100    ................
42000050: 00000004 00000001 00000000 00000000    ................
42000060: 00000000 00000000 3b9aca00 71b3e5e9    ...........;...q
42000070: fa46cc48 aabe65a2 32ec78f1 00000000    H.F..e...x.2....
42000080: 00000000 00000000 00000000 00000000    ................
42000090: 00000000 00000000 00000000 00000000    ................
420000a0: 00000000 00000000 00000000 00000000    ................
420000b0: 00000000 00000000 00000000 00000000    ................
420000c0: 00000000 00000000 00000000 00000000    ................
420000d0: 00000000 00000000 00000000 00000000    ................
420000e0: 00000000 00000000 00000000 00000000    ................
420000f0: 00000000 00000000 00000000 00000000    ................
=> ubifsload 0x42000000 README
Loading file 'README' to addr 0x42000000 with size 16 (0x00000010)...
Done
=> md 0x42000000
42000000: 7473754a 206e6120 6d617865 0a656c70    Just an example.
42000010: 2039303a 54534543 31303220 00000a32    :09 CEST 2012...
42000020: 00000800 0001f000 0000000e 000007c0    ................
42000030: 00800000 00000000 00000005 00000002    ................
42000040: 00000001 00000001 00000008 00000100    ................
42000050: 00000004 00000001 00000000 00000000    ................
42000060: 00000000 00000000 3b9aca00 71b3e5e9    ...........;...q
42000070: fa46cc48 aabe65a2 32ec78f1 00000000    H.F..e...x.2....
42000080: 00000000 00000000 00000000 00000000    ................
42000090: 00000000 00000000 00000000 00000000    ................
420000a0: 00000000 00000000 00000000 00000000    ................
420000b0: 00000000 00000000 00000000 00000000    ................
420000c0: 00000000 00000000 00000000 00000000    ................
420000d0: 00000000 00000000 00000000 00000000    ................
420000e0: 00000000 00000000 00000000 00000000    ................
420000f0: 00000000 00000000 00000000 00000000    ................
=>

* NOTE * In U-Boot, there is no clean way to detach an UBI device; all you can do is to attach a different device - assuming there is another one; alternatively, you can give a non-existent partition name: this will give an error, but it will detach the previously attached UBI device:

=> ubi part foobar
incorrect device type in foobar
Partition foobar not found!
=>

Update of an UBI Volume using U-Boot:

=> ubi part filesystem 2048
Unmounting UBIFS volume filesystem!
UBI: mtd1 is detached from ubi0
Creating 1 MTD partitions on "nand0":
0x000000800000-0x000010000000 : "mtd=4"
UBI: attaching mtd1 to ubi0
UBI: physical eraseblock size:   131072 bytes (128 KiB)
UBI: logical eraseblock size:    126976 bytes
UBI: smallest flash I/O unit:    2048
UBI: VID header offset:          2048 (aligned 2048)
UBI: data offset:                4096
UBI: attached mtd1 to ubi0
UBI: MTD device name:            "mtd=4"
UBI: MTD device size:            248 MiB
UBI: number of good PEBs:        1984
UBI: number of bad PEBs:         0
UBI: max. allowed volumes:       128
UBI: wear-leveling threshold:    4096
UBI: number of internal volumes: 1
UBI: number of user volumes:     1
UBI: available PEBs:             0
UBI: total number of reserved PEBs: 1984
UBI: number of PEBs reserved for bad PEB handling: 19
UBI: max/mean erase counter: 4/1
=> ubifsmount filesystem
UBIFS: mounted UBI device 0, volume 0, name "filesystem"
UBIFS: mounted read-only
UBIFS: file system size:   247603200 bytes (241800 KiB, 236 MiB, 1950 LEBs)
UBIFS: journal size:       9023488 bytes (8812 KiB, 8 MiB, 72 LEBs)
UBIFS: media format:       w4/r0 (latest is w4/r0)
UBIFS: default compressor: LZO
UBIFS: reserved for root:  0 bytes (0 KiB)
=> ubifsls
               30  Tue Oct 02 14:57:09 2012  date_of_creation
               30  Tue Oct 02 14:57:11 2012  date_of_modification
               16  Tue Oct 02 14:57:08 2012  README
=> tftp 0x42000000 duts/m28/image-filesystem-update.ubifs
Using FEC0 device
TFTP from server 192.168.1.1; our IP address is 192.168.20.33
Filename 'duts/m28/image-filesystem-update.ubifs'.
Load address: 0x42000000
Loading: #################################################################
         #########################################################
done
Bytes transferred = 1777664 (1b2000 hex)
=> ubi write 0x42000000 filesystem ${filesize}
1777664 bytes written to volume filesystem
=> ubifsmount filesystem
UBIFS: mounted UBI device 0, volume 0, name "filesystem"
UBIFS: mounted read-only
UBIFS: file system size:   247603200 bytes (241800 KiB, 236 MiB, 1950 LEBs)
UBIFS: journal size:       9023488 bytes (8812 KiB, 8 MiB, 72 LEBs)
UBIFS: media format:       w4/r0 (latest is w4/r0)
UBIFS: default compressor: LZO
UBIFS: reserved for root:  0 bytes (0 KiB)
=> ubifsls
               30  Tue Oct 02 14:57:17 2012  date_of_modification.update
               30  Tue Oct 02 14:57:15 2012  date_of_creation.update
               15  Tue Oct 02 14:57:13 2012  README.update
=>

Useful definitions:

update_data=ubi part data;tftp 40000000 m28/image-data.ubifs;ubi write 40000000 data ${filesize}
update_user=ubi part user;tftp 40000000 m28/image-user.ubifs;ubi write 40000000 user ${filesize}

5.9.4. Execution Control Commands

5.9.4.1. source - run script from memory

=> help source
source - run script from memory

Usage:
source [addr]
        - run script starting at addr
        - A valid image header must be present
=>

With the source command you can run "shell" scripts under U-Boot: You create a U-Boot script image by simply writing the commands you want to run into a text file; then you will have to use the mkimage tool to convert this text file into a U-Boot image (using the image type script).

This image can be loaded like any other image file, and with source you can run the commands in such an image. For instance, the following text file:

[marex@pollux]$ echo
echo Network Configuration:
echo ----------------------
echo Target:
printenv ipaddr hostname
echo
echo Server:
printenv serverip rootpath
echo

can be converted into a U-Boot script image using the mkimage command like this:

[marex@pollux]$

Now you can load and execute this script image in U-Boot:

=> tftp 0x42000000 /tftpboot/duts/m28/example.scr
Using FEC0 device
TFTP from server 192.168.1.1; our IP address is 192.168.20.33
Filename '/tftpboot/duts/m28/example.scr'.
Load address: 0x42000000
Loading: #
done
Bytes transferred = 221 (dd hex)
=> imi

## Checking Image at 42000000 ...
   Legacy image found
   Image Name:   U-Boot example script
   Created:      2012-09-05  10:24:29 UTC
   Image Type:   PowerPC Linux Script (uncompressed)
   Data Size:    157 Bytes = 157 Bytes
   Load Address: 00000000
   Entry Point:  00000000
   Contents:
      Image 0: 149 Bytes = 149 Bytes
   Verifying Checksum ... OK
=> source 0x42000000
## Executing script at 42000000

Network Configuration:
----------------------
Target:
ipaddr=192.168.20.33
hostname=m28

Server:
serverip=192.168.1.1
rootpath=/opt/eldk-5.2.1/armv5te/rootfs-qte-sdk

=>

5.9.4.2. bootm - boot application image from memory

=> help bootm
bootm - boot application image from memory

Usage:
bootm [addr [arg ...]]
    - boot application image stored in memory
        passing arguments 'arg ...'; when booting a Linux kernel,
        'arg' can be the address of an initrd image
        When booting a Linux kernel which requires a flat device-tree
        a third argument is required which is the address of the
        device-tree blob. To boot that kernel without an initrd image,
        use a '-' for the second argument. If you do not pass a third
        a bd_info struct will be passed instead

Sub-commands to do part of the bootm sequence.  The sub-commands must be
issued in the order below (it's ok to not issue all sub-commands):
        start [addr [arg ...]]
        loados  - load OS image
        fdt     - relocate flat device tree
        cmdline - OS specific command line processing/setup
        bdt     - OS specific bd_t processing
        prep    - OS specific prep before relocation or go
        go      - start OS
=>

The bootm command is used to start operating system images. From the image header it gets information about the type of the operating system, the file compression method used (if any), the load and entry point addresses, etc. The command will then load the image to the required memory address, uncompressing it on the fly if necessary. Depending on the OS it will pass the required boot arguments and start the OS at it's entry point.

The first argument to bootm is the memory address (in RAM, ROM or flash memory) where the image is stored, followed by optional arguments that depend on the OS.

For Linux, exactly one optional argument can be passed. If it is present, it is interpreted as the start address of a initrd ramdisk image (in RAM, ROM or flash memory). In this case the bootm command consists of three steps: first the Linux kernel image is uncompressed and copied into RAM, then the ramdisk image is loaded to RAM, and finally controll is passed to the Linux kernel, passing information about the location and size of the ramdisk image.

To boot a Linux kernel image without a initrd ramdisk image, the following command can be used:

=> bootm ${kernel_addr}

If a ramdisk image shall be used, you can type:

=> bootm ${kernel_addr} ${ramdisk_addr}

Both examples of course imply that the variables used are set to correct addresses for a kernel and a initrd ramdisk image.

When booting images that have been loaded to RAM (for instance using TFTP download) you have to be careful that the locations where the (compressed) images were stored do not overlap with the memory needed to load the uncompressed kernel. For instance, if you load a ramdisk image at a location in low memory, it may be overwritten when the Linux kernel gets loaded. This will cause undefined system crashes.

5.9.4.3. go - start application at address 'addr'

=> help go
go - start application at address 'addr'

Usage:
go addr [arg ...]
    - start application at address 'addr'
      passing 'arg' as arguments
=>

U-Boot has support for so-called standalone applications. These are programs that do not require the complex environment of an operating system to run. Instead they can be loaded and executed by U-Boot directly, utilizing U-Boot's service functions like console I/O or malloc() and free().

This can be used to dynamically load and run special extensions to U-Boot like special hardware test routines or bootstrap code to load an OS image from some filesystem.

The go command is used to start such standalone applications. The optional arguments are passed to the application without modification. For more informatoin see 5.12. U-Boot Standalone Applications.

5.9.5. Download Commands

5.9.5.1. bootp - boot image via network using BOOTP/TFTP protocol

=> help bootp
bootp - boot image via network using BOOTP/TFTP protocol

Usage:
bootp [loadAddress] [[hostIPaddr:]bootfilename]
=>

5.9.5.2. dhcp - invoke DHCP client to obtain IP/boot params

=> help dhcp
dhcp - boot image via network using DHCP/TFTP protocol

Usage:
dhcp [loadAddress] [[hostIPaddr:]bootfilename]
=>

5.9.5.3. loadb - load binary file over serial line (kermit mode)

=> help loadb
loadb - load binary file over serial line (kermit mode)

Usage:
loadb [ off ] [ baud ]
    - load binary file over serial line with offset 'off' and baudrate 'baud'
=>

With kermit you can download binary data via the serial line. Here we show how to download uImage, the Linux kernel image. Please make sure, that you have set up kermit as described in section 4.3. Configuring the "kermit" command and then type:

=> loadb 100000
## Ready for binary (kermit) download ...
Ctrl-\c
(Back at denx.denx.de)
----------------------------------------------------
C-Kermit 7.0.197, 8 Feb 2000, for Linux
 Copyright (C) 1985, 2000,
  Trustees of Columbia University in the City of New York.
Type ? or HELP for help.
Kermit> send /bin /tftpboot/pImage
...
Kermit> connect
Connecting to /dev/ttyS0, speed 115200.
The escape character is Ctrl-\ (ASCII 28, FS)
Type the escape character followed by C to get back,
or followed by ? to see other options.
----------------------------------------------------
= 550260 Bytes
## Start Addr      = 0x00100000
=> iminfo 100000

## Checking Image at 00100000 ...
   Image Name:   Linux-2.4.4
   Created:      2002-07-02  22:10:11 UTC
   Image Type:   PowerPC Linux Kernel Image (gzip compressed)
   Data Size:    550196 Bytes = 537 kB = 0 MB
   Load Address: 00000000
   Entry Point:  00000000
   Verifying Checksum ... OK

5.9.5.4. loads - load S-Record file over serial line

=> help loads
loads - load S-Record file over serial line

Usage:
loads [ off ]
    - load S-Record file over serial line with offset 'off'
=>

5.9.5.5. tftpboot- boot image via network using TFTP protocol

=> help tftp
tftpboot - boot image via network using TFTP protocol

Usage:
tftpboot [loadAddress] [[hostIPaddr:]bootfilename]
=>

5.9.6. Environment Variables Commands

5.9.6.1. printenv- print environment variables

=> help printenv
printenv - print environment variables

Usage:
printenv
    - print values of all environment variables
printenv name ...
    - print value of environment variable 'name'
=>

The printenv command prints one, several or all variables of the U-Boot environment. When arguments are given, these are interpreted as the names of environment variables which will be printed with their values:

=> printenv ipaddr hostname netmask
ipaddr=192.168.20.33
hostname=m28
netmask=255.255.0.0
=>

Without arguments, printenv prints all a list with all variables in the environment and their values, plus some statistics about the current usage and the total size of the memory available for the environment.

=> printenv
addcons=setenv bootargs ${bootargs} console=${consdev},${baudrate}
addip=setenv bootargs ${bootargs} ip=${ipaddr}:${serverip}:${gatewayip}:${netmask}:${hostname}:${netdev}:off
addmisc=setenv bootargs ${bootargs} ${miscargs}
addmtd=setenv bootargs ${bootargs} ${mtdparts}
addvideo=setenv bootargs ${bootargs} video=imx28-fb
base_offset=44040000
baudrate=115200
bootcmd=run net_nfs
bootdelay=1
bootdev=/dev/mmcblk0p2
bootfile=duts/m28/uImage
cmp_addr_r=0x44000000
cons_opts=console=tty0 console=ttyS0,${baudrate}
consdev=ttyAMA0
eth1addr=C0:E5:4E:02:00:01
ethact=FEC0
ethaddr=C0:E5:4E:02:00:00
ethprime=FEC0
fdt_addr_r=0x41000000
fdt_mmcload=mmc rescan ; fatload mmc 0:2 ${fdt_addr_r} ${fdtfile}
fdt_netload=tftp ${fdt_addr_r} ${fdtfile}
fdtfile=duts/m28/imx28-m28evk.dtb
fileaddr=43000000
filesize=216EA0
gatewayip=192.168.1.254
hostname=m28
ipaddr=192.168.20.33
kernel_addr_r=0x42000000
kernel_mmcload=mmc rescan ; fatload mmc 0:2 ${kernel_addr_r} uImage
kernel_nandload=nand read ${kernel_addr_r} 0 ${kernel_size}
kernel_netload=tftp ${kernel_addr_r} ${bootfile}
kernel_size=280000
load=echo $update_sd_firmware ; run update_sd_firmware
loadaddr=0x42000000
miscargs=fec_mac=${ethaddr} earlyprintk user_debug=29
mmc_mmc=run fdt_mmcload kernel_mmcload mmcargs addip addcons addmtd addmisc;bootm ${kernel_addr_r} - ${fdt_addr_r}
mmcargs=setenv bootargs root=${rootdev} rw rootwait
mtddevname=bootloader
mtddevnum=0
mtdids=nand0=gpmi-nand
mtdparts=mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel),-(filesystem)
nand_erasesize=20000
nand_mmc=run kernel_nandload mmcargs addip addcons addmtd addmisc;bootm ${kernel_addr_r} - ${fdt_addr_r}
nand_nfs=run kernel_nandload nfsargs addip addcons addmtd addmisc;bootm ${kernel_addr_r} - ${fdt_addr_r}
nand_oobsize=40
nand_writesize=800
net_mmc=run fdt_netload kernel_netload mmcargs addip addcons addmtd addmisc;bootm ${kernel_addr_r} - ${fdt_addr_r}
net_nfs=run fdt_netload kernel_netload nfsargs addip addcons addmtd addmisc;bootm ${kernel_addr_r} - ${fdt_addr_r}
net_nfs_nodt=run kernel_netload nfsargs addip addcons addmtd addmisc;bootm ${kernel_addr_r}
netdev=eth0
netmask=255.255.0.0
nfsargs=setenv bootargs root=/dev/nfs rw nfsroot=${serverip}:${rootpath},v3,tcp
partition=nand0,0
rootdev=/dev/mmcblk0p3
rootpath=/opt/eldk-5.2.1/armv5te/rootfs-qte-sdk
serverip=192.168.1.1
stderr=serial
stdin=serial
stdout=serial
update=echo done
update_nand_count=0x4
update_nand_firmware=if tftp ${update_nand_firmware_filename} ; then run update_nand_get_fcb_size ; setexpr fcb_sz ${update_nand_fcb} * 2 ; setexpr fw_sz ${update_nand_firmware_maxsz} * 2 ; setexpr fw_off ${fcb_sz} + ${update_nand_firmware_maxsz};nand erase ${fcb_sz} ${fw_sz} ; nand write ${loadaddr} ${fcb_sz} ${filesize} ; nand write ${loadaddr} ${fw_off} ${filesize} ; fi
update_nand_firmware_filename=u-boot.sb
update_nand_firmware_maxsz=0x100000
update_nand_full=if tftp ${update_nand_full_filename} ; then run update_nand_get_fcb_size ; nand scrub -y 0x0 ${filesize} ; nand write.raw ${loadaddr} 0x0 ${update_nand_fcb} ; setexpr update_off ${loadaddr} + ${update_nand_fcb} ; setexpr update_sz ${filesize} - ${update_nand_fcb} ; nand write ${update_off} ${update_nand_fcb} ${update_sz} ; fi
update_nand_full_filename=u-boot.nand
update_nand_get_fcb_size=nand device 0 ; nand info ; setexpr fcb_sz ${update_nand_stride} * ${update_nand_count};setexpr update_nand_fcb ${fcb_sz} * ${nand_writesize}
update_nand_stride=0x40
update_sd_firmware=if mmc rescan ; then if tftp ${update_sd_firmware_filename} ; then setexpr fw_sz ${filesize} / 0x200 ; setexpr fw_sz ${fw_sz} + 1 ; mmc write ${loadaddr} 0x800 ${fw_sz} ; fi ; fi
update_sd_firmware_filename=duts/m28/u-boot.mx28.sd
ver=U-Boot 2012.07-00471-ge8925d7-dirty (Oct 01 2012 - 18:20:02)
videomode=video=ctfb:x:800,y:480,depth:18,mode:0,pclk:30066,le:0,ri:256,up:0,lo:45,hs:1,vs:1,sync:100663296,vmode:0

Environment size: 4044/16379 bytes
=>

5.9.6.2. saveenv - save environment variables to persistent storage

=> help saveenv
saveenv - save environment variables to persistent storage

Usage:
saveenv
=>

All changes you make to the U-Boot environment are made in RAM only. They are lost as soon as you reboot the system. If you want to make your changes permanent you have to use the saveenv command to write a copy of the environment settings to persistent storage, from where they are automatically loaded during startup:

=> saveenv
Saving Environment to NAND...
Erasing NAND...
Erasing at 0x300000 --  25% complete.Erasing at 0x320000 --  50% complete.Erasing at 0x340000 --  75% complete.Erasing at 0x360000 -- 100% complete.
Writing to NAND... done
=>

5.9.6.3. setenv - set environment variables

=> help setenv
setenv - set environment variables

Usage:
setenv name value ...
    - set environment variable 'name' to 'value ...'
setenv name
    - delete environment variable 'name'
=>

To modify the U-Boot environment you have to use the setenv command. When called with exactly one argument, it will delete any variable of that name from U-Boot's environment, if such a variable exists. Any storage occupied for such a variable will be automatically reclaimed:

=> setenv foo This is an example value.
=> printenv foo
foo=This is an example value.
=> setenv foo
=> printenv foo
## Error: "foo" not defined
=>

When called with more arguments, the first one will again be the name of the variable, and all following arguments will (concatenated by single space characters) form the value that gets stored for this variable. New variables will be automatically created, existing ones overwritten.

=> printenv bar
## Error: "bar" not defined
=> setenv bar This is a new example.
=> printenv bar
bar=This is a new example.
=>

Remember standard shell quoting rules when the value of a variable shall contain characters that have a special meaning to the command line parser (like the $ character that is used for variable substitution or the semicolon which separates commands). Use the backslash (\) character to escape such special characters, or enclose the whole phrase in apstrophes ('). Use "${name}" for variable expansion (see 14.2.17. How the Command Line Parsing Works for details).

=> setenv cons_opts 'console=tty0 console=ttyS0,${baudrate}'
=> printenv cons_opts
cons_opts=console=tty0 console=ttyS0,${baudrate}
=>

There is no restriction on the characters that can be used in a variable name except the restrictions imposed by the command line parser (like using backslash for quoting, space and tab characters to separate arguments, or semicolon and newline to separate commands). Even strange input like "=-/|()+=" is a perfectly legal variable name in U-Boot.

A common mistake is to write

setenv name=value

instead of

setenv name value

There will be no error message, which lets you believe everything went OK, but it didn't: instead of setting the variable name to the value value you tried to delete a variable with the name name=value - this is probably not what you intended! Always remember that name and value have to be separated by space and/or tab characters!

5.9.6.4. run - run commands in an environment variable

=> help run
run - run commands in an environment variable

Usage:
run var [...]
    - run the commands in the environment variable(s) 'var'
=>

You can use U-Boot environment variables to store commands and even sequences of commands. To execute such a command, you use the run command:

=>  setenv test echo This is a test\;printenv ipaddr\;echo Done.
=> printenv test
test=echo This is a test;printenv ipaddr;echo Done.
=> run test
This is a test
ipaddr=192.168.20.33
Done.
=>

You can call run with several variables as arguments, in which case these commands will be executed in sequence:

=>  setenv test2 echo This is another Test\;printenv hostname\;echo Done.
=> printenv test test2
test=echo This is a test;printenv ipaddr;echo Done.
test2=echo This is another Test;printenv hostname;echo Done.
=> run test test2
This is a test
ipaddr=192.168.20.33
Done.
This is another Test
hostname=m28
Done.
=>

If a U-Boot variable contains several commands (separated by semicolon), and one of these commands fails when you "run" this variable, the remaining commands will be executed anyway.

If you execute several variables with one call to run, any failing command will cause "run" to terminate, i. e. the remaining variables are not executed.

5.9.6.5. bootd - boot default, i.e., run 'bootcmd'

=> help bootd
bootd - boot default, i.e., run 'bootcmd'

Usage:
bootd
=>

The bootd (short: boot) executes the default boot command, i. e. what happens when you don't interrupt the initial countdown. This is a synonym for the run bootcmd command.

5.9.7. Flattened Device Tree support

U-Boot is capable of quite comprehensive handling of the flattened device tree blob, implemented by the fdt family of commands:

=> help fdt
fdt - flattened device tree utility commands

Usage:
fdt addr   <addr> [<length>]        - Set the fdt location to <addr>
fdt move   <fdt> <newaddr> <length> - Copy the fdt to <addr> and make it active
fdt resize                          - Resize fdt to size + padding to 4k addr
fdt print  <path> [<prop>]          - Recursive print starting at <path>
fdt list   <path> [<prop>]          - Print one level starting at <path>
fdt set    <path> <prop> [<val>]    - Set <property> [to <val>]
fdt mknode <path> <node>            - Create a new node after <path>
fdt rm     <path> [<prop>]          - Delete the node or <property>
fdt header                          - Display header info
fdt bootcpu <id>                    - Set boot cpuid
fdt memory <addr> <size>            - Add/Update memory node
fdt rsvmem print                    - Show current mem reserves
fdt rsvmem add <addr> <size>        - Add a mem reserve
fdt rsvmem delete <index>           - Delete a mem reserves
fdt chosen [<start> <end>]          - Add/update the /chosen branch in the tree
                                        <start>/<end> - initrd start/end addr
NOTE: Dereference aliases by omiting the leading '/', e.g. fdt print ethernet0.
=>

5.9.7.1. fdt addr - select FDT to work on

First, the blob that is to be operated on should be stored in memory, and U-Boot has to be informed about its location by the fdt addr command. Once this command has been issued, all subsequent fdt handling commands will use the blob stored at the given address. This address can be changed later on by issuing fdt addr or fdt move command. Here's how to load the blob into memory and tell U-Boot its location:

=> print fdt_addr_r
fdt_addr_r=0x41000000
=> print fdt_file
fdt_file=/tftpboot/duts/m28/imx28-m28evk.dtb
=> tftp ${fdt_addr_r} ${fdt_file}
Using FEC0 device
TFTP from server 192.168.1.1; our IP address is 192.168.20.33
Filename '/tftpboot/duts/m28/imx28-m28evk.dtb'.
Load address: 0x41000000
Loading: ##
done
Bytes transferred = 16547 (40a3 hex)
=> fdt addr ${fdt_addr_r}
=>

5.9.7.2. fdt list - print one level

Having selected the device tree stored in the blob just loaded, we can inspect its contents. As an FDT usually is quite extensive, it is easier to get information about the structure by looking at selected levels rather than full hierarchies. fdt list allows us to do exactly this. Let's have a look at the hierarchy one level below the cpus node:

=> fdt list /cpus
cpus {
        cpu@0 {
        };
};
=>

5.9.7.3. fdt print - recursive print

To print a complete subtree we use fdt print. In comparison to the previous example it is obvious that the whole subtree is printed:

=> fdt print /cpus
cpus {
        cpu@0 {
                compatible = "arm,arm926ejs";
        };
};
=>

5.9.7.4. fdt mknode - create new nodes

fdt mknode can be used to attach a new node to the tree. We will use the fdt list command to verify that the new node has been created and that it is empty:

=> fdt list /
/ {
        #address-cells = <0x1>;
        #size-cells = <0x1>;
        interrupt-parent = <0x1>;
        model = "DENX M28EVK";
        compatible = "denx,m28evk", "fsl,imx28";
        chosen {
        };
        aliases {
        };
        memory {
        };
        cpus {
        };
        apb@80000000 {
        };
        ahb@80080000 {
        };
        regulators {
        };
        sound {
        };
};
=> fdt mknode / testnode
=> fdt list /
/ {
        #address-cells = <0x1>;
        #size-cells = <0x1>;
        interrupt-parent = <0x1>;
        model = "DENX M28EVK";
        compatible = "denx,m28evk", "fsl,imx28";
        testnode {
        };
        chosen {
        };
        aliases {
        };
        memory {
        };
        cpus {
        };
        apb@80000000 {
        };
        ahb@80080000 {
        };
        regulators {
        };
        sound {
        };
};
=> fdt list /testnode
testnode {
};
=>

5.9.7.5. fdt set - set node properties

Now, let's create a property at the newly created node; again we'll use fdt list for verification:

=> fdt set /testnode testprop testvalue
=> fdt list /testnode
testnode {
        testprop = "testvalue";
};
=>

5.9.7.6. fdt rm - remove nodes or properties

The fdt rm command is used to remove nodes and properties. Let's delete the test property created in the previous paragraph and verify the results:

=> fdt rm /testnode testprop
=> fdt list /testnode
testnode {
};
=> fdt rm /testnode
=> fdt list /
/ {
        #address-cells = <0x1>;
        #size-cells = <0x1>;
        interrupt-parent = <0x1>;
        model = "DENX M28EVK";
        compatible = "denx,m28evk", "fsl,imx28";
        chosen {
        };
        aliases {
        };
        memory {
        };
        cpus {
        };
        apb@80000000 {
        };
        ahb@80080000 {
        };
        regulators {
        };
        sound {
        };
};
=>

5.9.7.7. fdt move - move FDT blob to new address

To move the blob from one memory location to another we will use the fdt move command. Besides moving the blob, it makes the new address the "active" one - similar to fdt addr:

=> fdt move ${fdt_addr_r} 0x42000000
=> fdt list /
/ {
        #address-cells = <0x1>;
        #size-cells = <0x1>;
        interrupt-parent = <0x1>;
        model = "DENX M28EVK";
        compatible = "denx,m28evk", "fsl,imx28";
        chosen {
        };
        aliases {
        };
        memory {
        };
        cpus {
        };
        apb@80000000 {
        };
        ahb@80080000 {
        };
        regulators {
        };
        sound {
        };
};
=> fdt mknod / foobar
=> fdt list /
/ {
        #address-cells = <0x1>;
        #size-cells = <0x1>;
        interrupt-parent = <0x1>;
        model = "DENX M28EVK";
        compatible = "denx,m28evk", "fsl,imx28";
        foobar {
        };
        chosen {
        };
        aliases {
        };
        memory {
        };
        cpus {
        };
        apb@80000000 {
        };
        ahb@80080000 {
        };
        regulators {
        };
        sound {
        };
};
=> fdt addr ${fdt_addr_r}
=> fdt list /
/ {
        #address-cells = <0x1>;
        #size-cells = <0x1>;
        interrupt-parent = <0x1>;
        model = "DENX M28EVK";
        compatible = "denx,m28evk", "fsl,imx28";
        chosen {
        };
        aliases {
        };
        memory {
        };
        cpus {
        };
        apb@80000000 {
        };
        ahb@80080000 {
        };
        regulators {
        };
        sound {
        };
};
=>

5.9.7.8. fdt chosen - fixup dynamic info

One of the modifications made by U-Boot to the blob before passing it to the kernel is the addition of the /chosen node. Linux 2.6 Documentation/powerpc/booting-without-of.txt says that this node is used to store "some variable environment information, like the arguments, or the default input/output devices." To force U-Boot to add the /chosen node to the current blob, fdt chosen command can be used. Let's now verify its operation:

=> fdt list /
/ {
        #address-cells = <0x1>;
        #size-cells = <0x1>;
        interrupt-parent = <0x1>;
        model = "DENX M28EVK";
        compatible = "denx,m28evk", "fsl,imx28";
        chosen {
        };
        aliases {
        };
        memory {
        };
        cpus {
        };
        apb@80000000 {
        };
        ahb@80080000 {
        };
        regulators {
        };
        sound {
        };
};
=> fdt chosen
=> fdt list /
/ {
        #address-cells = <0x1>;
        #size-cells = <0x1>;
        interrupt-parent = <0x1>;
        model = "DENX M28EVK";
        compatible = "denx,m28evk", "fsl,imx28";
        chosen {
        };
        aliases {
        };
        memory {
        };
        cpus {
        };
        apb@80000000 {
        };
        ahb@80080000 {
        };
        regulators {
        };
        sound {
        };
};
=> fdt list /chosen
chosen {
};
=>

Note: fdt boardsetup performs board-specific blob updates, most commonly setting clock frequencies, etc. Discovering its operation is left as an excercise for the reader.

5.9.8. Special Commands

5.9.8.1. i2c - I2C sub-system

=> help i2c
i2c - I2C sub-system

Usage:
i2c crc32 chip address[.0, .1, .2] count - compute CRC32 checksum
i2c loop chip address[.0, .1, .2] [# of objects] - looping read of device
i2c md chip address[.0, .1, .2] [# of objects] - read from I2C device
i2c mm chip address[.0, .1, .2] - write to I2C device (auto-incrementing)
i2c mw chip address[.0, .1, .2] value [count] - write to I2C device (fill)
i2c nm chip address[.0, .1, .2] - write to I2C device (constant address)
i2c probe - show devices on the I2C bus
i2c read chip address[.0, .1, .2] length memaddress - read to memory
i2c reset - re-init the I2C Controller
i2c speed [speed] - show or set I2C bus speed
=>

5.9.9. Storage devices

This chapter introduces commands to work with storage devices, i.e. ATA, CF, SATA, SCSI, USB, NAND, etc. connected to the board.

5.9.9.1. MMC devices

=> 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
=>

The mmc dev command displays the current device

=> mmc dev
mmc0 is current device
=>

The mmc list command displays the available mmc devices

=> mmc list
MXS MMC: 0
=>

With the mmc rescan command you can rescan the actual mmc device.

=> mmc rescan
=>

The mmcinfo command displays the information about the actual mmc device

=> mmcinfo
Device: MXS MMC
Manufacturer ID: 1b
OEM: 534d
Name: 00000
Tran Speed: 50000000
Rd Block Len: 512
SD version 2.0
High Capacity: No
Capacity: 1.9 GiB
Bus Width: 4-bit
=>

The mmc part command displays the available partitions on the actual mmc device.

=> mmc part

Partition Map for MMC device 0  --   Partition Type: DOS

Partition     Start Sector     Num Sectors     Type
    1                 2048            2048      53
    2                 4096           61440       6
    3                65536         3620864      83
=>

You can read data from the mmc with the mmc read command:

=> mmc read 0x42000000 247 10

MMC read: dev # 0, block # 583, count 16 ... 16 blocks read: OK
=> md 0x42000000
42000000: 00000000 00000000 00000000 00000000    ................
42000010: 00000000 00000000 00000000 00000000    ................
42000020: 00000000 00000000 00000000 00000000    ................
42000030: 00000000 00000000 00000000 00000000    ................
42000040: 00000000 00000000 00000000 00000000    ................
42000050: 00000000 00000000 00000000 00000000    ................
42000060: 00000000 00000000 00000000 00000000    ................
42000070: 00000000 00000000 00000000 00000000    ................
42000080: 00000000 00000000 00000000 00000000    ................
42000090: 00000000 00000000 00000000 00000000    ................
420000a0: 00000000 00000000 00000000 00000000    ................
420000b0: 00000000 00000000 00000000 00000000    ................
420000c0: 00000000 00000000 00000000 00000000    ................
420000d0: 00000000 00000000 00000000 00000000    ................
420000e0: 00000000 00000000 00000000 00000000    ................
420000f0: 00000000 00000000 00000000 00000000    ................
=>

You can erase/write data from the mmc with the mmc erase / mmc write commands:

=> mmc read 0x43000000 247 10

MMC read: dev # 0, block # 583, count 16 ... 16 blocks read: OK
=> md 0x43000000
43000000: 00000000 00000000 00000000 00000000    ................
43000010: 00000000 00000000 00000000 00000000    ................
43000020: 00000000 00000000 00000000 00000000    ................
43000030: 00000000 00000000 00000000 00000000    ................
43000040: 00000000 00000000 00000000 00000000    ................
43000050: 00000000 00000000 00000000 00000000    ................
43000060: 00000000 00000000 00000000 00000000    ................
43000070: 00000000 00000000 00000000 00000000    ................
43000080: 00000000 00000000 00000000 00000000    ................
43000090: 00000000 00000000 00000000 00000000    ................
430000a0: 00000000 00000000 00000000 00000000    ................
430000b0: 00000000 00000000 00000000 00000000    ................
430000c0: 00000000 00000000 00000000 00000000    ................
430000d0: 00000000 00000000 00000000 00000000    ................
430000e0: 00000000 00000000 00000000 00000000    ................
430000f0: 00000000 00000000 00000000 00000000    ................
=> mw 0x42000000 0xdeadface 1000
=> md 0x42000000
42000000: deadface deadface deadface deadface    ................
42000010: deadface deadface deadface deadface    ................
42000020: deadface deadface deadface deadface    ................
42000030: deadface deadface deadface deadface    ................
42000040: deadface deadface deadface deadface    ................
42000050: deadface deadface deadface deadface    ................
42000060: deadface deadface deadface deadface    ................
42000070: deadface deadface deadface deadface    ................
42000080: deadface deadface deadface deadface    ................
42000090: deadface deadface deadface deadface    ................
420000a0: deadface deadface deadface deadface    ................
420000b0: deadface deadface deadface deadface    ................
420000c0: deadface deadface deadface deadface    ................
420000d0: deadface deadface deadface deadface    ................
420000e0: deadface deadface deadface deadface    ................
420000f0: deadface deadface deadface deadface    ................
=> mmc erase 247 10

MMC erase: dev # 0, block # 583, count 16 ... 16 blocks erase: OK
=> mmc write 0x42000000  247 10

MMC write: dev # 0, block # 583, count 16 ... 16 blocks write: OK
=> mw 0x42000000 0x0 1000
=> mmc read 0x42000000 247 10

MMC read: dev # 0, block # 583, count 16 ... 16 blocks read: OK
=> md 0x42000000
42000000: deadface deadface deadface deadface    ................
42000010: deadface deadface deadface deadface    ................
42000020: deadface deadface deadface deadface    ................
42000030: deadface deadface deadface deadface    ................
42000040: deadface deadface deadface deadface    ................
42000050: deadface deadface deadface deadface    ................
42000060: deadface deadface deadface deadface    ................
42000070: deadface deadface deadface deadface    ................
42000080: deadface deadface deadface deadface    ................
42000090: deadface deadface deadface deadface    ................
420000a0: deadface deadface deadface deadface    ................
420000b0: deadface deadface deadface deadface    ................
420000c0: deadface deadface deadface deadface    ................
420000d0: deadface deadface deadface deadface    ................
420000e0: deadface deadface deadface deadface    ................
420000f0: deadface deadface deadface deadface    ................
=> mmc erase 247 10

MMC erase: dev # 0, block # 583, count 16 ... 16 blocks erase: OK
=> mmc write 0x43000000  247 10

MMC write: dev # 0, block # 583, count 16 ... 16 blocks write: OK
=> mmc read 0x42000000 247 10

MMC read: dev # 0, block # 583, count 16 ... 16 blocks read: OK
=> md 0x42000000
42000000: 00000000 00000000 00000000 00000000    ................
42000010: 00000000 00000000 00000000 00000000    ................
42000020: 00000000 00000000 00000000 00000000    ................
42000030: 00000000 00000000 00000000 00000000    ................
42000040: 00000000 00000000 00000000 00000000    ................
42000050: 00000000 00000000 00000000 00000000    ................
42000060: 00000000 00000000 00000000 00000000    ................
42000070: 00000000 00000000 00000000 00000000    ................
42000080: 00000000 00000000 00000000 00000000    ................
42000090: 00000000 00000000 00000000 00000000    ................
420000a0: 00000000 00000000 00000000 00000000    ................
420000b0: 00000000 00000000 00000000 00000000    ................
420000c0: 00000000 00000000 00000000 00000000    ................
420000d0: 00000000 00000000 00000000 00000000    ................
420000e0: 00000000 00000000 00000000 00000000    ................
420000f0: 00000000 00000000 00000000 00000000    ................
=>

if you have a dos partition on the device, you can list the content with:

=> fatls mmc 0:2 /
  2340744   uimage
     5410   env.txt
  1152054   img1.bmp
  1152054   img.bmp
   233738   slide-1.bmp
  1152054   slide-2.bmp
  1152054   slide-3.bmp
  1152054   slide-4.bmp
  1152054   slide-5.bmp
  1152054   slide-6.bmp
  1152054   slide-7.bmp
  1152054   slide-8.bmp
     3754   env.txt.old
       29   date_of_creation
       29   date_of_modification
     1024   random.hex

16 file(s), 0 dir(s)

=>

and load a file in RAM with:

=> fatload mmc 0:2 0x42000000 date_of_creation
reading date_of_creation

29 bytes read
=> md 0x42000000
42000000: 20657554 2074634f 31203220 34343a35    Tue Oct  2 15:44
42000010: 2035353a 20435455 32313032 0000000a    :55 UTC 2012....
42000020: 00000000 00000000 00000000 00000000    ................
42000030: 00000000 00000000 00000000 00000000    ................
42000040: 00000000 00000000 00000000 00000000    ................
42000050: 00000000 00000000 00000000 00000000    ................
42000060: 00000000 00000000 00000000 00000000    ................
42000070: 00000000 00000000 00000000 00000000    ................
42000080: 00000000 00000000 00000000 00000000    ................
42000090: 00000000 00000000 00000000 00000000    ................
420000a0: 00000000 00000000 00000000 00000000    ................
420000b0: 00000000 00000000 00000000 00000000    ................
420000c0: 00000000 00000000 00000000 00000000    ................
420000d0: 00000000 00000000 00000000 00000000    ................
420000e0: 00000000 00000000 00000000 00000000    ................
420000f0: 00000000 00000000 00000000 00000000    ................
=>

5.9.9.2. NAND devices

U-Boot allows us to directly work with NAND devices attached directly or through a NAND controller. The commands are grouped under the nand subsystem:

=> help nand
nand - NAND sub-system

Usage:
nand info - show available NAND devices
nand device [dev] - show or set current device
nand read - addr off|partition size
nand write - addr off|partition size
    read/write 'size' bytes starting at offset 'off'
    to/from memory address 'addr', skipping bad blocks.
nand read.raw - addr off|partition [count]
nand write.raw - addr off|partition [count]
    Use read.raw/write.raw to avoid ECC and access the flash as-is.
nand erase[.spread] [clean] off size - erase 'size' bytes from offset 'off'
    With '.spread', erase enough for given file size, otherwise,
    'size' includes skipped bad blocks.
nand erase.part [clean] partition - erase entire mtd partition'
nand erase.chip [clean] - erase entire chip'
nand bad - show bad blocks
nand dump[.oob] off - dump page
nand scrub [-y] off size | scrub.part partition | scrub.chip
    really clean NAND erasing bad blocks (UNSAFE)
nand markbad off [...] - mark bad block(s) at offset (UNSAFE)
nand biterr off - make a bit error at offset (UNSAFE)
=>

5.9.9.2.1. nand bad - show bad block information

As NAND devices can develop bad blocks over their lifetime and usually even are delivered with bad blocks already, the NAND commands need to be aware of this fact. Getting information of the current bad block list is easy:

=> nand bad

Device 0 bad blocks:
=>

5.9.9.2.2. nand erase - erase region

Ensuring that the NAND device functions properly, we will use the basic commands to construct a testpattern in memory, write that to the device, of course erasing it first, and read the data back. A final comparison will show if all data was transferred correctly.

We begin be erasing 64k at the start of the NAND device:

=> nand erase 0x00400000 0x10000

NAND erase: device 0 offset 0x400000, size 0x10000
Erasing at 0x400000 -- 100% complete.
OK
=>

5.9.9.2.3. nand write - write to NAND device

Let's create a testpattern in memory and write that to the previously erased NAND area:

=> mw 0x42000000 0x55aa55aa 0x4000
=> nand write 0x42000000 0x00400000 0x10000

NAND write: device 0 offset 0x400000, size 0x10000
 65536 bytes written: OK
=>

5.9.9.2.4. nand read - read from NAND device

As everything worked ok, we ensure everything was fine by transferring the data to a different location in RAM and check this against the original written content:

=> nand read 0x43000000 0x00400000 0x10000

NAND read: device 0 offset 0x400000, size 0x10000
 65536 bytes read: OK
=> md 0x43000000
43000000: 55aa55aa 55aa55aa 55aa55aa 55aa55aa    .U.U.U.U.U.U.U.U
43000010: 55aa55aa 55aa55aa 55aa55aa 55aa55aa    .U.U.U.U.U.U.U.U
=> cmp 0x42000000 0x43000000 0x4000
Total of 16384 word(s) were the same
=>

5.9.10. Miscellaneous Commands

5.9.10.1. date - get/set/reset date & time

=> help date
date - get/set/reset date & time

Usage:
date [MMDDhhmm[[CC]YY][.ss]]
date reset
  - without arguments: print date & time
  - with numeric argument: set the system date & time
  - with 'reset' argument: reset the RTC
=>

The date command is used to display the current time in a standard format, or to set the system date. On some systems it can also be used to reset (initialize) the system clock:

=> date reset
Reset RTC...
Date: 1970-01-01 (Thursday)    Time:  0:00:00
=> date 100216462012
Date: 2012-10-02 (Tuesday)    Time: 16:46:00
=> date
Date: 2012-10-02 (Tuesday)    Time: 16:46:00
=>

5.9.10.2. echo - echo args to console

=> help echo
echo - echo args to console

Usage:
echo [args..]
    - echo args to console; \c suppresses newline
=>

The echo command echoes the arguments to the console:

=> echo The quick brown fox jumped over the lazy dog.
The quick brown fox jumped over the lazy dog.
=>

5.9.10.3. reset - Perform RESET of the CPU

=> help reset
reset - Perform RESET of the CPU

Usage:
reset
=>

The reset command reboots the system.

=>
=> reset
resetting ...


U-Boot 2012.07-00471-ge8925d7-dirty (Oct 01 2012 - 18:20:02)

CPU:   Freescale i.MX28 rev1.2 at 454 MHz
BOOT:  SSP SD/MMC #0, 3V3
I2C:   ready
DRAM:  256 MiB
NAND:  256 MiB
MMC:   MXS MMC: 0
In:    serial
Out:   serial
Err:   serial
Net:   FEC0 [PRIME], FEC1
Hit any key to stop autoboot:  0
=>

5.9.10.4. sleep - delay execution for some time

=> help sleep
sleep - delay execution for some time

Usage:
sleep N
    - delay execution for N seconds (N is _decimal_ !!!)
=>

The sleep command pauses execution for the number of seconds given as the argument:

=> sleep 5
=>

5.9.10.5. version - print monitor version

=> help version
version - print monitor, compiler and linker version

Usage:
version
=>

You can print the version and build date of the U-Boot image running on your system using the version command (short: vers):

=> version

U-Boot 2012.07-00471-ge8925d7-dirty (Oct 01 2012 - 18:20:02)
arm-linux-gnueabi-gcc (Debian 4.7.2-2) 4.7.2
GNU ld (GNU Binutils for Debian) 2.22
=>

5.9.10.6. ? - alias for 'help'

You can use ? as a short form for the help command (see description above).

5.10. U-Boot Environment Variables

The U-Boot environment is a block of memory that is kept on persistent storage and copied to RAM when U-Boot starts. It is used to store environment variables which can be used to configure the system. The environment is protected by a CRC32 checksum.

This section lists the most important environment variables, some of which have a special meaning to U-Boot. You can use these variables to configure the behaviour of U-Boot to your liking.

• autoload: if set to "no" (or any string beginning with 'n'), the rarpb, bootp or dhcp commands will perform only a configuration lookup from the BOOTP / DHCP server, but not try to load any image using TFTP.

• autostart: if set to "yes", an image loaded using the rarpb, bootp, dhcp, tftp, disk, or docb commands will be automatically started (by internally calling the bootm command).

• baudrate: a decimal number that selects the console baudrate (in bps). Only a predefined list of baudrate settings is available.
  When you change the baudrate (using the "setenv baudrate ..." command), U-Boot will switch the baudrate of the console terminal and wait for a newline which must be entered with the new speed setting. This is to make sure you can actually type at the new speed. If this fails, you have to reset the board (which will operate at the old speed since you were not able to saveenv the new settings.)
  If no "baudrate" variable is defined, the default baudrate of 115200 is used.

• bootargs: The contents of this variable are passed to the Linux kernel as boot arguments (aka "command line").

• bootcmd: This variable defines a command string that is automatically executed when the initial countdown is not interrupted.
  This command is only executed when the variable bootdelay is also defined!

• bootdelay: After reset, U-Boot will wait this number of seconds before it executes the contents of the bootcmd variable. During this time a countdown is printed, which can be interrupted by pressing any key.
  Set this variable to 0 boot without delay. Be careful: depending on the contents of your bootcmd variable, this can prevent you from entering interactive commands again forever!
  Set this variable to -1 to disable autoboot.

• bootfile: name of the default image to load with TFTP

• cpuclk: (Only with MPC859 / MPC866 / MPC885 processors) On some processors, the CPU clock frequency can be adjusted by the user (for example to optimize performance versus power dissipation). On such systems the cpuclk variable can be set to the desired CPU clock value, in MHz. If the cpuclk variable exists and its value is within the compile-time defined limits (CFG_866_CPUCLK_MIN and CFG_866_CPUCLK_MAX = minimum resp. maximum allowed CPU clock), then the specified value is used. Otherwise, the default CPU clock value is set.

• ethaddr: Ethernet MAC address for first/only ethernet interface (= eth0 in Linux).
  This variable can be set only once (usually during manufacturing of the board). U-Boot refuses to delete or overwrite this variable once it has been set.

• eth1addr: Ethernet MAC address for second ethernet interface (= eth1 in Linux).

• eth2addr: Ethernet MAC address for third ethernet interface (= eth2 in Linux).
  ...

• initrd_high: used to restrict positioning of initrd ramdisk images:
  If this variable is not set, initrd images will be copied to the highest possible address in RAM; this is usually what you want since it allows for maximum initrd size. If for some reason you want to make sure that the initrd image is loaded below the CFG_BOOTMAPSZ limit, you can set this environment variable to a value of "no" or "off" or "0". Alternatively, you can set it to a maximum upper address to use (U-Boot will still check that it does not overwrite the U-Boot stack and data).
  For instance, when you have a system with 16 MB RAM, and want to reserve 4 MB from use by Linux, you can do this by adding "mem=12M" to the value of the "bootargs" variable. However, now you must make sure that the initrd image is placed in the first 12 MB as well - this can be done with
  

=> setenv initrd_high 00c00000
     

• ipaddr: IP address; needed for tftp command

• loadaddr: Default load address for commands like tftp or loads.

• loads_echo: If set to 1, all characters received during a serial download (using the loads command) are echoed back. This might be needed by some terminal emulations (like cu), but may as well just take time on others.

• mtdparts: This variable (usually defined using the mtdparts command) allows to share a common MTD partition scheme between U-Boot and the Linux kernel.

• pram: If the "Protected RAM" feature is enabled in your board's configuration, this variable can be defined to enable the reservation of such "protected RAM", i. e. RAM which is not overwritten by U-Boot. Define this variable to hold the number of kB you want to reserve for pRAM. Note that the board info structure will still show the full amount of RAM. If pRAM is reserved, a new environment variable "mem" will automatically be defined to hold the amount of remaining RAM in a form that can be passed as boot argument to Linux, for instance like that:

=> setenv bootargs ${bootargs} mem=\${mem}
=> saveenv
     

• serverip: TFTP server IP address; needed for tftp command.

• serial#: contains hardware identification information such as type string and/or serial number.
  This variable can be set only once (usually during manufacturing of the board). U-Boot refuses to delete or overwrite this variable once it hass been set.

• silent: If the configuration option CONFIG_SILENT_CONSOLE has been enabled for your board, setting this variable to any value will suppress all console messages. Please see doc/README.silent for details.

• verify: If set to n or no disables the checksum calculation over the complete image in the bootm command to trade speed for safety in the boot process. Note that the header checksum is still verified.

The following environment variables may be used and automatically updated by the network boot commands (bootp, dhcp, or tftp), depending the information provided by your boot server:

• bootfile: see above
• dnsip: IP address of your Domain Name Server
• gatewayip: IP address of the Gateway (Router) to use
• hostname: Target hostname
• ipaddr: see above
• netmask: Subnet Mask
• rootpath: Pathname of the root filesystem on the NFS server
• serverip: see above
• filesize: Size (as hex number in bytes) of the file downloaded using the last bootp, dhcp, or tftp command.

5.11. U-Boot Scripting Capabilities

For example, assume that you will have to run the following sequence of commands on many boards, so you store them in a text file, say "setenv-commands":

bash$ cat setenv-commands
setenv loadaddr 00200000
echo ===== U-Boot settings =====
setenv u-boot /tftpboot/TQM860L/u-boot.bin
setenv u-boot_addr 40000000
setenv load_u-boot 'tftp ${loadaddr} ${u-boot}'
setenv install_u-boot 'protect off ${u-boot_addr} +${filesize};era ${u-boot_addr} +${filesize};cp.b ${loadaddr} ${u-boot_addr} ${filesize};saveenv'
setenv update_u-boot run load_u-boot install_u-boot
echo ===== Linux Kernel settings =====
setenv bootfile /tftpboot/TQM860L/uImage
setenv kernel_addr 40040000
setenv load_kernel 'tftp ${loadaddr} ${bootfile};'
setenv install_kernel 'era ${kernel_addr} +${filesize};cp.b ${loadaddr} ${kernel_addr} ${filesize}'
setenv update_kernel run load_kernel install_kernel
echo ===== Ramdisk settings =====
setenv ramdisk /tftpboot/TQM860L/uRamdisk
setenv ramdisk_addr 40100000
setenv load_ramdisk 'tftp ${loadaddr} ${ramdisk};'
setenv install_ramdisk 'era ${ramdisk_addr} +${filesize};cp.b ${loadaddr} ${ramdisk_addr} ${filesize}'
setenv update_ramdisk run load_ramdisk install_ramdisk
echo ===== Save new definitions =====
saveenv
bash$ 

bash$ mkimage -T script -C none -n 'Demo Script File' -d setenv-commands setenv.img
Image Name:   Demo Script File
Created:      Mon Jun  6 13:33:14 2005
Image Type:   PowerPC Linux Script (uncompressed)
Data Size:    1147 Bytes = 1.12 kB = 0.00 MB
Load Address: 0x00000000
Entry Point:  0x00000000
Contents:
   Image 0:     1139 Bytes =    1 kB = 0 MB
bash$ 

=> tftp 100000 /tftpboot/TQM860L/setenv.img
Using FEC ETHERNET device
TFTP from server 192.168.3.1; our IP address is 192.168.3.80
Filename '/tftpboot/TQM860L/setenv.img'.
Load address: 0x100000
Loading: #
done
Bytes transferred = 1211 (4bb hex)
=> imi 100000

## Checking Image at 00100000 ...
   Image Name:   Demo Script File
   Created:      2005-06-06  11:33:14 UTC
   Image Type:   PowerPC Linux Script (uncompressed)
   Data Size:    1147 Bytes =  1.1 kB
   Load Address: 00000000
   Entry Point:  00000000
   Verifying Checksum ... OK
=> autoscr 100000
## Executing script at 00100000
===== U-Boot settings =====
===== Linux Kernel settings =====
===== Ramdisk settings =====
===== Save new definitions =====
Saving Environment to Flash...
Un-Protected 1 sectors
Un-Protected 1 sectors
Erasing Flash...
. done
Erased 1 sectors
Writing to Flash... done
Protected 1 sectors
Protected 1 sectors
=> 

5.12. U-Boot Standalone Applications

U-Boot supports "standalone" applications, which are loaded dynamically; these applications can have access to the U-Boot console I/O functions, memory allocation and interrupt services.

A couple of simple examples are included with the U-Boot source code:

5.12.1. "Hello World" Demo

examples/hello_world.c contains a small "Hello World" Demo application; it is automatically compiled when you build U-Boot. It's configured to run at address 0x00040004, so you can play with it like that:

=> loads
## Ready for S-Record download ...
~>examples/hello_world.srec
1 2 3 4 5 6 7 8 9 10 11 ...
[file transfer complete]
[connected]
## Start Addr = 0x00040004

=> go 40004 Hello World! This is a test.
## Starting application at 0x00040004 ...
Hello World
argc = 7
argv[0] = "40004"
argv[1] = "Hello"
argv[2] = "World!"
argv[3] = "This"
argv[4] = "is"
argv[5] = "a"
argv[6] = "test."
argv[7] = ""
Hit any key to exit ...

## Application terminated, rc = 0x0

Alternatively, you can of course use TFTP to download the image over the network. In this case the binary image (hello_world.bin) is used.

Note that the entry point of the program is at offset 0x0004 from the start of file, i. e. the download address and the entry point address differ by four bytes.

=> tftp 40000 /tftpboot/hello_world.bin
...
=> go 40004 This is another test.
## Starting application at 0x00040004 ...
Hello World
argc = 5
argv[0] = "40004"
argv[1] = "This"
argv[2] = "is"
argv[3] = "another"
argv[4] = "test."
argv[5] = ""
Hit any key to exit ...

## Application terminated, rc = 0x0

5.12.2. Timer Demo

This example is only available on MPC8xx CPUs.

5.13. U-Boot Image Formats

U-Boot operates on "image" files which can be basically anything, preceeded by a special header; see the definitions in include/image.h for details; basically, the header defines the following image properties:

• Target Operating System (Provisions for OpenBSD, NetBSD, FreeBSD, 4.4BSD, Linux, SVR4, Esix, Solaris, Irix, SCO, Dell, NCR, LynxOS, pSOS, QNX, RTEMS, U-Boot, ARTOS, Unity OS, Integrity; Currently supported: Linux, NetBSD, VxWorks, QNX, RTEMS, ARTOS, Unity OS, Integrity).

• Target CPU Architecture (Provisions for Alpha, ARM, AVR32, BlackFin, IA64, M68K, Microblaze, MIPS, MIPS64, NIOS, NIOS2, Power Architecture®, IBM S390, SuperH, Sparc, Sparc 64 Bit, Intel x86; Currently supported: ARM, AVR32, BlackFin, M68K, Microblaze, MIPS, MIPS64, NIOS, NIOS2, Power Architecture®, SuperH, Sparc, Sparc 64 Bit, Intel x86).

• Compression Type (Provisions for uncompressed, gzip, bzip2, lzo; Currently supported: uncompressed, gzip, bzip2, lzo).

• Load Address

• Entry Point

• Image Name

• Image Timestamp

5.14. U-Boot Advanced Features

5.14.1. Boot Count Limit

The Open Source Development Labs Carrier Grade Linux Requirements Definition version 2.0 (http://www.osdl.org/docs/carrier_grade_linux_requirements_definition___version_20_final_public_draft.pdf) contains the following requirement definition (ID PLT.4.0, p. 44):

CGL shall provide support for detecting a repeating reboot cycle due to recurring failures and will go to an offline state if this occurs.

This feature is available in U-Boot if you enable the CONFIG_BOOTCOUNT_LIMIT configuration option. The implementation uses the following environment variables:

bootcount:

This variable will be automatically created if it does not exist, and it will be updated at each reset of the processor. After a power-on reset, it will be initialized with 1, and each reboot will increment the value by 1.

bootlimit:

If this variable exists, its contents are taken as the maximum number of reboot cycles allowed.

altbootcmd:

If, after a reboot, the new value of bootcount exceeds the value of bootlimit, then instead of the standard boot action (executing the contents of bootcmd) an alternate boot action will be performed, and the contents of altbootcmd will be executed.

If the variable bootlimit is not defined in the environment, the Boot Count Limit feature is disabled. If it is enabled, but altbootcmd is not defined, then U-Boot will drop into interactive mode and remain there.

It is the responsibility of some application code (typically a Linux application) to reset the variable bootcount, thus allowing for more boot cycles.

At the moment, the Boot Count Limit feature is available only for MPC8xx, MPC82xx and MPC5200 Power Architecture® processors.

• ubootBootcountAccess.c: C-source: bootcount access through /proc file system

6. Embedded Linux Configuration

6.1. Download and Unpack the Linux Kernel Sources

You can download the Linux Kernel Sources from our anonymous git server at http://git.denx.de/. To checkout the module for the first time, proceed as follows:

bash$ cd /opt/eldk/usr/src
bash$ git clone git://git.denx.de/linux-denx.git linux-denx
bash$ cd linux-denx
bash$ git checkout -b duts remotes/origin/m28
Branch duts set up to track remote branch m28 from origin.
Switched to a new branch 'duts'
bash$

6.2. Kernel Configuration and Compilation

The m28 board is fully supported by DENX Software Engineering. This means that you will always be able to build a working default configuration with just minimal interaction.

Please be aware that you will need the "arm" cross development tools for the following steps. Make sure that the directory which contains the binaries of your ELDK are in your PATH.

To be sure that no intermediate results of previous builds are left in your Linux kernel source tree you can clean it up as follows:

bash$ make mrproper

The following command selects a standard configuration for the m28 board that has been extensively tested. It is recommended to use this as a starting point for other, customized configurations:

bash$ make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi- m28_defconfig

[marex@pollux]$

Note: The name of this default configuration file is arch/arm/configs/XXX . By (recursively) listing the contents of the arch/arm/configs/ directory you can easily find out which other default configurations are available.

If you don't want to change the default configuration you can now continue to use it to build a kernel image:

bash$ make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi- uImage

Otherwise you can modify the kernel configuration as follows:

bash$ make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi-  config

or

bash$ make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi-  menuconfig

Note: Because of problems (especially with some older Linux kernel versions) the use of "make xconfig" is not recommended.

bash$ make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi- uImage

The make target uImage uses the tool mkimage (from the U-Boot package) to create a Linux kernel image in arch/arm/boot/uImage which is immediately usable for download and booting with U-Boot.

In case you need a DTB to boot your linux kernel, you need the following step:

bash$ make m28.dtb

In case you configured modules you will also need to compile the modules:

make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi- modules

add install the modules (make sure to pass the correct root path for module installation):

bash$ make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi- INSTALL_MOD_PATH=/opt/eldk-5.2/armv5te/rootfs modules_install

6.3. Installation

For now it is sufficient to copy the Linux kernel image into the directory used by your TFTP server:

bash$ cp arch/arm/boot/uImage /tftpboot/uImage

7. Booting Embedded Linux

7.1. Introduction

In principle, if you have a Linux kernel image and the flattened device tree blob somewhere in system memory (RAM, ROM, flash...), then all you need to boot the system is the bootm command. Assume a Linux kernel image has been stored at address 0x42000000 and the flattened device tree blob has been stored at address 0x41000000 - then you can boot this image with the following command:

=> bootm 42000000 - 41000000

7.2. Flattened Device Tree Blob

Linux kernel expects certain information on the hardware that it runs on. For kernels compiled with fdt support, this information has the form of a device tree, which is based on the Open Firmware specification. Bootloaders like U-Boot that do not implement the Open Firmware API, are expected to pass to the kernel a binary form of the flattened device tree, commonly referred to as FDT blob or simply the blob.

Device trees are defined in human-readable text files, which are part of the Linux 2.6 source tree. Device tree source for the m28 board is found in arch/arm/boot/dts/m28.dts file. Before the device tree can be passed to the kernel, it has to be compiled to the binary form by the dtc compiler. The dtc compiler is included with the Linux kernel since 2.6.25. Since 2.6.26 there is also a simple makefile rule to generate the blob:

make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi- m28.dtb

After the blob has been compiled, it has to be transferred from where it was built ("arch/arm/boot/m28.dtb") to target's memory, for example over the TFTP protocol using U-Boot's tftp command. Then, the blob is passed to the kernel by the bootm command, and its address in memory is one of the arguments to bootm - refer to the description of this command in UBootCmdGroupExec for more details.

Note that U-Boot makes some automatic modifications to the blob before passing it to the kernel - mainly adding and modifying information that is learnt at run-time. See the board-specific function ft_board_setup() and related routines.

U-Boot also has provisions to alter a flattened device tree in arbitrary ways from the command line, refer to the description of the fdt commands found in UBootCmdFDT.

Notes:

• Flattened Device Tree custodian's page at http://www.denx.de/wiki/U-Boot/UBootFdtInfo contains useful information, and a number of references.
• At the time of this writing (September 2007) blob handling is still a very fresh feature and undergoing frequent changes. Reader is encouraged to watch the u-boot-users and linuxppc-dev mailing lists for important news (required version of the dtc compiler, blob compilation options, flattened device tree source file structure, etc.).

7.3. Passing Kernel Arguments

In nearly all cases, you will want to pass additional information to the Linux kernel; for instance, information about the root device or network configuration.

In U-Boot, this is supported using the bootargs environment variable. Its contents are automatically passed to the Linux kernel as boot arguments (or "command line" arguments). This allows the use of the same Linux kernel image in a wide range of configurations. For instance, by just changing the contents of the bootargs variable you can use the very same Linux kernel image to boot with an initrd ramdisk image, with a root filesystem over NFS, with a CompactFlash disk or from a flash filesystem.

As one example, to boot the Linux kernel image at address 0x42000000 using the initrd ramdisk image at address 0x44100000 as root filesystem, and with the flattened device tree blob at address 0x41000000, you can use the following commands:

=> setenv bootargs root=/dev/ram rw
=> bootm 0x42000000 0x44100000  0x41000000 

To boot the same kernel image with a root filesystem over NFS, the following command sequence can be used. This example assumes that your NFS server has the IP address "192.168.1.1" and exports the directory "/opt/eldk-5.2/armv5te/rootfs" as root filesystem for the target. The target has been assigned the IP address "192.168.20.38" and the hostname "m28". A netmask of "255.255.0.0" is used:

=> setenv bootargs root=/dev/nfs rw nfsroot=192.168.1.1:/opt/eldk-5.2/armv5te/rootfs ip=192.168.20.38:192.168.1.1:192.168.1.1:255.255.0.0:m28::off
=> bootm 0x42000000  - 0x41000000 

Please see also the files Documentation/initrd.txt and Documentation/nfsroot.txt in your Linux kernel source directory for more information about which options can be passed to the Linux kernel.

Note: Once your system is up and running, if you have a simple shell login, you can normally examine the boot arguments that were used by the kernel for the most recent boot with the command:

$ cat /proc/cmdline

7.4. Boot Arguments Unleashed

Passing command line arguments to the Linux kernel allows for very flexible and efficient configuration which is especially important in Embedded Systems. It is somewhat strange that these features are nearly undocumented everywhere else. One reason for that is certainly the very limited capabilities of other boot loaders.

It is especially U-Boot's capability to easily define, store, and use environment variables that makes it such a powerful tool in this area. In the examples above we have already seen how we can use for instance the root and ip boot arguments to pass information about the root filesystem or network configuration. The ip argument is not only useful in configurations with root filesystem over NFS; if the Linux kernel has the CONFIG_IP_PNP configuration enabled (IP kernel level autoconfiguration), this can be used to enable automatic configuration of IP addresses of devices and of the routing table during kernel boot, based on either information supplied on the kernel command line or by BOOTP or RARP protocols.

The advantage of this mechanism is that you don't have to spend precious system memory (RAM and flash) for network configuration tools like ifconfig or route - especially in Embedded Systems where you seldom have to change the network configuration while the system is running.

We can use U-Boot environment variables to store all necessary configuration parameters:

=> setenv ipaddr 192.168.20.38
=> setenv serverip 192.168.1.1
=> setenv netmask 255.255.0.0
=> setenv hostname m28
=> setenv rootpath /opt/eldk-5.2/armv5te/rootfs
=> saveenv

Then you can use these variables to build the boot arguments to be passed to the Linux kernel:

=> setenv nfsargs 'root=/dev/nfs rw nfsroot=${serverip}:${rootpath}'

Note how apostrophes are used to delay the substitution of the referenced environment variables. This way, the current values of these variables get inserted when assigning values to the "bootargs" variable itself later, i. e. when it gets assembled from the given parts before passing it to the kernel. This allows us to simply redefine any of the variables (say, the value of "ipaddr" if it has to be changed), and the changes will automatically propagate to the Linux kernel.

Note: You cannot use this method directly to define for example the "bootargs" environment variable, as the implicit usage of this variable by the "bootm" command will not trigger variable expansion - this happens only when using the "setenv" command.

In the next step, this can be used for a flexible method to define the "bootargs" environment variable by using a function-like approach to build the boot arguments step by step:

=> setenv ramargs setenv bootargs root=/dev/ram rw
=> setenv nfsargs 'setenv bootargs root=/dev/nfs rw nfsroot=${serverip}:${rootpath}'
=> setenv addip 'setenv bootargs ${bootargs} ip=${ipaddr}:${serverip}:${gatewayip}:${netmask}:${hostname}::off'
=> setenv ram_root 'run ramargs addip;bootm ${kernel_addr} ${ramdisk_addr} ${fdt_addr} '
=> setenv nfs_root 'run nfsargs addip;bootm ${kernel_addr} - ${fdt_addr} '

In this setup we define two variables, ram_root and nfs_root, to boot with root filesystem from a ramdisk image or over NFS, respecively. The variables can be executed using U-Boot's run command. These variables make use of the run command itself:

• First, either run ramargs or run nfsargs is used to initialize the bootargs environment variable as needed to boot with ramdisk image or with root over NFS.

• Then, in both cases, run addip is used to append the ip parameter to use the Linux kernel IP autoconfiguration mechanism for configuration of the network settings.

• Finally, the bootm command is used with three resp. two address arguments to boot the Linux kernel image with resp. without a ramdisk image. (We assume here that the variables kernel_addr , ramdisk_addr and fdt_addr have already been set.)

This method can be easily extended to add more customization options when needed.

If you have used U-Boot's network commands before (and/or read the documentation), you will probably have recognized that the names of the U-Boot environment variables we used in the examples above are exactly the same as those used with the U-Boot commands to boot over a network using DHCP or BOOTP. That means that, instead of manually setting network configuration parameters like IP address, etc., these variables will be set automatically to the values retrieved with the network boot protocols. This is explained in detail in the sections about the respective U-Boot commands.

7.5. Networked Operation with Root Filesystem over NFS

This section will show how to boot the target into Linux with no more than U-Boot residing on it. For this we will use the tftp command of U-Boot to transfer a Linux kernel and boot it with the NFS rootfilesystem provided by the ELDK.

For this to work, we rely on some U-Boot environment variables to be set up correctly, i.e. the network parameters, the names of files to transfer via tftp and last but not least some scripts easing the assembly of the Linux command line. The whole process is packaged up into one script shown before we actually execute it.

Note that the Linux kernel will also output the command line used, so you can easily check if everything worked like expected. The command line in this example passes at least the following information to the:

• root=/dev/nfs rw: the root filesystem will be mounted using NFS, and it will be writable.

• nfsroot=192.168.1.1:/opt/eldk-5.2/armv5te/rootfs: the NFS server has the IP address 192.168.1.1, and exports the directory /opt/eldk-5.2/armv5te/rootfs for our system to use as root filesystem.

• ip=192.168.20.38:192.168.1.1:192.168.1.1:255.255.0.0:m28::off: the target has the IP address 192.168.20.38; the NFS server is 192.168.1.1; there is a gateway at IP address 192.168.1.1; the netmask is 255.255.0.0 and our hostname is m28. The first ethernet interface (eth0) willbe used, and the Linux kernel will immediately use this network configuration and not try to re-negotiate it (IP autoconfiguration is off).

See Documentation/nfsroot.txt in you Linux kernel source directory for more information about these parameters and other options.

7.5.1. Bootlog of tftp'd Linux kernel with Root Filesystem over NFS

=> setenv bootfile /tftpboot/duts/m28/uImage
=> setenv rootpath /opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk/
=> printenv net_nfs
net_nfs=run fdt_netload kernel_netload nfsargs addip addcons addmtd addmisc;bootm ${kernel_addr_r} - ${fdt_addr_r}
=> run net_nfs
Using FEC0 device
TFTP from server 192.168.1.1; our IP address is 192.168.20.33
Filename 'duts/m28/imx28-m28evk.dtb'.
Load address: 0x41000000
Loading: ##
done
Bytes transferred = 16547 (40a3 hex)
Using FEC0 device
TFTP from server 192.168.1.1; our IP address is 192.168.20.33
Filename '/tftpboot/duts/m28/uImage'.
Load address: 0x42000000
Loading: #################################################################
         #################################################################
         ####################
done
Bytes transferred = 2191008 (216ea0 hex)
## Booting kernel from Legacy Image at 42000000 ...
   Image Name:   Linux-3.6.0-next-20121001-00014-
   Created:      2012-10-02  13:23:40 UTC
   Image Type:   ARM Linux Kernel Image (uncompressed)
   Data Size:    2190944 Bytes = 2.1 MiB
   Load Address: 40008000
   Entry Point:  40008000
   Verifying Checksum ... OK
## Flattened Device Tree blob at 41000000
   Booting using the fdt blob at 0x41000000
   Loading Kernel Image ... OK
OK
   Loading Device Tree to 4fb3c000, end 4fb430a2 ... OK

Starting kernel ...

Uncompressing Linux... done, booting the kernel.
[    0.000000] Booting Linux on physical CPU 0
[    0.000000] Initializing cgroup subsys cpu
[    0.000000] Linux version 3.6.0-next-20121001-00014-g61a70ae (root@mashiro) (gcc version 4.7.2 (Debian 4.7.2-2) ) #2577 PREEMPT Tue Oct 2 15:23:26 CEST 2012
[    0.000000] CPU: ARM926EJ-S [41069265] revision 5 (ARMv5TEJ), cr=00053177
[    0.000000] CPU: VIVT data cache, VIVT instruction cache
[    0.000000] Machine: Freescale i.MX28 (Device Tree), model: DENX M28EVK
[    0.000000] bootconsole [earlycon0] enabled
[    0.000000] Memory policy: ECC disabled, Data cache writeback
[    0.000000] Built 1 zonelists in Zone order, mobility grouping on.  Total pages: 65024
[    0.000000] Kernel command line: root=/dev/nfs rw nfsroot=192.168.1.1:/opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk/,v3,tcp ip=192.168.20.33:192.168.1.1:192.168.1.254:255.255.0.0:m28:eth0:off console=ttyAMA0,115200 mtdparts=gpmi-nand:3m(bootloader)ro,512k(environment),512k(redundant-environment),4m(kernel),-(filesystem) fec_mac=${ethaddr} earlyprintk user_debug=29
[    0.000000] PID hash table entries: 1024 (order: 0, 4096 bytes)
[    0.000000] Dentry cache hash table entries: 32768 (order: 5, 131072 bytes)
[    0.000000] Inode-cache hash table entries: 16384 (order: 4, 65536 bytes)
[    0.000000] Memory: 256MB = 256MB total
[    0.000000] Memory: 254044k/254044k available, 8100k reserved, 0K highmem
[    0.000000] Virtual kernel memory layout:
[    0.000000]     vector  : 0xffff0000 - 0xffff1000   (   4 kB)
[    0.000000]     fixmap  : 0xfff00000 - 0xfffe0000   ( 896 kB)
[    0.000000]     vmalloc : 0x90800000 - 0xff000000   (1768 MB)
[    0.000000]     lowmem  : 0x80000000 - 0x90000000   ( 256 MB)
[    0.000000]       .text : 0x80008000 - 0x8050df88   (5144 kB)
[    0.000000]       .init : 0x8050e000 - 0x80538b64   ( 171 kB)
[    0.000000]       .data : 0x8053a000 - 0x80565e40   ( 176 kB)
[    0.000000]        .bss : 0x80565e64 - 0x8059b8f4   ( 215 kB)
[    0.000000] SLUB: Genslabs=13, HWalign=32, Order=0-3, MinObjects=0, CPUs=1, Nodes=1
[    0.000000] NR_IRQS:16 nr_irqs:16 16
[    0.000000] of_irq_init: children remain, but no parents
[    0.000000] sched_clock: 32 bits at 100 Hz, resolution 10000000ns, wraps every 4294967286ms
[    0.000000] Console: colour dummy device 80x30
[    0.010000] Calibrating delay loop... 226.09 BogoMIPS (lpj=1130496)
[    0.090000] pid_max: default: 32768 minimum: 301
[    0.090000] Mount-cache hash table entries: 512
[    0.100000] CPU: Testing write buffer coherency: ok
[    0.110000] Setting up static identity map for 0x40404d20 - 0x40404d5c
[    0.120000] devtmpfs: initialized
[    0.120000] pinctrl core: initialized pinctrl subsystem
[    0.130000] regulator-dummy: no parameters
[    0.130000] NET: Registered protocol family 16
[    0.140000] DMA: preallocated 256 KiB pool for atomic coherent allocations
[    0.170000] Serial: AMBA PL011 UART driver
[    0.170000] 80074000.serial: ttyAMA0 at MMIO 0x80074000 (irq = 218) is a PL011 rev2
[    0.180000] console [ttyAMA0] enabled, bootconsole disabled
[    0.180000] console [ttyAMA0] enabled, bootconsole disabled
[    0.200000] bio: create slab <bio-0> at 0
[    0.200000] mxs-dma 80004000.dma-apbh: initialized
[    0.210000] mxs-dma 80024000.dma-apbx: initialized
[    0.210000] 3P3V: 3300 mV
[    0.220000] vddio-sd0: 3300 mV
[    0.220000] usb0_vbus: 5000 mV
[    0.220000] usb1_vbus: 5000 mV
[    0.230000] SCSI subsystem initialized
[    0.230000] usbcore: registered new interface driver usbfs
[    0.240000] usbcore: registered new interface driver hub
[    0.240000] usbcore: registered new device driver usb
[    0.250000] pps_core: LinuxPPS API ver. 1 registered
[    0.250000] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it>
[    0.260000] PTP clock support registered
[    0.260000] Advanced Linux Sound Architecture Driver Initialized.
[    0.270000] Switching to clocksource mxs_timer
[    0.290000] NET: Registered protocol family 2
[    0.300000] TCP established hash table entries: 8192 (order: 4, 65536 bytes)
[    0.300000] TCP bind hash table entries: 8192 (order: 3, 32768 bytes)
[    0.310000] TCP: Hash tables configured (established 8192 bind 8192)
[    0.320000] TCP: reno registered
[    0.320000] UDP hash table entries: 256 (order: 0, 4096 bytes)
[    0.330000] UDP-Lite hash table entries: 256 (order: 0, 4096 bytes)
[    0.330000] NET: Registered protocol family 1
[    0.340000] RPC: Registered named UNIX socket transport module.
[    0.340000] RPC: Registered udp transport module.
[    0.350000] RPC: Registered tcp transport module.
[    0.350000] RPC: Registered tcp NFSv4.1 backchannel transport module.
[    0.360000] NetWinder Floating Point Emulator V0.97 (double precision)
[    0.400000] NFS: Registering the id_resolver key type
[    0.400000] Key type id_resolver registered
[    0.410000] Key type id_legacy registered
[    0.410000] msgmni has been set to 496
[    0.420000] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 250)
[    0.420000] io scheduler noop registered (default)
[    0.440000] Console: switching to colour frame buffer device 100x30
[    0.460000] mxsfb 80030000.lcdif: initialized
[    0.630000] uart-pl011 80074000.serial: no DMA platform data
[    0.640000] 8006a000.serial: ttyAPP0 at MMIO 0x8006a000 (irq = 215) is a 8006a000.serial
[    0.650000] mxs-auart 8006a000.serial: Found APPUART 3.1.0
[    0.670000] brd: module loaded
[    0.680000] loop: module loaded
[    0.680000] at24 0-0051: 16384 byte 24c128 EEPROM, writable, 32 bytes/write
[    0.690000] ONFI param page 0 valid
[    0.700000] ONFI flash detected
[    0.700000] NAND device: Manufacturer ID: 0x2c, Chip ID: 0xda (Micron MT29F2G08ABAEAWP), page size: 2048, OOB size: 64
[    0.710000] Scanning device for bad blocks
[    1.050000] 5 cmdlinepart partitions found on MTD device gpmi-nand
[    1.050000] Creating 5 MTD partitions on "gpmi-nand":
[    1.060000] 0x000000000000-0x000000300000 : "bootloader"
[    1.070000] 0x000000300000-0x000000380000 : "environment"
[    1.080000] 0x000000380000-0x000000400000 : "redundant-environment"
[    1.080000] 0x000000400000-0x000000800000 : "kernel"
[    1.090000] 0x000000800000-0x000010000000 : "filesystem"
[    1.100000] gpmi-nand 8000c000.gpmi-nand: driver registered.
[    1.130000] m25p80 spi32766.0: m25p80 (1024 Kbytes)
[    1.140000] libphy: Fixed MDIO Bus: probed
[    1.140000] CAN device driver interface
[    1.150000] flexcan 80032000.can: device registered (reg_base=f5032000, irq=190)
[    1.160000] flexcan 80034000.can: device registered (reg_base=f5034000, irq=191)
[    1.170000] libphy: fec_enet_mii_bus: probed
[    1.180000] Initializing USB Mass Storage driver...
[    1.190000] usbcore: registered new interface driver usb-storage
[    1.190000] USB Mass Storage support registered.
[    1.200000] ci_hdrc ci_hdrc.0: doesn't support gadget
[    1.200000] ci_hdrc ci_hdrc.0: ChipIdea HDRC EHCI
[    1.210000] ci_hdrc ci_hdrc.0: new USB bus registered, assigned bus number 1
[    1.260000] ci_hdrc ci_hdrc.0: USB 2.0 started, EHCI 1.00
[    1.260000] hub 1-0:1.0: USB hub found
[    1.270000] hub 1-0:1.0: 1 port detected
[    1.270000] ci_hdrc ci_hdrc.1: doesn't support gadget
[    1.280000] ci_hdrc ci_hdrc.1: ChipIdea HDRC EHCI
[    1.280000] ci_hdrc ci_hdrc.1: new USB bus registered, assigned bus number 2
[    1.340000] ci_hdrc ci_hdrc.1: USB 2.0 started, EHCI 1.00
[    1.340000] hub 2-0:1.0: USB hub found
[    1.350000] hub 2-0:1.0: 1 port detected
[    1.350000] mousedev: PS/2 mouse device common for all mice
[    1.360000] stmp3xxx-rtc 80056000.rtc: rtc core: registered 80056000.rtc as rtc0
[    1.370000] i2c /dev entries driver
[    1.410000] mxs-mmc 80010000.ssp: initialized
[    1.410000] usbcore: registered new interface driver usbhid
[    1.420000] usbhid: USB HID core driver
[    1.440000] sgtl5000 0-000a: Failed to get supply 'VDDD': -517
[    1.450000] 0-000a: 1200 mV normal
[    1.450000] sgtl5000 0-000a: Using internal LDO instead of VDDD
[    1.460000] sgtl5000 0-000a: sgtl5000 revision 0x11
[    1.470000] mxs-sgtl5000 sound.10:  sgtl5000 <-> 80042000.saif mapping ok
[    1.480000] mxs-sgtl5000 sound.10:  sgtl5000 <-> 80046000.saif mapping ok
[    1.490000] TCP: cubic registered
[    1.490000] Initializing XFRM netlink socket
[    1.500000] NET: Registered protocol family 17
[    1.500000] NET: Registered protocol family 15
[    1.510000] can: controller area network core (rev 20120528 abi 9)
[    1.510000] NET: Registered protocol family 29
[    1.520000] can: raw protocol (rev 20120528)
[    1.520000] can: broadcast manager protocol (rev 20120528 t)
[    1.530000] Key type dns_resolver registered
[    1.540000] stmp3xxx-rtc 80056000.rtc: setting system clock to 2012-10-02 16:55:53 UTC (1349196953)
[    1.550000] eth0: Freescale FEC PHY driver [Micrel KSZ8021] (mii_bus:phy_addr=800f0000.etherne:00, irq=-1)
[    1.660000] mmc0: new high speed SD card at address 0002
[    1.670000] mmcblk0: mmc0:0002 00000 1.90 GiB
[    1.680000]  mmcblk0: p1 p2 p3
[    3.550000] libphy: 800f0000.etherne:00 - Link is Up - 100/Full
[    3.580000] IP-Config: Complete:
[    3.580000]      device=eth0, addr=192.168.20.33, mask=255.255.0.0, gw=192.168.1.254
[    3.590000]      host=m28, domain=, nis-domain=(none)
[    3.590000]      bootserver=192.168.1.1, rootserver=192.168.1.1, rootpath=
[    3.600000] ALSA device list:
[    3.600000]   #0: mxs_sgtl5000
[    3.620000] VFS: Mounted root (nfs filesystem) on device 0:12.
[    3.630000] Freeing init memory: 168K
INIT: version 2.88 booting
Starting udev
[   20.630000] EXT4-fs (mmcblk0): VFS: Can't find ext4 filesystem
[   20.640000] EXT3-fs (mmcblk0): error: can't find ext3 filesystem on dev mmcblk0.
[   20.650000] EXT2-fs (mmcblk0): error: can't find an ext2 filesystem on dev mmcblk0.
[   20.660000] FAT-fs (mmcblk0): bogus number of reserved sectors
[   20.660000] FAT-fs (mmcblk0): Can't find a valid FAT filesystem
[   20.680000] FAT-fs (mmcblk0): bogus number of reserved sectors
[   20.690000] FAT-fs (mmcblk0): Can't find a valid FAT filesystem
[   22.250000] EXT4-fs (mmcblk0p1): VFS: Can't find ext4 filesystem
[   22.260000] EXT3-fs (mmcblk0p1): error: can't find ext3 filesystem on dev mmcblk0p1.
[   22.270000] EXT2-fs (mmcblk0p1): error: can't find an ext2 filesystem on dev mmcblk0p1.
[   22.290000] FAT-fs (mmcblk0p1): bogus number of reserved sectors
[   22.290000] FAT-fs (mmcblk0p1): Can't find a valid FAT filesystem
[   22.330000] FAT-fs (mmcblk0p1): bogus number of reserved sectors
[   22.330000] FAT-fs (mmcblk0p1): Can't find a valid FAT filesystem
[   22.800000] EXT4-fs (mmcblk0p3): warning: maximal mount count reached, running e2fsck is recommended
[   22.860000] EXT4-fs (mmcblk0p3): mounted filesystem with ordered data mode. Opts: (null)
Starting Bootlog daemon: bootlogd.
Configuring network interfaces... ifup skipped for nfsroot interface eth0
run-parts: /etc/network/if-pre-up.d/nfsroot exited with code 1
done.
Starting portmap daemon...
Unknown HZ value! (91) Assume 100.
net.ipv4.conf.default.rp_filter = 1
net.ipv4.conf.all.rp_filter = 1
Starting atd: OK
INIT: Entering runlevel: 5
Starting system message bus: dbus.
Starting OpenBSD Secure Shell server: sshd
done.
Starting Distributed Compiler Daemon: distcc.
creating NFS state directory: done
starting 8 nfsd kernel threads: rpc.nfsd: Unable to access /proc/fs/nfsd errno 2 (No such file or directory).
Please try, as root, 'mount -t nfsd nfsd /proc/fs/nfsd' and then restart rpc.nfsd to correct the problem
done
starting mountd: done
starting statd: done
Starting system log daemon...0
Starting kernel log daemon...0
Starting internet superserver: xinetd.
Starting Lighttpd Web Server: lighttpd.
Unknown HZ value! (90) Assume 100.
cups: started scheduler.
Starting crond: OK
Starting tcf-agent: OK
Stopping Bootlog daemon: bootlogd.
getty: ioctl() TIOCSPGR
ELDK 5.2.1 generic-armv5te console

generic-armv5te login: root
root@generic-armv5te:~#

7.6. Boot from NAND Flash Memory

The previous section described how to load the Linux kernel image over ethernet using TFTP. This is especially well suited for your development and test environment, when the kernel image is still undergoing frequent changes, for instance because you are modifying kernel code or configuration.

Later in your development cycle you will work on application code or device drivers, which can be loaded dynamically as modules. If the Linux kernel remains the same then you can save the time needed for the TFTP download and put the kernel image into the NAND flash memory of your m28 board.

After having deleted the target flash area, you can download the Linux image and write it to flash. Below is a transcript of the complete operation with a final iminfo command to check the newly placed Linux kernel image in the flash memory.

=> setenv kernel_addr_r 0x42000000
=> setenv nand_off 0x400000
=> nand erase 0x400000 0x3f8000

NAND erase: device 0 offset 0x400000, size 0x3f8000
Erasing at 0x400000 --   3% complete.Erasing at 0x420000 --   6% complete.Erasing at 0x440000 --   9% complete.Erasing at 0x460000 --  12% complete.Erasing at 0x480000 --  15% complete.Erasing at 0x4a0000 --  18% complete.Erasing at 0x4c0000 --  21% complete.Erasing at 0x4e0000 --  25% complete.Erasing at 0x500000 --  28% complete.Erasing at 0x520000 --  31% complete.Erasing at 0x540000 --  34% complete.Erasing at 0x560000 --  37% complete.Erasing at 0x580000 --  40% complete.Erasing at 0x5a0000 --  43% complete.Erasing at 0x5c0000 --  46% complete.Erasing at 0x5e0000 --  50% complete.Erasing at 0x600000 --  53% complete.Erasing at 0x620000 --  56% complete.Erasing at 0x640000 --  59% complete.Erasing at 0x660000 --  62% complete.Erasing at 0x680000 --  65% complete.Erasing at 0x6a0000 --  68% complete.Erasing at 0x6c0000 --  71% complete.Erasing at 0x6e0000 --  75% complete.Erasing at 0x700000 --  78% complete.Erasing at 0x720000 --  81% complete.Erasing at 0x740000 --  84% complete.Erasing at 0x760000 --  87% complete.Erasing at 0x780000 --  90% complete.Erasing at 0x7a0000 --  93% complete.Erasing at 0x7c0000 --  96% complete.Erasing at 0x7e0000 -- 100% complete.
OK
=> tftp 0x42000000 /tftpboot/duts/m28/uImage
Using FEC0 device
TFTP from server 192.168.1.1; our IP address is 192.168.20.33
Filename '/tftpboot/duts/m28/uImage'.
Load address: 0x42000000
Loading: #################################################################
         #################################################################
         ####################
done
Bytes transferred = 2191008 (216ea0 hex)
=> iminfo 0x42000000

## Checking Image at 42000000 ...
   Legacy image found
   Image Name:   Linux-3.6.0-next-20121001-00014-
   Created:      2012-10-02  13:23:40 UTC
   Image Type:   ARM Linux Kernel Image (uncompressed)
   Data Size:    2190944 Bytes = 2.1 MiB
   Load Address: 40008000
   Entry Point:  40008000
   Verifying Checksum ... OK
=> nand write ${kernel_addr_r} ${nand_off} ${filesize}

NAND write: device 0 offset 0x400000, size 0x216ea0
 2191008 bytes written: OK
=> setenv nand_off
=> saveenv
Saving Environment to NAND...
Erasing redundant NAND...
Erasing at 0x380000 --  25% complete.Erasing at 0x3a0000 --  50% complete.Erasing at 0x3c0000 --  75% complete.Erasing at 0x3e0000 -- 100% complete.
Writing to redundant NAND... done
=>

Note how the filesize variable (which gets set by the TFTP transfer) is used to automatically adjust for the actual image size.

Since kernel requires the flattened device tree blob to be passed at boot time, you have to also write the blob to the flash memory. Below is a transcript of this operation.

=> setenv fdt_addr_r 0x41000000
=> setenv nand_off 0x007e0000
=> setenv nand_len 0x20000
=> setenv cmp_addr_r 0x44000000
=> nand erase ${nand_off} ${nand_len}

NAND erase: device 0 offset 0x7e0000, size 0x20000
Erasing at 0x7e0000 -- 100% complete.
OK
=> tftp 0x41000000 /tftpboot/duts/m28/imx28-m28evk.dtb
Using FEC0 device
TFTP from server 192.168.1.1; our IP address is 192.168.20.33
Filename '/tftpboot/duts/m28/imx28-m28evk.dtb'.
Load address: 0x41000000
Loading: ##
done
Bytes transferred = 16547 (40a3 hex)
=> nand write ${fdt_addr_r} ${nand_off} ${filesize}

NAND write: device 0 offset 0x7e0000, size 0x40a3
 16547 bytes written: OK
=> setenv nand_off
=> setenv nand_len
=> saveenv
Saving Environment to NAND...
Erasing NAND...
Erasing at 0x300000 --  25% complete.Erasing at 0x320000 --  50% complete.Erasing at 0x340000 --  75% complete.Erasing at 0x360000 -- 100% complete.
Writing to NAND... done
=>

Now we can boot directly from flash. All we need to do is passing the in-flash address of the image (42000000) and the in-flash address of the flattened device tree (41000000) with the bootm command; we also make the definition of the bootargs variable permanent now:

=> setenv bootcmd bootm 42000000 - 41000000 
=> setenv bootargs root=/dev/nfs rw nfsroot=${serverip}:${rootpath} ip=${ipaddr}:${serverip}:${gatewayip}:${netmask}:${hostname}::off

Use printenv to verify that everything is OK before you save the environment settings:

=> printenv
bootdelay=5
baudrate=115200
stdin=serial
stdout=serial
stderr=serial
bootcmd=bootm 42000000 - 41000000 
bootargs=root=/dev/nfs rw nfsroot=192.168.1.1:/opt/eldk-5.2/armv5te/rootfs
ip=192.168.20.38:192.168.1.1:192.168.1.1:255.255.0.0:m28::off
....

=> saveenv

To test booting from flash you can now reset the board (either by power-cycling it, or using the U-Boot command reset), or you can manually call the boot command which will run the commands in the bootcmd variable:

Note: Included topic DULGData_m28.LinuxBootSelfNand does not exist yet

-- HeikoSchocher - 04 Nov 2011

7.7. Standalone Operation with Ramdisk Image

When your application development is completed, you usually will want to run your Embedded System standalone, i. e. independent from external resources like NFS filesystems. Instead of mounting the root filesystem from a remote server you can use a compressed ramdisk image, which is stored in flash memory and loaded into RAM when the system boots.

Load the ramdisk image into RAM and write it to flash as follows:

Note: Included topic DULGData_m28.UBootInstallNandRamdisk does not exist yet

To tell the Linux kernel to use the ramdisk image as root filesystem you have to modify the command line arguments passed to the kernel, and to pass two arguments to the bootm command, the first is the memory address of the Linux kernel image, the second that of the ramdisk image:

Note: Included topic DULGData_m28.LinuxBootSelfNand does not exist yet

8. Building and Using Modules

This section still needs to be written (this is a wiki, so please feel free to contribute!).

In the meantime, please refer to file Documentation/kbuild/modules.txt in the Linux source tree.

9. Advanced Topics

This section lists some advanced topics of interest to users of U-Boot and Linux.

9.1. Flash Filesystems

9.1.1. Memory Technology Devices

All currently available flash filesystems are based on the Memory Technology Devices MTD layer, so you must enable (at least) the following configuration options to get flash filesystem support in your system:

CONFIG_MTD=y
CONFIG_MTD_PARTITIONS=y
CONFIG_MTD_CHAR=y
CONFIG_MTD_BLOCK=y
CONFIG_MTD_CFI=y
CONFIG_MTD_GEN_PROBE=y
CONFIG_MTD_CFI_AMDSTD=y
CONFIG_MTD_ROM=y
CONFIG_MTD_m28=y

Note: this configuration uses CFI conformant AMD flash chips; you may need to adjust these settings on other boards.

The partition layout of the flash devices is contained in the flat device tree for the system (see 13.1. Flat Device Tree).

Informational messages of the MTD subsystem can be found in the Linux bootlog, i.e. see section 7.5.1. Bootlog of tftp'd Linux kernel with Root Filesystem over NFS.

One can discover this information in a running system using the proc filesystem:

root@generic-armv5te:~#
root@generic-armv5te:~# cat /proc/mt

Now we can run some basic tests to verify that the flash driver routines and the partitioning works as expected:

root@generic-armv5te:~#
root@generic-armv5te:~# hexdump -C /dev/mtd4 | head

In the hex-dumps of the MTD devices you can identify some strings that verify that we indeed see an U-Boot environment, a Linux kernel, a ramdisk image and an empty partition to play wih.

The last output shows the partition to be empty. We can try write some data into it:

0

root@generic-armv5te:~#
root@generic-armv5te:~# date > /tmp/tempfile
root@generic-armv5te:~# dd if=/dev/zero of=/tmp/tempfile bs=1 count=4096 seek=50 Aroot@generic-armv5te:~# dd if=/dev/zero of=/tmp/tempfile bs=1 count=4096 seek=5K0
d4096+0 records in
4096+0 records out
4096 bytes (4.1 kB) copied, 0.0759063 s, 54.0 kB/s
root@generic-armv5te:~# dd if=/tmp/tempfile of=/dev/mtd4 bs=4096 count=1
1+0 records in
1+0 records out
4096 bytes (4.1 kB) copied, 0.0023125 s, 1.8 MB/s
root@generic-armv5te:~# head -1 /dev/mtd4
Tue Oct  2 16:57:16 UTC 2012
root@generic-armv5te:~# dd if=/tmp/tempfile of=/dev/mtd4 bs=4096 count=

As you can see it worked the first time. When we tried to write the (new date) again, we got an error. The reason is that the date has changed (probably at least the seconds) and flash memory cannot be simply overwritten - it has to be erased first.

You can use the eraseall Linux commands to erase a whole MTD partition:

p
root@generic-armv5te:~#
root@generic-armv5te:~# flash_erase -q /dev/mtd4 0 0
date > /tmp/tempfile
root@generic-armv5te:~# date > /tmp/tempfile
root@generic-armv5te:~# dd if=/dev/zero of=/tmp/tempfile bs=1 count=4096 seek=50 Aroot@generic-armv5te:~# dd if=/dev/zero of=/tmp/tempfile bs=1 count=4096 seek=5K0
d 4096+0 records in
4096+0 records out
4096 bytes (4.1 kB) copied, 0.0732188 s, 55.9 kB/s
root@generic-armv5te:~# dd if=/tmp/tempfile of=/dev/mtd4 bs=4096 count=1
1+0 records in
1+0 records out
4096 bytes (4.1 kB) copied, 0.00228125 s, 1.8 MB/s
root@generic-armv5te:~# head -1 /dev/mtd

We have now sufficient proof that the MTD layer is working as expected, so we can try creating a flash filesystem.

9.1.2. Journalling Flash File System

At the moment it seems that the Journalling Flash File System JFFS is the best choice for filesystems in flash memory of embedded devices. You must enable the following configuration options to get JFFS support in your system:

CONFIG_JFFS_FS=y
CONFIG_JFFS_FS_VERBOSE=0

If the flash device is erased, we can simply mount it, and the creation of the JFFS filesystem is performed automagically.

Note: For simple accesses like direct read or write operations or erasing you use the character device interface (/dev/mtd*) of the MTD layer, while for filesystem operations like mounting we must use the block device interface (/dev/mtdblock*).

# eraseall /dev/mtd2
Erased 4096 Kibyte @ 0 -- 100% complete.       
# mount -t jffs /dev/mtdblock2 /mnt
# mount
/dev/root on / type nfs (rw,v2,rsize=4096,wsize=4096,hard,udp,nolock,addr=10.0.0.2)
proc on /proc type proc (rw)
devpts on /dev/pts type devpts (rw)
/dev/mtdblock2 on /mnt type jffs (rw)
# df
Filesystem           1k-blocks      Used Available Use% Mounted on
/dev/root              2087212   1232060    855152  60% /
/dev/mtdblock2            3584         0      3584   0% /mnt

Now you can access the files in the JFFS filesystem in the /mnt directory.

9.1.3. Second Version of JFFS

Probably even more interesting for embedded systems is the second version of JFFS, JFFS2, since it not only fixes a few design issues with JFFS, but also adds transparent compression, so that you can save a lot of precious flash memory.

The mkfs.jffs2 tool is used to create a JFFS2 filesystem image; it populates the image with files from a given directory. For instance, to create a JFFS2 image for a flash partition of 3 MB total size and to populate it with the files from the /tmp/flashtools directory you would use:

# mkfs.jffs2 --pad=3145728 --eraseblock=262144 \
--root=/tmp/flashtools/ --output image.jffs2
# eraseall /dev/mtd4
Erased 3072 Kibyte @ 0 -- 100% complete.       
\# dd if=image.jffs2 of=/dev/mtd4 bs=256k
12+0 records in
12+0 records out
# mount -t jffs2 /dev/mtdblock4 /mnt
# df /mnt
Filesystem           1k-blocks      Used Available Use% Mounted on
/dev/mtdblock4            3072      2488       584  81% /mnt

Note: Especially when you are running time-critical applications on your system you should carefully study if the behaviour of the flash filesystem might have any negative impact on your application. After all, a flash device is not a normal harddisk. This is especially important when your flash filesystem gets full; JFFS2 acts a bit weird then:

• You will note that an increasing amount of CPU time is spent by the filesystem's garbage collection kernel thread.

• Access times to the files on the flash filesystem may increase drastically.

• Attempts to truncate a file (to free space) or to rename it may fail:

  ...
  # cp /bin/bash file
  cp: writing `file': No space left on device
  # >file
  bash: file: No space left on device
  # mv file foo
  mv: cannot create regular file `foo': No space left on device
  

  You will have to use rm to actually delete a file in this situation.

This is especially critical when you are using the flash filesystem to store log files: when your application detects some abnormal condition and produces lots of log messages (which usually are especially important in this situation) the filesystem may fill up and cause extreme long delays - if your system crashes, the most important messages may never be logged at all.

9.1.4. Compressed ROM Filesystem

In some cases it is sufficent to have read-only access to some files, and if the files are big enough it becomes desirable to use some method of compression. The Compressed ROM Filesystem CramFs might be a solution here.

Please note that CramFs has - beside the fact that it is a read-only filesystem - some severe limitations (like missing support for timestamps, hard links, and 16/32 bit uid/gids), but there are many situations in Embedded Systems where it's still useful.

To create a CramFs filesystem a special tool mkcramfs is used to create a file which contains the CramFs image. Note that the CramFs filesystem can be written and read only by kernels with PAGE_CACHE_SIZE == 4096, and some versions of the mkcramfs program may have other restrictions like that the filesystem must be written and read with architectures of the same endianness. Especially the endianness requirement makes it impossible to build the CramFs image on x86 PC host when you want to use it on a Power Architecture® target. The endianness problem has been fixed in the version of mkcramfs that comes with the ELDK.

In some cases you can use a target system running with root filesystem mounted over NFS to create the CramFs image on the native system and store it to flash for further use.

Note: The normal version of the mkcramfs program tries to initialize some entries in the filesystem's superblock with random numbers by reading /dev/random; this may hang permanently on your target because there is not enough input (like mouse movement) to the entropy pool. You may want to use a modified version of mkcramfs which does not depend on /dev/random.

To create a CramFs image, you put all files you want in the filesystem into one directory, and then use the mkcramfs= program as follows:

$ mkdir /tmp/test
$ cp ... /tmp/test
$ du -sk /tmp/test
64      /tmp/test
$ mkcramfs /tmp/test test.cramfs.img
Super block: 76 bytes
  erase
  eraseall
  mkfs.jffs
  lock
  unlock
Directory data: 176 bytes
-54.96% (-4784 bytes)   erase
-55.46% (-5010 bytes)   eraseall
-51.94% (-8863 bytes)   mkfs.jffs
-58.76% (-4383 bytes)   lock
-59.68% (-4215 bytes)   unlock
Everything: 24 kilobytes
$ ls -l test.cramfs.img
-rw-r--r--    1 wd       users       24576 Nov 10 23:44 test.cramfs.img

As you can see, the CramFs image test.cramfs.img takes just 24 kB, while the input directory contained 64 kB of data. Savings of some 60% like in this case are typical CramFs.

Now we write the CramFs image to a partition in flash and test it:

# cp test.cramfs.img /dev/mtd3
# mount -t cramfs /dev/mtdblock3 /mnt
# mount
/dev/root on / type nfs (rw,v2,rsize=4096,wsize=4096,hard,udp,nolock,addr=10.0.0.2)
proc on /proc type proc (rw)
devpts on /dev/pts type devpts (rw)
/dev/mtdblock3 on /mnt type cramfs (rw)
# ls -l /mnt
total 54
-rwxr-xr-x    1 wd       users        8704 Jan  9 16:32 erase
-rwxr-xr-x    1 wd       users        9034 Jan  1 01:00 eraseall
-rwxr-xr-x    1 wd       users        7459 Jan  1 01:00 lock
-rwxr-xr-x    1 wd       users       17063 Jan  1 01:00 mkfs.jffs
-rwxr-xr-x    1 wd       users        7063 Jan  1 01:00 unlock

Note that all the timestamps in the CramFs filesyste are bogus, and so is for instance the output of the df command for such filesystems:

# df /mnt
Filesystem           1k-blocks      Used Available Use% Mounted on
/dev/mtdblock3               0         0         0   -  /mnt

9.1.5. UBI and UBIFS file systems

UBIFS is a flash filesystem, which work on top of the Linux MTD layer. UBI itself is a software layer which basically is a volume management and wear-leveling layer. It provides so called UBI volumes which is a higher level abstraction than a MTD device.

For more documentation about UBI/UBIFS see:

• linux source:Documentation/filesystems/ubifs.txt
• http://www.linux-mtd.infradead.org/doc/ubi.html

This document illustrates the usage of UBI/UBIFS for the m28 board.

9.1.5.1. Create Device Files

First we have to create some device files, which are necessary for using UBI/UBIFS:

ot
root@generic-armv5te:~#
root@generic-armv5te:~# mknod /dev/ubi_ctrl c 10 63
mknod: `/dev/ubi_ctrl': File exists
root@generic-armv5te:~# mknod /dev/ubi0 c 253 0
root@generic-armv5te:~# for i in $(seq 0 9); do mknod /dev/ubi0_$i c 253 $((i +  1)); done
knodroot@generic-armv5te:~# mknod /dev/ubi1 c 252 0
root@generic-armv5te:~# for i in $(seq 0 9); do mknod /dev/ubi1_$i c 252 $((i +  1)); done
ls -lroot@generic-armv5te:~# ls -l /dev/ubi

9.1.5.2. Using UBI on NAND Flash:

Erase the flash partition:

root@generic-armv5te:~# flash_erase -q /dev/mtd4 0 0
root@generic-armv5te:~#

and attach it to UBI:

root@generic-armv5te:~# ubiattach /dev/ubi_ctrl -m 4 -O 2048
[  115.210000] UBI: attaching mtd4 to ubi0
[  116.580000] UBI: scanning is finished
[  116.590000] UBI: empty MTD device detected
[  116.630000] UBI: attached mtd4 (name "filesystem", size 248 MiB) to ubi0
[  116.630000] UBI: PEB size: 131072 bytes (128 KiB), LEB size: 126976 bytes
[  116.650000] UBI: min./max. I/O unit sizes: 2048/2048, sub-page size 2048
[  116.650000] UBI: VID header offset: 2048 (aligned 2048), data offset: 4096
[  116.660000] UBI: good PEBs: 1984, bad PEBs: 0, corrupted PEBs: 0
[  116.670000] UBI: user volume: 0, internal volumes: 1, max. volumes count: 128
[  116.680000] UBI: max/mean erase counter: 0/0, WL threshold: 4096, image sequence number: 1951905100
[  116.690000] UBI: available PEBs: 1976, total reserved PEBs: 8, PEBs reserved for bad PEB handling: 4
[  116.700000] UBI: background thread "ubi_bgt0d" started, PID 719
UBI device number 0, total 1984 LEBs (251920384 bytes, 240.2 MiB), available 1976 LEBs (250904576 bytes, 239.3 MiB), LEB size 126976 bytes (124.0 KiB)
root@generic-armv5te:~#

As this is done, we check if things are done correct:

root@generic-armv5te:~# ubinfo
UBI version:                    1
Count of UBI devices:           1
UBI control device major/minor: 10:59
Present UBI devices:            ubi0
root@generic-armv5te:~#

Create a Volume on the UBI Device:

root@generic-armv5te:~# ubimkvol /dev/ubi0 -N filesystem -m
Set volume size to 250904576
Volume ID 0, size 1976 LEBs (250904576 bytes, 239.3 MiB), LEB size 126976 bytes (124.0 KiB), dynamic, name "filesystem", alignment 1
root@generic-armv5te:~#

As we intend to create just a single volume, we use maximum size ("-m" option).

Mount and use it:

root@generic-armv5te:~# mkdir /mnt/filesystem
root@generic-armv5te:~# mount -t ubifs /dev/ubi0_0 /mnt/filesystem
[  126.720000] UBIFS: default file-system created
[  126.720000] UBIFS: background thread "ubifs_bgt0_0" started, PID 726
[  126.860000] UBIFS: mounted UBI device 0, volume 0, name "filesystem"(null)
[  126.870000] UBIFS: LEB size: 126976 bytes (124 KiB), min./max. I/O unit sizes: 2048 bytes/2048 bytes
[  126.880000] UBIFS: FS size: 249253888 bytes (237 MiB, 1963 LEBs), journal size 12443648 bytes (11 MiB, 98 LEBs)
[  126.890000] UBIFS: reserved for root: 4952683 bytes (4836 KiB)
[  126.900000] UBIFS: media format: w4/r0 (latest is w4/r0), UUID F5A040E7-EFDA-4732-A803-99C0066F2309, small LPT model
root@generic-armv5te:~#

Check with "-a" option:

root@generic-armv5te:~# ubinfo -a
UBI version:                    1
Count of UBI devices:           1
UBI control device major/minor: 10:59
Present UBI devices:            ubi0

ubi0
Volumes count:                           1
Logical eraseblock size:                 126976 bytes, 124.0 KiB
Total amount of logical eraseblocks:     1984 (251920384 bytes, 240.2 MiB)
Amount of available logical eraseblocks: 0 (0 bytes)
Maximum count of volumes                 128
Count of bad physical eraseblocks:       0
Count of reserved physical eraseblocks:  4
Current maximum erase counter value:     2
Minimum input/output unit size:          2048 bytes
Character device major/minor:            247:0
Present volumes:                         0

Volume ID:   0 (on ubi0)
Type:        dynamic
Alignment:   1
Size:        1976 LEBs (250904576 bytes, 239.3 MiB)
State:       OK
Name:        filesystem
Character device major/minor: 247:1
root@generic-armv5te:~#

List which partitions we have mounted, and how many space we have available:

root@generic-armv5te:~# df -h
Filesystem                                           Size  Used Avail Use% Mounted on
rootfs                                               315G  148G  152G  50% /
192.168.1.1:/opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk/  315G  148G  152G  50% /
none                                                 125M  140K  124M   1% /dev
/dev/mmcblk0p2                                        30M   13M   18M  42% /media/mmcblk0p2
/dev/mmcblk0p3                                       1.7G  1.2G  493M  71% /media/mmcblk0p3
tmpfs                                                125M  152K  124M   1% /var/volatile
tmpfs                                                125M     0  125M   0% /media/ram
/dev/ubi0_0                                          220M   24K  215M   1% /mnt/filesystem
root@generic-armv5te:~#

9.1.5.3. Creating UBIFS File System Images

9.1.5.3.1. Determining the Parameters of the used Flash Types:

The "mkfs.ubifs" requires a few parameters that describe the specific features of the underlying flash chips. The easiest way to determine these parameters is to run the "mtdinfo" utility on a running Linux system./dev/mtd7 is NAND flash:

root@generic-armv5te:~# cat /proc/mtd
dev:    size   erasesize  name
mtd0: 00300000 00020000 "bootloader"
mtd1: 00080000 00020000 "environment"
mtd2: 00080000 00020000 "redundant-environment"
mtd3: 00400000 00020000 "kernel"
mtd4: 0f800000 00020000 "filesystem"
mtd5: 00100000 00010000 "spi32766.0"
root@generic-armv5te:~#

root@generic-armv5te:~# mtdinfo -u /dev/mtd4
mtd4
Name:                           filesystem
Type:                           nand
Eraseblock size:                131072 bytes, 128.0 KiB
Amount of eraseblocks:          1984 (260046848 bytes, 248.0 MiB)
Minimum input/output unit size: 2048 bytes
Sub-page size:                  2048 bytes
OOB size:                       64 bytes
Character device major/minor:   90:8
Bad blocks are allowed:         true
Device is writable:             true
Default UBI VID header offset:  2048
Default UBI data offset:        4096
Default UBI LEB size:           126976 bytes, 124.0 KiB
Maximum UBI volumes count:      128

root@generic-armv5te:~# ubinfo /dev/ubi0
ubi0
Volumes count:                           1
Logical eraseblock size:                 126976 bytes, 124.0 KiB
Total amount of logical eraseblocks:     1984 (251920384 bytes, 240.2 MiB)
Amount of available logical eraseblocks: 0 (0 bytes)
Maximum count of volumes                 128
Count of bad physical eraseblocks:       0
Count of reserved physical eraseblocks:  4
Current maximum erase counter value:     2
Minimum input/output unit size:          2048 bytes
Character device major/minor:            247:0
Present volumes:                         0
root@generic-armv5te:~#

The interesting parameters are:

- min-io-size:  corresponds to "Minimum input/output unit size"
- max-leb-cnt:  corresponds to "Amount of eraseblocks"

One more needed parameter from the "ubinfo -a" command:

- leb-size:     corresponds to "Logical eraseblock size"

9.1.5.3.2. Create some Test File System Hierarchy

[marex@pollux]$ cd /tmp/duts-${pid}
[marex@pollux]$ mkdir fs
[marex@pollux]$ chmod 777 fs
[marex@pollux]$ echo Just an example >fs/README
[marex@pollux]$ date >fs/date_of_creation
[marex@pollux]$ ls -l fs
total 8
-rw-rw-r-- 1 marex marex 16 Oct  2 17:01 README
-rw-rw-r-- 1 marex marex 30 Oct  2 17:01 date_of_creation
[marex@pollux]$

Create UBIFS Images for ""userfs"" (NAND)

[marex@pollux]$ mkfs.ubifs --root=fs --min-io-size=2048 --leb-size=126976 --max-leb-cnt=1984 -o image-filesystem.ubifs
WARNING: setting root UBIFS inode UID=GID=0 (root) and permissions to u+rwx,go+rx; use --squash-rino-perm or --nosquash-rino-perm to suppress this warning
[marex@pollux]$ ls -lh image-filesystem.ubifs
-rw-rw-r-- 1 marex marex 1.7M Oct  2 17:01 image-filesystem.ubifs
[marex@pollux]$ cp image-filesystem.ubifs /opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk//home/duts/
[marex@pollux]$

9.1.5.3.3. Installing UBIFS images into existing UBI Volume:

root@generic-armv5te:~# ubiattach /dev/ubi_ctrl -m 4 -O 2048
[  158.760000] UBI: attaching mtd4 to ubi0
[  161.500000] UBI: scanning is finished
[  161.530000] UBI: attached mtd4 (name "filesystem", size 248 MiB) to ubi0
[  161.530000] UBI: PEB size: 131072 bytes (128 KiB), LEB size: 126976 bytes
[  161.550000] UBI: min./max. I/O unit sizes: 2048/2048, sub-page size 2048
[  161.550000] UBI: VID header offset: 2048 (aligned 2048), data offset: 4096
[  161.560000] UBI: good PEBs: 1984, bad PEBs: 0, corrupted PEBs: 0
[  161.580000] UBI: user volume: 1, internal volumes: 1, max. volumes count: 128
[  161.580000] UBI: max/mean erase counter: 2/1, WL threshold: 4096, image sequence number: 1951905100
[  161.600000] UBI: available PEBs: 0, total reserved PEBs: 1984, PEBs reserved for bad PEB handling: 4
[  161.600000] UBI: background thread "ubi_bgt0d" started, PID 735
UBI device number 0, total 1984 LEBs (251920384 bytes, 240.2 MiB), available 0 LEBs (0 bytes), LEB size 126976 bytes (124.0 KiB)
root@generic-armv5te:~# ubiupdatevol /dev/ubi0_0 /home/duts/image-filesystem.ubi fs
root@generic-armv5te:~# mount -t ubifs /dev/ubi0_0 /mnt/filesystem
[  167.500000] UBIFS: background thread "ubifs_bgt0_0" started, PID 738
[  167.640000] UBIFS: mounted UBI device 0, volume 0, name "filesystem"(null)
[  167.640000] UBIFS: LEB size: 126976 bytes (124 KiB), min./max. I/O unit sizes: 2048 bytes/2048 bytes
[  167.650000] UBIFS: FS size: 249507840 bytes (237 MiB, 1965 LEBs), journal size 9023488 bytes (8 MiB, 72 LEBs)
[  167.670000] UBIFS: reserved for root: 0 bytes (0 KiB)
[  167.670000] UBIFS: media format: w4/r0 (latest is w4/r0), UUID F31EAD94-B13A-4F8F-9927-E150346C2DEB, small LPT model
root@generic-armv5te:~# ls -l /mnt/filesystem
total 8
-rw-rw-r-- 1 555 555 16 Oct  2 15:01 README
-rw-rw-r-- 1 555 555 30 Oct  2 15:01 date_of_creation
root@generic-armv5te:~# hexdump -C /mnt/filesystem/date_of_creation
00000000  54 75 65 20 4f 63 74 20  20 32 20 31 37 3a 30 31  |Tue Oct  2 17:01|
00000010  3a 32 38 20 43 45 53 54  20 32 30 31 32 0a        |:28 CEST 2012.|
0000001e
root@generic-armv5te:~#

9.1.5.3.4. Installing UBI images (if no UBI Volumes exist):

If the UBI device is not already formatted to contain suitable volumes, we have to generate UBI images. An UBI image may contain one or more UBI volumes, which in turn may be used to store an UBIFS file system. This is done using the "ubinize" tool, which unfortunately is a bit weird to use: an input configuration ini-file defines all the UBI volumes - their characteristics and the contents, but it does not define the characteristics of the flash flash device - these have to be specified as command-line options.

Also, we should keep in mind that we should no longer use "flash_eraseall" to erase the MTD device, as this has no knowledge about the UBI erase counters. Instead, we should use "ubiformat" to erase the flash and install a new UBI image.

Unmount and detach previously used UBI devices:

root@generic-armv5te:~# umount /mnt/filesystem
[  154.740000] UBIFS: un-mount UBI device 0, volume 0
[  154.750000] UBIFS: background thread "ubifs_bgt0_0" stops
root@generic-armv5te:~# ubidetach /dev/ubi_ctrl -m 4
[  156.310000] UBI: detaching mtd4 from ubi0
[  156.320000] UBI: mtd4 is detached from ubi0
root@generic-armv5te:~#

Create UBI ini-file and create UBI images:

Note: for the volume sizes we use the byte count previously determined by running "ubinfo -a" above.

[marex@pollux]$ cat ubi-filesystem.cfg
[ubifs]
mode=ubi
image=image-filesystem.ubifs
vol_id=0
vol_size=251920384
vol_type=dynamic
vol_name=filesystem
[marex@pollux]$ ubinize --min-io-size=2048 --peb-size=128KiB -s 2048 -o image-filesystem.ubi ubi-filesystem.cfg
[marex@pollux]$ cp image-filesystem.ubi /opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk//home/duts/
[marex@pollux]$

Note that for the NAND device we must pass the "-s 2048" option to ubinize; if we don't, an attempt to attach the created UBI device will result in error messages like these:

root@generic-armv5te:~# ubiattach /dev/ubi_ctrl -m 4
[  196.320000] UBI: attaching mtd4 to ubi0
[  199.060000] UBI: scanning is finished
[  199.090000] UBI: attached mtd4 (name "filesystem", size 248 MiB) to ubi0
[  199.090000] UBI: PEB size: 131072 bytes (128 KiB), LEB size: 126976 bytes
[  199.110000] UBI: min./max. I/O unit sizes: 2048/2048, sub-page size 2048
[  199.110000] UBI: VID header offset: 2048 (aligned 2048), data offset: 4096
[  199.130000] UBI: good PEBs: 1984, bad PEBs: 0, corrupted PEBs: 0
[  199.130000] UBI: user volume: 1, internal volumes: 1, max. volumes count: 128
[  199.150000] UBI: max/mean erase counter: 2/1, WL threshold: 4096, image sequence number: 1951905100
[  199.150000] UBI: available PEBs: 0, total reserved PEBs: 1984, PEBs reserved for bad PEB handling: 4
[  199.170000] UBI: background thread "ubi_bgt0d" started, PID 745
UBI device number 0, total 1984 LEBs (251920384 bytes, 240.2 MiB), available 0 LEBs (0 bytes), LEB size 126976 bytes (124.0 KiB)
root@generic-armv5te:~#

For background information please see http://www.linux-mtd.infradead.org/faq/ubi.html#L_vid_offset_mismatch

Erase MTD partitions and install UBI images:

root@generic-armv5te:/home/duts# ubiformat -q -s 2048 -f /home/duts/image-filesy stem.ubi /dev/mtd4
root@generic-armv5te:/home/duts#

Verify that it worked:

root@generic-armv5te:/home/duts# ubiattach /dev/ubi_ctrl -m 4 -O 2048
[  213.240000] UBI: attaching mtd4 to ubi0
[  215.970000] UBI: scanning is finished
[  216.000000] UBI error: init_volumes: not enough PEBs, required 1986, available 1984
[  216.010000] UBI error: ubi_wl_init: no enough physical eraseblocks (-2, need 1)
[  216.030000] UBI error: ubi_attach_mtd_dev: failed to attach mtd4, error -12
ubiattach: error!: cannot attach mtd4
           error 12 (Cannot allocate memory)
root@generic-armv5te:/home/duts# mount -t ubifs /dev/ubi0_0 /mnt/filesystem
[  218.290000] UBIFS error (pid 749): ubifs_mount: cannot open "/dev/ubi0_0", error -22
mount: wrong fs type, bad option, bad superblock on /dev/ubi0_0,
       missing codepage or helper program, or other error
       (for several filesystems (e.g. nfs, cifs) you might
       need a /sbin/mount.<type> helper program)
       In some cases useful info is found in syslog - try
       dmesg | tail  or so

root@generic-armv5te:/home/duts# ls -al /mnt/filesystem
total 8
drwxr-xr-x 2 root root 4096 Oct  2 15:01 .
drwxr-xr-x 4 root root 4096 Oct  2 15:01 ..
root@generic-armv5te:/home/duts# hexdump -C /mnt/filesystem/date_of_creation
hexdump: /mnt/filesystem/date_of_creation: No such file or directory
hexdump: /mnt/filesystem/date_of_creation: Bad file descriptor
root@generic-armv5te:/home/duts#

9.2. The TMPFS Virtual Memory Filesystem

The tmpfs filesystem, formerly known as shmfs, is a filesystem keeping all files in virtual memory.

Everything in tmpfs is temporary in the sense that no files will be created on any device. If you unmount a tmpfs instance, everything stored therein is lost.

tmpfs puts everything into the kernel internal caches and grows and shrinks to accommodate the files it contains and is able to swap unneeded pages out to swap space. It has maximum size limits which can be adjusted on the fly via 'mount -o remount ...'

If you compare it to ramfs (which was the template to create tmpfs) you gain swapping and limit checking. Another similar thing is the RAM disk (/dev/ram*), which simulates a fixed size hard disk in physical RAM, where you have to create an ordinary filesystem on top. Ramdisks cannot swap and you do not have the possibility to resize them.

9.2.1. Mount Parameters

tmpfs has a couple of mount options:

• size: The limit of allocated bytes for this tmpfs instance. The default is half of your physical RAM without swap. If you oversize your tmpfs instances the machine will deadlock since the OOM handler will not be able to free that memory.
• nr_blocks: The same as size, but in blocks of PAGECACHE_SIZE.
• nr_inodes: The maximum number of inodes for this instance. The default is half of the number of your physical RAM pages.

These parameters accept a suffix k, m or g for kilo, mega and giga and can be changed on remount.

To specify the initial root directory you can use the following mount options:

• mode: The permissions as an octal number
• uid: The user id
• gid: The group id

These options do not have any effect on remount. You can change these parameters with chmod(1), chown(1) and chgrp(1) on a mounted filesystem.

So the following mount command will give you a tmpfs instance on /mytmpfs which can allocate 12MB of RAM/SWAP and it is only accessible by root.

mount -t tmpfs -o size=12M,mode=700 tmpfs /mytmpfs

9.2.2. Kernel Support for tmpfs

In order to use a tmpfs filesystem, the CONFIG_TMPFS option has to be enabled for your kernel configuration. It can be found in the Filesystems configuration group. You can simply check if a running kernel supports tmpfs by searching the contents of /proc/fileysystems:

bash# grep tmpfs /proc/filesystems
nodev   tmpfs
bash#

9.2.3. Usage of tmpfs in Embedded Systems

In embedded systems tmpfs is very well suited to provide read and write space (e.g. /tmp and /var) for a read-only root file system such as CramFs described in section 9.1.4. Compressed ROM Filesystem. One way to achieve this is to use symbolic links. The following code could be part of the startup file /etc/rc.sh of the read-only ramdisk:

#!/bin/sh
...
# Won't work on read-only root: mkdir /tmpfs
mount -t tmpfs tmpfs /tmpfs
mkdir /tmpfs/tmp /tmpfs/var
# Won't work on read-only root: ln -sf /tmpfs/tmp /tmpfs/var /
...

The commented out sections will of course fail on a read-only root filesystem, so you have to create the /tmpfs mount-point and the symbolic links in your root filesystem beforehand in order to successfully use this setup.

9.3. Using MultiMediaCards in Linux"

The MultiMediaCard (MMC) is a flash memory card standard.

Booting Linux with a MMC card connected to the m28 you should find in the bootlog something like that:

mmc0: new high speed SD card at address 0002
mmcblk0: mmc0:0002 00000 1.90 GiB
 mmcblk0: p1

If the mmc card is detected, you should see at least the following device files:

/dev/mmcblk0

If there are partitions on it, you see the following device files:

/dev/mmcblk0pX

X =[1..n] with n=number of partitions

mount the partition:

root@generic-armv5te:~# cd /tmp/duts
root@generic-armv5te:/tmp/duts# ls -al mmc
total 0
drwxr-xr-x 2 root root 40 Oct  2 16:59 .
drwxr-xr-x 3 root root 60 Oct  2 16:59 ..
root@generic-armv5te:/tmp/duts# mount
rootfs on / type rootfs (rw)
192.168.1.1:/opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk/ on / type nfs (rw,relatime,vers=3,rsize=4096,wsize=4096,namlen=255,hard,nolock,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=192.168.1.1,mountvers=3,mountproto=tcp,local_lock=all,addr=192.168.1.1)
proc on /proc type proc (rw,relatime)
sysfs on /sys type sysfs (rw,relatime)
none on /dev type tmpfs (rw,relatime,mode=755)
/dev/mmcblk0p2 on /media/mmcblk0p2 type vfat (rw,sync,relatime,fmask=0022,dmask=0022,codepage=cp437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro)
/dev/mmcblk0p3 on /media/mmcblk0p3 type ext4 (rw,sync,relatime,data=ordered)
devpts on /dev/pts type devpts (rw,relatime,gid=5,mode=620,ptmxmode=000)
tmpfs on /var/volatile type tmpfs (rw,relatime)
tmpfs on /media/ram type tmpfs (rw,relatime)
root@generic-armv5te:/tmp/duts# df
Filesystem                                          1K-blocks      Used Available Use% Mounted on
rootfs                                              330279532 155153328 158348988  50% /
192.168.1.1:/opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk/ 330279532 155153328 158348988  50% /
none                                                   127104       140    126964   1% /dev
/dev/mmcblk0p2                                          30642     12666     17976  42% /media/mmcblk0p2
/dev/mmcblk0p3                                        1781936   1187500    503916  71% /media/mmcblk0p3
tmpfs                                                  127104       168    126936   1% /var/volatile
tmpfs                                                  127104         0    127104   0% /media/ram
root@generic-armv5te:/tmp/duts# mount -t vfat /dev/mmcblk0p2 /tmp/duts/mmc
root@generic-armv5te:/tmp/duts# mount
rootfs on / type rootfs (rw)
192.168.1.1:/opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk/ on / type nfs (rw,relatime,vers=3,rsize=4096,wsize=4096,namlen=255,hard,nolock,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=192.168.1.1,mountvers=3,mountproto=tcp,local_lock=all,addr=192.168.1.1)
proc on /proc type proc (rw,relatime)
sysfs on /sys type sysfs (rw,relatime)
none on /dev type tmpfs (rw,relatime,mode=755)
/dev/mmcblk0p2 on /media/mmcblk0p2 type vfat (rw,sync,relatime,fmask=0022,dmask=0022,codepage=cp437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro)
/dev/mmcblk0p3 on /media/mmcblk0p3 type ext4 (rw,sync,relatime,data=ordered)
devpts on /dev/pts type devpts (rw,relatime,gid=5,mode=620,ptmxmode=000)
tmpfs on /var/volatile type tmpfs (rw,relatime)
tmpfs on /media/ram type tmpfs (rw,relatime)
/dev/mmcblk0p2 on /var/volatile/tmp/duts/mmc type vfat (rw,sync,relatime,fmask=0022,dmask=0022,codepage=cp437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro)
root@generic-armv5te:/tmp/duts# df
Filesystem                                          1K-blocks      Used Available Use% Mounted on
rootfs                                              330279532 155153328 158348988  50% /
192.168.1.1:/opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk/ 330279532 155153328 158348988  50% /
none                                                   127104       140    126964   1% /dev
/dev/mmcblk0p2                                          30642     12666     17976  42% /media/mmcblk0p2
/dev/mmcblk0p3                                        1781936   1187500    503916  71% /media/mmcblk0p3
tmpfs                                                  127104       168    126936   1% /var/volatile
tmpfs                                                  127104         0    127104   0% /media/ram
/dev/mmcblk0p2                                          30642     12666     17976  42% /var/volatile/tmp/duts/mmc
root@generic-armv5te:/tmp/duts# ls -al mmc
total 12682
drwxr-xr-x 2 root root   16384 Jan  1  1970 .
drwxr-xr-x 3 root root      60 Oct  2 16:59 ..
-rwxr-xr-x 1 root root      29 Oct  2 15:44 date_of_creation
-rwxr-xr-x 1 root root      29 Sep  5 04:12 date_of_modification
-rwxr-xr-x 1 root root    5410 May 22 11:32 env.txt
-rwxr-xr-x 1 root root    3754 Apr 17 10:13 env.txt.OLD
-rwxr-xr-x 1 root root 1152054 Feb  7  2012 img.bmp
-rwxr-xr-x 1 root root 1152054 Feb  7  2012 img1.bmp
-rwxr-xr-x 1 root root    1024 Oct  2 15:44 random.hex
-rwxr-xr-x 1 root root  233738 Feb  8  2012 slide-1.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-2.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-3.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-4.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-5.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-6.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-7.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-8.bmp
-rwxr-xr-x 1 root root 2340744 Apr 18 07:07 uImage
root@generic-armv5te:/tmp/duts#

write some files to the MMC:

root@generic-armv5te:/tmp/duts# date > /tmp/duts/mmc/date_of_creation
root@generic-armv5te:/tmp/duts# dd if=/dev/urandom of=/tmp/duts/random.hex bs=10 24 count=1
1+0 records in
1+0 records out
1024 bytes (1.0 kB) copied, 0.00196875 s, 520 kB/s
root@generic-armv5te:/tmp/duts# cp random.hex mmc/
root@generic-armv5te:/tmp/duts# cmp random.hex mmc/random.hex
root@generic-armv5te:/tmp/duts# ls -al mmc/
total 12682
drwxr-xr-x 2 root root   16384 Jan  1  1970 .
drwxr-xr-x 3 root root      80 Oct  2 16:59 ..
-rwxr-xr-x 1 root root      29 Oct  2 16:59 date_of_creation
-rwxr-xr-x 1 root root      29 Sep  5 04:12 date_of_modification
-rwxr-xr-x 1 root root    5410 May 22 11:32 env.txt
-rwxr-xr-x 1 root root    3754 Apr 17 10:13 env.txt.OLD
-rwxr-xr-x 1 root root 1152054 Feb  7  2012 img.bmp
-rwxr-xr-x 1 root root 1152054 Feb  7  2012 img1.bmp
-rwxr-xr-x 1 root root    1024 Oct  2 16:59 random.hex
-rwxr-xr-x 1 root root  233738 Feb  8  2012 slide-1.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-2.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-3.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-4.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-5.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-6.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-7.bmp
-rwxr-xr-x 1 root root 1152054 Feb  8  2012 slide-8.bmp
-rwxr-xr-x 1 root root 2340744 Apr 18 07:07 uImage
root@generic-armv5te:/tmp/duts#

unmount the partition with:

root@generic-armv5te:/tmp/duts# umount /tmp/duts/mmc
root@generic-armv5te:/tmp/duts# mount
rootfs on / type rootfs (rw)
192.168.1.1:/opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk/ on / type nfs (rw,relatime,vers=3,rsize=4096,wsize=4096,namlen=255,hard,nolock,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=192.168.1.1,mountvers=3,mountproto=tcp,local_lock=all,addr=192.168.1.1)
proc on /proc type proc (rw,relatime)
sysfs on /sys type sysfs (rw,relatime)
none on /dev type tmpfs (rw,relatime,mode=755)
/dev/mmcblk0p2 on /media/mmcblk0p2 type vfat (rw,sync,relatime,fmask=0022,dmask=0022,codepage=cp437,iocharset=iso8859-1,shortname=mixed,errors=remount-ro)
/dev/mmcblk0p3 on /media/mmcblk0p3 type ext4 (rw,sync,relatime,data=ordered)
devpts on /dev/pts type devpts (rw,relatime,gid=5,mode=620,ptmxmode=000)
tmpfs on /var/volatile type tmpfs (rw,relatime)
tmpfs on /media/ram type tmpfs (rw,relatime)
root@generic-armv5te:/tmp/duts# df
Filesystem                                          1K-blocks      Used Available Use% Mounted on
rootfs                                              330279532 155153328 158348988  50% /
192.168.1.1:/opt/eldk-5.2.1/armv5te/rootfs-lsb-sdk/ 330279532 155153328 158348988  50% /
none                                                   127104       140    126964   1% /dev
/dev/mmcblk0p2                                          30642     12666     17976  42% /media/mmcblk0p2
/dev/mmcblk0p3                                        1781936   1187500    503916  71% /media/mmcblk0p3
tmpfs                                                  127104       172    126932   1% /var/volatile
tmpfs                                                  127104         0    127104   0% /media/ram
root@generic-armv5te:/tmp/duts#

9.4. Splash Screen Support in Linux

To complement the U-Boot Splash Screen feature the new configuration option "CONFIG_FB_PRE_INIT_FB" was added to the Linux kernel. This allows the Linux kernel to skip certain parts of the framebuffer initialization and to reuse the framebuffer contents that was set up by the U-Boot firmware. This allows to have an image displayed nearly immediately after power-on, so the delay needed to boot the Linux kernel is masked to the user.

The current implementation has some limitations:

• We did not succeed in reusing the previously allocated framebuffer contents directly. Instead, Linux will allocate a new framebuffer, copy the contents, and then switch the display. This adds a minimal delay to the boot time, but is otherwise invisible to the user.
• Linux manages its own colormap, and we considered it too much effort to keep the same settings as used by U-Boot. Instead we use the "trick" that U-Boot will fill the color map table backwards (top down). This works pretty well for images which use no more than 200...225 colors. If the images uses more colors, a bad color mapping may result.
  We strongly recommend to convert all images that will be loaded as Linux splash screens to use no more than 225 colors. The "ppmquant" tool can be used for this purpose (see Bitmap Support in U-Boot for details).
• Usually there will be a Linux device driver that is used to adjust the brightness and contrast of the display. When this driver starts, a visible change of brightness will happen if the default settings as used by U-Boot differ.
  We recommend to store settings of brightness and contrast in U-Boot environment variables that can be shared between U-Boot and Linux. This way it is possible (assuming adequate driver support) to adjust the display settings correctly already in U-Boot and thus to avoid any flicker of the display when Linux takes over control.

9.5. Root File System: Design and Building

It is not an easy task to design the root file system for an embedded system. There are three major problems to be solved:

1. what to put in it
2. which file system type to use
3. where to store and how to boot it

For now we will assume that the contents of the root file system is aready known; for example, it is given to us as a directory tree or a tarball which contains all the required files.

We will also assume that our system is a typical resource-limited embedded system so we will especially look for solutions where the root file system can be stored on on-board flash memory or other flash memory based devices like CompactFlash or SD cards, MMC or USB memory sticks.

A widespread approach to build a root file system is to use some Linux distribution (like the ELDK) and to remove things not needed. This approach may be pretty common, but it is almost always terribly wrong. You also don't build a family home by taking a skyscraper and removing parts. Like a house, a root file system should be built bottom up, starting from scratch and adding things you know you need. Never add anything where you don't exactly know what it's needed for.

But our focus here is on the second item: the options we have for chosing a file system type and the consequences this has.

In all cases we will base our experiments on the same content of the root filesystem; we use the images of the SELF (Simple Embedded Linux Framework) that come with the ELDK. In a first step we will transform the SELF images into a tarball to meet the requirements mentioned above:

In a ELDK installation, the SELF images can be found in the /opt/eldk/<architecture>/images/ directory. There is already a compressed ramdisk image in this directory, which we will use (ramdisk_image.gz):

1. Uncompress ramdisk image:

   bash$ gzip -d -c -v /opt/eldk/ppc_8xx/images/ramdisk_image.gz >/tmp/ramdisk_image
   /opt/eldk/ppc_8xx/images/ramdisk_image.gz:       61.4%
   

   
   Note: The following steps require root permissions!
2. Mount ramdisk image:

   bash# mount -o loop /tmp/ramdisk_image /mnt/tmp
   

3. Create tarball; to avoid the need for root permissions in the following steps we don't include the device files in our tarball:

   bash# cd /mnt/tmp
   bash# tar -zc --exclude='dev/*' -f /tmp/rootfs.tar.gz *
   

4. Instead, we create a separate tarball which contains only the device entries so we can use them when necessary (with cramfs):

   bash# tar -zcf /tmp/devices.tar.gz dev/
   bash# cd /tmp
   

5. Unmount ramdisk image:

   bash# umount /mnt/tmp
   

9.5.1. Root File System on a Ramdisk

Ram disks are used very often to hold the root file system of embedded systems. They have several advantages:

• well-known
• well-supported by the Linux kernel
• simple to build
• simple to use - you can even combine the ramdisk with the Linux kernel into a single image file
• RAM based, thus pretty fast
• writable file system
• original state of file system after each reboot = easy recovery from accidental or malicious data corruption etc.

• big memory footprint: you always have to load the complete filesystem into RAM, even if only small parts of are actually used
• slow boot time: you have to load (and uncompress) the whole image before the first application process can start
• only the whole image can be replaced (not individual files)
• additional storage needed for writable persistent data

In almost all cases you will use an ext2 file system in your ramdisk image. The following steps are needed to create it:

1. Create a directory tree with the content of the target root filesystem. We do this by unpacking our master tarball:

   $ mkdir rootfs
   $ cd rootfs
   $ tar zxf /tmp/rootfs.tar.gz
   

2. We use the genext2fs tool to create the ramdisk image as this allows to use a simple text file to describe which devices shall be created in the generated file system image. That means that no root permissions are required at all. We use the following device table rootfs_devices.tab:

   #<name>    <type> <mode> <uid> <gid> <major> <minor> <start>  <inc>  <count>
   /dev            d  755  0       0        -      -       -       -       -
   /dev/console    c  640  0       0        5      1       -       -       -
   /dev/fb0        c  640  0       0       29      0       -       -       -
   /dev/full       c  640  0       0        1      7       -       -       -
   /dev/hda        b  640  0       0        3      0       -       -       -
   /dev/hda        b  640  0       0        3      1       1       1       16
   /dev/kmem       c  640  0       0        1      2       -       -       -
   /dev/mem        c  640  0       0        1      1       -       -       -
   /dev/mtd        c  640  0       0       90      0       0       2       16
   /dev/mtdblock   b  640  0       0       31      0       0       1       16
   /dev/mtdr       c  640  0       0       90      1       0       2       16
   /dev/nftla      b  640  0       0       93      0       -       -       -
   /dev/nftla      b  640  0       0       93      1       1       1       8
   /dev/nftlb      b  640  0       0       93      16      -       -       -
   /dev/nftlb      b  640  0       0       93      17      1       1       8
   /dev/null       c  640  0       0        1      3       -       -       -
   /dev/ptyp       c  640  0       0        2      0       0       1       10
   /dev/ptypa      c  640  0       0        2      10      -       -       -
   /dev/ptypb      c  640  0       0        2      11      -       -       -
   /dev/ptypc      c  640  0       0        2      12      -       -       -
   /dev/ptypd      c  640  0       0        2      13      -       -       -
   /dev/ptype      c  640  0       0        2      14      -       -       -
   /dev/ptypf      c  640  0       0        2      15      -       -       -
   /dev/ram        b  640  0       0        1      0       0       1       2
   /dev/ram        b  640  0       0        1      1       -       -       -
   /dev/rtc        c  640  0       0       10      135     -       -       -
   /dev/tty        c  640  0       0        4      0       0       1       4
   /dev/tty        c  640  0       0        5      0       -       -       -
   /dev/ttyS       c  640  0       0        4      64      0       1       8
   /dev/ttyp       c  640  0       0        3      0       0       1       10
   /dev/ttypa      c  640  0       0        3      10      -       -       -
   /dev/ttypb      c  640  0       0        3      11      -       -       -
   /dev/ttypc      c  640  0       0        3      12      -       -       -
   /dev/ttypd      c  640  0       0        3      13      -       -       -
   /dev/ttype      c  640  0       0        3      14      -       -       -
   /dev/ttypf      c  640  0       0        3      15      -       -       -
   /dev/zero       c  640  0       0        1      5       -       -       -
   

   A description of the format of this table is part of the manual page for the genext2fs tool, genext2fs(8).
3. We can now create an ext2 file system image using the genext2fs tool:

   $ ROOTFS_DIR=rootfs                 # directory with root file system content
   $ ROOTFS_SIZE=3700                  # size of file system image
   $ ROOTFS_FREE=100                   # free space wanted
   $ ROOTFS_INODES=380                 # number of inodes
   $ ROOTFS_DEVICES=rootfs_devices.tab # device description file
   $ ROOTFS_IMAGE=ramdisk.img          # generated file system image
   
   $ genext2fs -U \
           -d ${ROOTFS_DIR} \
           -D ${ROOTFS_DEVICES} \
           -b ${ROOTFS_SIZE} \
           -r ${ROOTFS_FREE} \
           -i ${ROOTFS_INODES} \
           ${ROOTFS_IMAGE}
   

4. Compress the file system image:

   $ gzip -v9 ramdisk.img
   rootfs.img:      55.6% -- replaced with ramdisk.img.gz
   

5. Create an U-Boot image file from it:

   $ mkimage -T ramdisk -C gzip -n 'Test Ramdisk Image' \
   > -d ramdisk.img.gz uRamdisk
   Image Name:   Test Ramdisk Image
   Created:      Sun Jun 12 16:58:06 2005
   Image Type:   PowerPC Linux RAMDisk Image (gzip compressed)
   Data Size:    1618547 Bytes = 1580.61 kB = 1.54 MB
   Load Address: 0x00000000
   Entry Point:  0x00000000
   

We now have a root file system image uRamdisk that can be used with U-Boot.

9.5.2. Root File System on a JFFS2 File System

JFFS2 (Journalling Flash File System version 2) was specifically designed for use on flash memory devices in embedded systems. It is a log-structured file system which means that it is robust against loss of power, crashes or other unorderly shutdowns of the system ("robust" means that data that is just being written when the system goes down may be lost, but the file system itself does not get corrupted and the system can be rebootet without need for any kind of file system check).

Some of the advantages of using JFFS2 as root file system in embedded systems are:

• file system uses compression, thus making efficient use of flash memory
• log-structured file system, thus robust against unorderly shutdown
• flash sector wear-leveling
• writable flash file system

Disadvantages are:

• long mount times (especially older versions)
• slow when reading: files to be read get uncompressed on the fly which eats CPU cycles and takes time
• slow when writing: files to be written get compressed, which eats CPU cycles and takes time, but it may even take much longer until data gets actually stored in flash if the file system becomes full and blocks must be erased first or - even worse - if garbage collection becomes necessary
• The garbage collector thread may run at any time, consuming CPU cycles and blocking accesses to the file system.

Despite the aforementioned disadvantages, systems using a JFFS2 based root file system are easy to build, make efficient use of the available resources and can run pretty reliably.

To create a JFFS2 based root file system please proceed as follows:

1. Create a directory tree with the content of the target root filesystem. We do this by unpacking our master tarball:

   $ mkdir rootfs
   $ cd rootfs
   $ tar zxf /tmp/rootfs.tar.gz
   

2. We can now create a JFFS2 file system image using the mkfs.jffs2 tool:

   $ ROOTFS_DIR=rootfs                 # directory with root file system content
   $ ROOTFS_EBSIZE=0x20000             # erase block size of flash memory
   $ ROOTFS_ENDIAN=b                   # target system is big endian
   $ ROOTFS_DEVICES=rootfs_devices.tab # device description file
   $ ROOTFS_IMAGE=jffs2.img            # generated file system image
   
   $ mkfs.jffs2 -U \
           -d ${ROOTFS_DIR} \
           -D ${ROOTFS_DEVICES} \
           -${ROOTFS_ENDIAN} \
           -e ${ROOTFS_EBSIZE} \
           -o ${ROOTFS_IMAGE}
   mkfs.jffs2: skipping device_table entry '/dev': no parent directory!
   

When booting the Linux kernel prints the following messages showing the default partition map which is used for the flash memory on the TQM8xxL boards:

 TQM flash bank 0: Using static image partition definition
Creating 7 MTD partitions on "TQM8xxL0":
0x00000000-0x00040000 : "u-boot"
0x00040000-0x00100000 : "kernel"
0x00100000-0x00200000 : "user"
0x00200000-0x00400000 : "initrd"
0x00400000-0x00600000 : "cramfs"
0x00600000-0x00800000 : "jffs"
0x00400000-0x00800000 : "big_fs"

1. Erase flash:

   => era 40400000 407FFFFF
   
   ................. done
   Erased 35 sectors
   

2. Download JFFS2 image:

   => tftp 100000 /tftpboot/TQM860L/jffs2.img
   Using FEC ETHERNET device
   TFTP from server 192.168.3.1; our IP address is 192.168.3.80
   Filename '/tftpboot/TQM860L/jffs2.img'.
   Load address: 0x100000
   Loading: #################################################################
            #################################################################
            #################################################################
            #################################################################
            #################################################################
            #################################################################
            ########
   done
   Bytes transferred = 2033888 (1f08e0 hex)
   

3. Copy image to flash:

   => cp.b 100000 40400000 ${filesize}
   Copy to Flash... done
   

4. set up boot arguments to use flash partition 6 as root device:

   => setenv mtd_args setenv bootargs root=/dev/mtdblock6 rw rootfstype=jffs2
   => printenv addip
   addip=setenv bootargs ${bootargs} ip=${ipaddr}:${serverip}:${gatewayip}:${netmask}:${hostname}:${netdev}:off panic=1
   => setenv flash_mtd 'run mtd_args addip;bootm ${kernel_addr}'
   => run flash_mtd
   Using FEC ETHERNET device
   TFTP from server 192.168.3.1; our IP address is 192.168.3.80
   Filename '/tftpboot/TQM860L/uImage'.
   Load address: 0x200000
   Loading: #################################################################
            #################################################################
            ###########
   done
   Bytes transferred = 719233 (af981 hex)
   ## Booting image at 40040000 ...
      Image Name:   Linux-2.4.25
      Created:      2005-06-12  16:32:24 UTC
      Image Type:   PowerPC Linux Kernel Image (gzip compressed)
      Data Size:    782219 Bytes = 763.9 kB
      Load Address: 00000000
      Entry Point:  00000000
      Verifying Checksum ... OK
      Uncompressing Kernel Image ... OK
   Linux version 2.4.25 (wd@xpert) (gcc version 3.3.3 (DENX ELDK 3.1.1 3.3.3-9)) #1 Sun Jun 12 18:32:18 MEST 2005
   On node 0 totalpages: 4096
   zone(0): 4096 pages.
   zone(1): 0 pages.
   zone(2): 0 pages.
   Kernel command line: root=/dev/mtdblock6 rw rootfstype=jffs2 ip=192.168.3.80:192.168.3.1::255.255.255.0:tqm860l:eth1:off panic=1
   Decrementer Frequency = 187500000/60
   Calibrating delay loop... 49.86 BogoMIPS
   ...
   NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
   VFS: Mounted root (jffs2 filesystem).
   Freeing unused kernel memory: 56k init
   
   
   BusyBox v0.60.5 (2005.03.07-06:54+0000) Built-in shell (msh)
   Enter 'help' for a list of built-in commands.
   
   # ### Application running ...
   # mount
   rootfs on / type rootfs (rw)
   /dev/mtdblock6 on / type jffs2 (rw)
   /proc on /proc type proc (rw)
   # df /
   Filesystem           1k-blocks      Used Available Use% Mounted on
   rootfs                    4096      2372      1724  58% /
   

9.5.3. Root File System on a cramfs File System

cramfs is a compressed, read-only file system.

Advantages are:

• file system uses compression, thus making efficient use of flash memory
• Allows for quick boot times as only used files get loaded and uncompressed

Disadvantages are:

• only the whole image can be replaced (not individual files)
• additional storage needed for writable persistent data
• mkcramfs tool does not support device table, so we need root permissions to create the required device files

To create a cramfs based root file system please proceed as follows:

1. Create a directory tree with the content of the target root filesystem. We do this by unpacking our master tarball:

   $ mkdir rootfs
   $ cd rootfs
   $ tar -zxf /tmp/rootfs.tar.gz
   

2. Create the required device files. We do this here by unpacking a special tarball which holds only the device file entries. Note: this requires root permissions!

   # cd rootfs
   # tar -zxf /tmp/devices.tar.gz
   

3. Many tools require some storage place in a filesystem, so we must provide at least one (small) writable filesystem. For all data which may be lost when the system goes down, a "tmpfs" filesystem is the optimal choice. To create such a writable tmpfs filesystem we add the following lines to the /etc/rc.sh script:

   # mount TMPFS because root-fs is readonly
   /bin/mount -t tmpfs -o size=2M tmpfs /tmpfs
   

   Some tools require write permissions on some device nodes (for example, to change ownership and permissions), or dynamically (re-) create such files (for example, /dev/log which is usually a Unix Domain socket). The files are placed in a writable filesystem; in the root filesystem symbolic links are used to point to their new locations:

   dev/ptyp0	→	/tmpfs/dev/ptyp0		dev/ttyp0	→	/tmpfs/dev/ttyp0
   dev/ptyp1	→	/tmpfs/dev/ptyp1		dev/ttyp1	→	/tmpfs/dev/ttyp1
   dev/ptyp2	→	/tmpfs/dev/ptyp2		dev/ttyp2	→	/tmpfs/dev/ttyp2
   dev/ptyp3	→	/tmpfs/dev/ptyp3		dev/ttyp3	→	/tmpfs/dev/ttyp3
   dev/ptyp4	→	/tmpfs/dev/ptyp4		dev/ttyp4	→	/tmpfs/dev/ttyp4
   dev/ptyp5	→	/tmpfs/dev/ptyp5		dev/ttyp5	→	/tmpfs/dev/ttyp5
   dev/ptyp6	→	/tmpfs/dev/ptyp6		dev/ttyp6	→	/tmpfs/dev/ttyp6
   dev/ptyp7	→	/tmpfs/dev/ptyp7		dev/ttyp7	→	/tmpfs/dev/ttyp7
   dev/ptyp8	→	/tmpfs/dev/ptyp8		dev/ttyp8	→	/tmpfs/dev/ttyp8
   dev/ptyp9	→	/tmpfs/dev/ptyp9		dev/ttyp9	→	/tmpfs/dev/ttyp9
   dev/ptypa	→	/tmpfs/dev/ptypa		dev/ttypa	→	/tmpfs/dev/ttypa
   dev/ptypb	→	/tmpfs/dev/ptypb		dev/ttypb	→	/tmpfs/dev/ttypb
   dev/ptypc	→	/tmpfs/dev/ptypc		dev/ttypc	→	/tmpfs/dev/ttypc
   dev/ptypd	→	/tmpfs/dev/ptypd		dev/ttypd	→	/tmpfs/dev/ttypd
   dev/ptype	→	/tmpfs/dev/ptype		dev/ttype	→	/tmpfs/dev/ttype
   dev/ptypf	→	/tmpfs/dev/ptypf		dev/ttypf	→	/tmpfs/dev/ttypf
   tmp	→	/tmpfs/tmp		var	→	/tmpfs/var
   dev/log	→	/var/log/log
   In case you use dhclient also:
   etc/dhclient.conf	→	/tmpfs/var/lib/dhclient.conf		etc/resolv.conf	→	/tmpfs/var/lib/resolv.conf

   
   To place the corresponding directories and device files in the tmpfs file system, the following code is added to the /etc/rc.sh script:

   mkdir -p /tmpfs/tmp /tmpfs/dev \
            /tmpfs/var/lib/dhcp /tmpfs/var/lock /tmpfs/var/run
   
   while read name minor
   do      
           mknod /tmpfs/dev/ptyp$name c 2 $minor
           mknod /tmpfs/dev/ttyp$name c 3 $minor
   done <<__EOD__
   0  0    
   1  1    
   2  2    
   3  3    
   4  4    
   5  5    
   6  6    
   7  7    
   8  8    
   9  9    
   a 10    
   b 11    
   c 12    
   d 13    
   e 14    
   f 15    
   __EOD__ 
   chmod 0666 /tmpfs/dev/*
   

4. We can now create a cramfs file system image using the mkcramfs tool:

   $ ROOTFS_DIR=rootfs                 # directory with root file system content
   $ ROOTFS_ENDIAN="-r"                # target system has reversed (big) endianess
   $ ROOTFS_IMAGE=cramfs.img           # generated file system image
   
   PATH=/opt/eldk/usr/bin:$PATH
   mkcramfs ${ROOTFS_ENDIAN} ${DEVICES} ${ROOTFS_DIR} ${ROOTFS_IMAGE}
   Swapping filesystem endian-ness
     bin
     dev
     etc
   ...
   -48.78% (-86348 bytes)  in.ftpd
   -46.02% (-16280 bytes)  in.telnetd
   -45.31% (-74444 bytes)  xinetd
   Everything: 1864 kilobytes
   Super block: 76 bytes
   CRC: c166be6d
   warning: gids truncated to 8 bits.  (This may be a security concern.)
   

5. We can use the same setup as before for the JFFS2 filesystem, just changing the bootargument to "rootfstype=cramfs"

9.5.4. Root File System on a Read-Only ext2 File System

When storing the root file system in on-board flash memory it seems only natural to look for special flash filesystems like JFFS2, or for other file system types that are designed for such environments like cramfs. It seems to be a bad idea to use a standard ext2 file system because it contains neither any type of wear leveling which is needed for writable file systems in flash memory, nor is it robust against unorderly shutdowns.

The situation changes if we use an ext2 file system which we mount read-only. Such a configuration can be very useful in some situations.

Advantages:

• very fast
• low RAM memory footprint

Disadvantages:

• high flash memory footprint because no compression

To create an ext2 image that can be used as a read-only root file system the following steps are necessary:

1. Create a directory tree with the content of the target root filesystem. We do this by unpacking our master tarball:

   $ mkdir rootfs
   $ cd rootfs
   $ tar -zxf /tmp/rootfs.tar.gz
   

2. Like with the cramfs root file system, we use "tmpfs" for cases where a writable file system is needed and add the following lines to the /etc/rc.sh script:

   # mount TMPFS because root-fs is readonly
   /bin/mount -t tmpfs -o size=2M tmpfs /tmpfs
   

   We also create the same symbolic links for device files that must be placed in a writable filesystem:

   dev/ptyp0	→	/tmpfs/dev/ptyp0		dev/ttyp0	→	/tmpfs/dev/ttyp0
   dev/ptyp1	→	/tmpfs/dev/ptyp1		dev/ttyp1	→	/tmpfs/dev/ttyp1
   dev/ptyp2	→	/tmpfs/dev/ptyp2		dev/ttyp2	→	/tmpfs/dev/ttyp2
   dev/ptyp3	→	/tmpfs/dev/ptyp3		dev/ttyp3	→	/tmpfs/dev/ttyp3
   dev/ptyp4	→	/tmpfs/dev/ptyp4		dev/ttyp4	→	/tmpfs/dev/ttyp4
   dev/ptyp5	→	/tmpfs/dev/ptyp5		dev/ttyp5	→	/tmpfs/dev/ttyp5
   dev/ptyp6	→	/tmpfs/dev/ptyp6		dev/ttyp6	→	/tmpfs/dev/ttyp6
   dev/ptyp7	→	/tmpfs/dev/ptyp7		dev/ttyp7	→	/tmpfs/dev/ttyp7
   dev/ptyp8	→	/tmpfs/dev/ptyp8		dev/ttyp8	→	/tmpfs/dev/ttyp8
   dev/ptyp9	→	/tmpfs/dev/ptyp9		dev/ttyp9	→	/tmpfs/dev/ttyp9
   dev/ptypa	→	/tmpfs/dev/ptypa		dev/ttypa	→	/tmpfs/dev/ttypa
   dev/ptypb	→	/tmpfs/dev/ptypb		dev/ttypb	→	/tmpfs/dev/ttypb
   dev/ptypc	→	/tmpfs/dev/ptypc		dev/ttypc	→	/tmpfs/dev/ttypc
   dev/ptypd	→	/tmpfs/dev/ptypd		dev/ttypd	→	/tmpfs/dev/ttypd
   dev/ptype	→	/tmpfs/dev/ptype		dev/ttype	→	/tmpfs/dev/ttype
   dev/ptypf	→	/tmpfs/dev/ptypf		dev/ttypf	→	/tmpfs/dev/ttypf
   tmp	→	/tmpfs/tmp		var	→	/tmpfs/var
   dev/log	→	/var/log/log
   In case you use dhclient also:
   etc/dhclient.conf	→	/tmpfs/var/lib/dhclient.conf		etc/resolv.conf	→	/tmpfs/var/lib/resolv.conf

   
   To place the corresponding directories and device files in the tmpfs file system, the following code is added to the /etc/rc.sh script:

   mkdir -p /tmpfs/tmp /tmpfs/dev \
            /tmpfs/var/lib/dhcp /tmpfs/var/lock /tmpfs/var/run
   
   while read name minor
   do      
           mknod /tmpfs/dev/ptyp$name c 2 $minor
           mknod /tmpfs/dev/ttyp$name c 3 $minor
   done <<__EOD__
   0  0    
   1  1    
   2  2    
   3  3    
   4  4    
   5  5    
   6  6    
   7  7    
   8  8    
   9  9    
   a 10    
   b 11    
   c 12    
   d 13    
   e 14    
   f 15    
   __EOD__ 
   chmod 0666 /tmpfs/dev/*
   

3. Like we did for the ramdisk, we now create an ext2 file system image using the genext2fs tool:

   $ ROOTFS_DIR=rootfs                 # directory with root file system content
   $ ROOTFS_SIZE=3700                  # size of file system image
   $ ROOTFS_FREE=100                   # free space wanted
   $ ROOTFS_INODES=380                 # number of inodes
   $ ROOTFS_DEVICES=rootfs_devices.tab # device description file
   $ ROOTFS_IMAGE=ext2.img             # generated file system image
   
   $ genext2fs -U \
           -d ${ROOTFS_DIR} \
           -D ${ROOTFS_DEVICES} \
           -b ${ROOTFS_SIZE} \
           -r ${ROOTFS_FREE} \
           -i ${ROOTFS_INODES} \
           ${ROOTFS_IMAGE}
   

4. We can again use the same setup as before for the JFFS2 filesystem, just changing the boot argument to "rootfstype=ext2" (or simply omit it completely as this is the default anyway), and we must change the "rw" argument into "ro" to mount our root file system really read-only:

   ...
   Linux version 2.4.25 (wd@xpert) (gcc version 3.3.3 (DENX ELDK 3.1.1 3.3.3-9)) #1 Sun Jun 12 18:32:18 MEST 2005
   On node 0 totalpages: 4096
   zone(0): 4096 pages.
   zone(1): 0 pages.
   zone(2): 0 pages.
   Kernel command line: root=/dev/mtdblock6 ro rootfstype=ext2 ip=192.168.3.80:192.168.3.1::255.255.255.0:tqm860l:eth1:off panic=1
   Decrementer Frequency = 187500000/60
   Calibrating delay loop... 49.86 BogoMIPS
   ...
   

9.5.5. Root File System on a Flash Card

Using an ext2 file system on a flash memory card (like CompactFlash, SD, MMC or a USB memory stick) is standard technology. To avoid unnecessary flash wear it is a good idea to mount the root file system read-only, or at least using the "noatime" mount option.

For our test we can use the "ext2.img" file from the previous step without changes:

1. In this test we use a standard CompactFlash card which comes with a single partition on it. We use U-Boot to copy the ext2 file system image into this partition:

   => tftp 100000 /tftpboot/TQM860L/ext2.img
   Using FEC ETHERNET device
   TFTP from server 192.168.3.1; our IP address is 192.168.3.80
   Filename '/tftpboot/TQM860L/ext2.img'.
   Load address: 0x100000
   Loading: #################################################################
            #################################################################
            #################################################################
            #################################################################
            #################################################################
            #################################################################
            #################################################################
            #################################################################
            #################################################################
            #################################################################
            #################################################################
            ##########################
   done
   Bytes transferred = 3788800 (39d000 hex)
   => ide part
   
   Partition Map for IDE device 0  --   Partition Type: DOS
   
   Partition     Start Sector     Num Sectors     Type
       1                   32          500704       6
   => ide write 100000 20 1ce8
   
   IDE write: device 0 block # 32, count 7400 ... 7400 blocks written: OK
   

   Note that the "ide write" command takes parameters as hex numbers, and the write count is in terms of disk blocks of 512 bytes each. So we have to use 0x20 for the starts sector of the first partition, and 3788800 / 512 = 7400 = 0x1CE8 for the block count.
2. We now prepare the Linux boot arguments to take this partition as read-only root device:

   => setenv cf_args setenv bootargs root=/dev/hda1 ro
   => setenv flash_cf 'run cf_args addip;bootm ${kernel_addr}'
   => setenv bootcmd run flash_cf
   

3. ...and boot the system:

   ...
   Linux version 2.4.25 (wd@xpert) (gcc version 3.3.3 (DENX ELDK 3.1.1 3.3.3-9)) #1 Sun Jun 12 18:32:18 MEST 2005
   On node 0 totalpages: 4096
   zone(0): 4096 pages.
   zone(1): 0 pages.
   zone(2): 0 pages.
   Kernel command line: root=/dev/hda1 ro ip=192.168.3.80:192.168.3.1::255.255.255.0:tqm860l:eth1:off panic=1
   Decrementer Frequency = 187500000/60
   Calibrating delay loop... 49.86 BogoMIPS
   ...
   

9.5.6. Root File System in a Read-Only File in a FAT File System

This is a more complicated example that shows that - depending on project requirements - many other alternatives for chosing a root file system for your embedded system exist.

The scenario is as follows: on your embedded device you use a cheap and popular storage medium like CompactFlash, MMC or SD cards or USB memory sticks to store both the Linux kernel and your root file system. You want to distribute software updates over the internet: your customers can download the file from your web site, or you sent the images by email. Your customers may use any flash card or memory stick they happen to find, so you have no information about brand or size of the storage device.

Unfortunately most of your customers use Windows systems. And they don't want to be bothered with long instructions how to create special partitions on the storage device or how to write binary images or things like that. A simple "copy file" operation is nearly exhausting their capabilities.

What to do? Well, if copying a file is all your customers can do we should not ask for more. Storage devices like CompactFlash cards etc. typically come with a single partition on it, which holds a FAT or VFAT file system. This cannot be used as a Linux root file system directly, so we have to use some trickery.

Here is one possible solution: Your software distribution consistes of two files: The first file is the Linux kernel with a minimal ramdisk image attached (using the multi-file image format for U-Boot); U-Boot can load and boot such files from a FAT or VFAT file system. The second file is your root file system. For convenience and speed we use again an image of an ext2 file system. When Linux boots, it will initially use the attached ramdisk as root file system. The programs in this ramdisk will mount the FAT or VFAT file system - read-only. Then we can use a loop device (see losetup(8)) to associate the root file system image with a block device which can be used as a mount point. And finally we use pivot_root(8) to change the root file system to our image on the CF card.

This sounds not so complicated, and actually it is quite simple once you understand what needs to be done. Here is a more detailed description:

1. The root file system image is easy: as mantioned before, we will use an ext2 file system image, and to avoid wearing the flash storage device we will use it in read-only mode - we did a read-only ext2 root file system image before, and here we can just re-use the existing image file.
2. The initial ramdisk image that performs the pivot_root step must be created from scratch, but we already know how to create ramdisk images, so we just have to figure out what to put in it.
   
   The most important tool here is nash, a script interpreter that was specifically designed for such purposes (see nash(8)). We don't need any additional tools, and if we use static linking, then the nash binary plus a small script to control it is all we need for our initial ramdisk.
   
   To be precise, we need a couple of (empty) directories (bin, dev, etc, lib, loopfs, mnt, proc, and sysroot), the bin/nash binary, the linuxrc script and a symbolic link sbin pointing to bin:

   drwxr-xr-x    2 wd       users        4096 Apr 13 01:11 bin
   -rwxr-xr-x    1 wd       users      469512 Apr 11 22:47 bin/nash
   drwxr-xr-x    2 wd       users        4096 Apr 12 00:04 dev
   drwxr-xr-x    2 wd       users        4096 Apr 12 00:04 etc
   drwxr-xr-x    2 wd       users        4096 Apr 12 00:04 lib
   -rwxr-xr-x    1 wd       users         511 Apr 13 01:28 linuxrc
   drwxr-xr-x    2 wd       users        4096 Apr 12 00:04 loopfs
   drwxr-xr-x    2 wd       users        4096 Apr 12 00:09 mnt
   drwxr-xr-x    2 wd       users        4096 Apr 12 00:04 proc
   lrwxrwxrwx    1 wd       users           3 Jun 12 18:54 sbin -> bin
   drwxr-xr-x    2 wd       users        4096 Apr 12 00:04 sysroot
   

3. We also need only a minimal device table for creating the initial ramdisk:

   #<name>    <type> <mode> <uid> <gid> <major> <minor> <start>  <inc>  <count>
   /dev            d  755  0       0       -       -       -       -       -
   /dev/console    c  640  0       0        5      1       -       -       -
   /dev/hda        b  640  0       0        3      0       -       -       -
   /dev/hda        b  640  0       0        3      1       1       1       8
   /dev/loop       b  640  0       0        7      0       0       1       4
   /dev/null       c  640  0       0        1      3       -       -       -
   /dev/ram        b  640  0       0        1      0       0       1       2
   /dev/ram        b  640  0       0        1      1       -       -       -
   /dev/tty        c  640  0       0        4      0       0       1       4
   /dev/tty        c  640  0       0        5      0       -       -       -
   /dev/ttyS       c  640  0       0        4      64      0       1       4
   /dev/zero       c  640  0       0        1      5       -       -       -
   

4. To create the initial ramdisk we perform the usual steps:

   $ INITRD_DIR=initrd
   $ INITRD_SIZE=490
   $ INITRD_FREE=0
   $ INITRD_INODES=54
   $ INITRD_DEVICES=initrd_devices.tab
   $ INITRD_IMAGE=initrd.img
   
   $ genext2fs -U \
           -d ${INITRD_DIR} \
           -D ${INITRD_DEVICES} \
           -b ${INITRD_SIZE} \
           -r ${INITRD_FREE} \
           -i ${INITRD_INODES} \
           ${INITRD_IMAGE}
   
   $ gzip -v9 ${INITRD_IMAGE}
   

   The result is a really small (233 kB) compressed ramdisk image.
5. Assuming you already have your Linux kernel image, you can now use mkimage to build an U-Boot multi-file image that combines the Linux kernel and the initial ramdisk:

   $ LINUX_KERNEL=linuxppc_2_4_devel/arch/ppc/boot/images/vmlinux.gz
   $ mkimage -A ppc -O Linux -T multi -C gzip \
   > -n 'Linux with Pivot Root Helper' \
   > -d ${LINUX_KERNEL}:${INITRD_IMAGE}.gz linux.img
   Image Name:   Linux with Pivot Root Helper
   Created:      Mon Jun 13 01:48:11 2005
   Image Type:   PowerPC Linux Multi-File Image (gzip compressed)
   Data Size:    1020665 Bytes = 996.74 kB = 0.97 MB
   Load Address: 0x00000000
   Entry Point:  0x00000000
   Contents:
      Image 0:   782219 Bytes =  763 kB = 0 MB
      Image 1:   238433 Bytes =  232 kB = 0 MB
   

   The newly created file linux.img is the second image we have to copy to the CF card.
   
   We are done.

But wait - one essential part was not mentioned yet: the linuxrc script in our initial ramdisk image which contains all the magic. This script is quite simple:

#!/bin/nash

echo Mounting /proc filesystem
mount -t proc /proc /proc

echo Creating block devices
mkdevices /dev

echo Creating root device
mkrootdev /dev/root
echo 0x0100 > /proc/sys/kernel/real-root-dev

echo Mounting flash card
mount -o noatime -t vfat /dev/hda1 /mnt

echo losetup for filesystem image
losetup /dev/loop0 /mnt/rootfs.img

echo Mounting root filesystem image
mount -o defaults --ro -t ext2 /dev/loop0 /sysroot

echo Running pivot_root
pivot_root /sysroot /sysroot/initrd
umount /initrd/proc

• The first line says that it's a script file for the /bin/nash interpreter.
  Note: even if this file looks like a shell script it is NOT interpreted by a shell, but by the nash interpreter. For a complete list of available nash commands and their syntax please refer to the manual page, nash(8).
• The first action is to mount the /proc pseudo file system which is needed to find out some required information.
• Then we create block device entries for all partitions listed in /proc/partitions (mkdevices command).
• In the next step a block device for our new root file system is created (mkrootdev command).
• Then we mount the CF card. We assume that there is only a single partition on it (/dev/hda1) which is of type VFAT (which also will work with FAT file systems). These assumptions work fine with basicly all memory devices used under Windows.
• We further assume that the file name of the root file system image on the CF card is "rootfs.img" - this file now gets mounted using a loop device (losetup and mount commands).
• Our file system image, is now mounted on the /sysroot directory. In the last step we use pivot_root to make this the new root file system.
• As a final cleanup we unmount the /proc file system which is not needed any more.

And how does this work on our target?

1. First we copy the two images to the CF card; we do this on the target under Linux:

   bash-2.05b# fdisk -l /dev/hda
   
   Disk /dev/hda: 256 MB, 256376832 bytes
   16 heads, 32 sectors/track, 978 cylinders
   Units = cylinders of 512 * 512 = 262144 bytes
   
      Device Boot    Start       End    Blocks   Id  System
   /dev/hda1   *         1       978    250352    6  FAT16
   bash-2.05b# mkfs.vfat /dev/hda1
   mkfs.vfat 2.8 (28 Feb 2001)
   bash-2.05b# mount -t vfat /dev/hda1 /mnt
   bash-2.05b# cp -v linux.img rootfs.img /mnt/
   `linux.img' -> `/mnt/linux.img'
   `rootfs.img' -> `/mnt/rootfs.img'
   bash-2.05b# ls -l /mnt
   total 4700
   -rwxr--r--    1 root     root      1020729 Jun 14 05:36 linux.img
   -rwxr--r--    1 root     root      3788800 Jun 14 05:36 rootfs.img
   bash-2.05b# umount /mnt
   

2. We now prepare U-Boot to load the "uMulti" file (combined Linux kernel and initial ramdisk) from the CF card and boot it:

   => setenv fat_args setenv bootargs rw
   => setenv fat_boot 'run fat_args addip;fatload ide 0:1 200000 linux.img;bootm'
   => setenv bootcmd run fat_boot
   

3. And finally we try it out:

   U-Boot 1.1.3 (Jun 13 2005 - 02:24:00)
   
   CPU:   XPC86xxxZPnnD4 at 50 MHz: 4 kB I-Cache 4 kB D-Cache FEC present
   Board: TQM860LDB0A3-T50.202
   DRAM:  16 MB
   FLASH:  8 MB
   In:    serial
   Out:   serial
   Err:   serial
   Net:   SCC ETHERNET, FEC ETHERNET [PRIME]
   PCMCIA: 3.3V card found: Transcend    256M
               Fixed Disk Card
               IDE interface 
               [silicon] [unique] [single] [sleep] [standby] [idle] [low power]
   Bus 0: OK 
     Device 0: Model: Transcend    256M Firm: 1.1 Ser#: SSSC256M04Z27A25906T
               Type: Removable Hard Disk
               Capacity: 244.5 MB = 0.2 GB (500736 x 512)
   
   Type "run flash_nfs" to mount root filesystem over NFS
   
   Hit any key to stop autoboot:  0 
   reading linux.img
   
   1025657 bytes read
   ## Booting image at 00200000 ...
      Image Name:   Linux with Pivot Root Helper
      Created:      2005-06-13   0:32:41 UTC
      Image Type:   PowerPC Linux Multi-File Image (gzip compressed)
      Data Size:    1025593 Bytes = 1001.6 kB
      Load Address: 00000000
      Entry Point:  00000000
      Contents:
      Image 0:   787146 Bytes = 768.7 kB
      Image 1:   238433 Bytes = 232.8 kB
      Verifying Checksum ... OK
      Uncompressing Multi-File Image ... OK
      Loading Ramdisk to 00f3d000, end 00f77361 ... OK
   Linux version 2.4.25 (wd@xpert) (gcc version 3.3.3 (DENX ELDK 3.1.1 3.3.3-9)) #1 Mon Jun 13 02:32:10 MEST 2005
   On node 0 totalpages: 4096
   zone(0): 4096 pages.
   zone(1): 0 pages.
   zone(2): 0 pages.
   Kernel command line: rw ip=192.168.3.80:192.168.3.1::255.255.255.0:tqm860l:eth1:off panic=1
   Decrementer Frequency = 187500000/60
   Calibrating delay loop... 49.86 BogoMIPS
   ...
   NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.
   RAMDISK: Compressed image found at block 0
   Freeing initrd memory: 232k freed
   VFS: Mounted root (ext2 filesystem).
   Red Hat nash version 4.1.18 starting
   Mounting /proc filesystem
   Creating block devices
   Creating root device
   Mounting flash card
    hda: hda1
    hda: hda1
   losetup for filesystem image
   Mounting root filesystem image
   Running pivot_root
   Freeing unused kernel memory: 60k init
   
   
   BusyBox v0.60.5 (2005.03.07-06:54+0000) Built-in shell (msh)
   Enter 'help' for a list of built-in commands.
   
   # ### Application running ...
   

9.6. Root File System Selection

Now we know several options for file systems we can use, and know how to create the corresponding images. But how can we decide which one to chose?

For practical purposes in embedded systems the following criteria are often essential:

• boot time (i. e. time needed from power on until application code is running)
• flash memory footprint
• RAM memory footprint
• effects on software updates

The following data was measured for the different configurations. All measurements were performed on the same TQM860L board (MPC860 CPU at 50 MHz, 16 MB RAM, 8 MB flash, 256 MB CompactFlash card):

As you can see, the ramdisk solution is the worst of all in terms of RAM memory footprint; also it takes a pretty long time to boot. However, it is one of the few solutions that allow an in-situ update while the system is running.

JFFS2 is easy to use as it's a writable file system but it takes a long time to boot.

A read-only ext2 file system shines when boot time and RAM memory footprint are important; you pay for this with an increased flash memory footprint.

External flash memory devices like CompactFlash cards or USB memory sticks can be cheap and efficient solutions especially when lots of data need to be stored or when easy update procedures are required. -

9.7. Overlay File Systems

Introduction

Overlay File Systems provide an interesting approach to several frequent problems in Embedded Systems. For example, mini_fo is a virtual kernel file system that can make read-only file systems writable. This is done by redirecting modifying operations to a writeable location called "storage directory", and leaving the original data in the "base directory" untouched. When reading, the file system merges the modifed and original data so that only the newest versions will appear. This occurs transparently to the user, who can access the data like on any other read-write file system.

What it is good for?

In embedded systems the main use of mini_fo is to overlay the root file system. This means it is mounted on top of the regular root file system, thereby allowing applications or users to transparently make modifications to it but redirecting these to a different location.

Some examples of why this is usefull are explained in the following sections.

Making a read-only root filesystem writeable

Root file systems stored in flash are often read only, such as cramfs or read only ext2. While this offers major advantages in terms of speed and flash memory footprint, it nevertheless is often desireable to be able to modify the root file system, for example to

• apply (small) software updates without having to burn a whole new root file system image to flash
• make modifications during developement when frequent changes to the root file system occur.

This can be achieved by mounting mini_fo on top of the root file system and using a (probably small) writeable partition as the storage file system. This could be either a JFFS2 flash file system, or during development even an external hard disk. This has the following advantages:

• read-only file systems (fast, small memory footprint) can be used like persistent writable file systems (in contrast to a ramdisk)
• slow flash journalling file systems with large flash memory footprint can be avoided.

Non persistant changes

Ramdisks are often used when the root file system needs to be modified non-persistantly. This works well, but downsides are the large RAM memory footprint and the time costly operation of copying the ramdisk into RAM during startup. These can be avoided by overlaying the root file system as in the previous example but with the difference that the tmpfs file system is used as storage. Thus only modified files are stored in RAM, and can even be swapped out if neccessary. This saves boot time and RAM!

Resetable changes

Mini_fo can be easily used to implement a "reset to factory defaults" function by overlaying the default root file system. When configuration changes are made, these are automatically directed to the storage file system and take precedence over the original files. Now, to restore the system to factory defaults, all that needs to be done is delete the contents of the storage directory. This will remove all changes made to the root file system and return it to the original state.

Note: Deleting the contents of the storage directory should only be done when the overlay file system is unmounted.

Examples

Generally, there are two different ways of overlaying the root file system, which both make sense in different scenarios.

Starting a single application in a chrooted overlayed environment

This is easy. Let's assume "/" is the read-only root file system and /dev/mtdblock5 contains a small JFFS2 flash partition that shall be used to store modifications made by application "/usr/bin/autoPilot":

# mount -t jffs2 /dev/mtdblock5 /tmp/sto
# insmod mini_fo.o
# mount -t mini_fo -o base=/,sto=/tmp/sto/ / /mnt/mini_fo/
# cd /mnt/mini_fo/
# chroot . /usr/bin/autoPilot

Starting the whole system system in chrooted overlayed environment

This is more interesting, and a bit trickier, as mounting needs to be done during system startup after the root file system has been mounted, but before init is started. The best way to do this is to have a script that mounts the mini_fo file system on top of root and then starts init in the chrooted overlayed environment. For example assume the following script "overlay_init", stored in /sbin/:

#!/bin/bash
#
# mount mini_fo overlay file system and execute init
#

# make sure these exist in the read-only file system
STORAGE=/tmp/sto
MOUNT_POINT=/mnt/mini_fo/

# mount tmpfs as storage file system with a maximum size of 32MB
mount -t tmpfs -o rw,size=32M none $STORAGE

/sbin/modprobe mini_fo
mount -t mini_fo -o base=/,sto=$STORAGE / $MOUNT_POINT

exec /usr/sbin/chroot $MOUNT_POINT /sbin/init

echo "exec chroot failed, bad!"
exec /bin/sh

exit 1

Now its easy to choose between a mini_fo overlayed and the regular non overlayed system just by setting the "init" kernel parameter in the boot loader to "init=/sbin/overlay_init".

Tips

• pivot_root(1) can be used with chroot if there is need to access the original non overlayed root file system from the chrooted overlayed environment.

Performance overhead

The mini_fo file system is inserted as an additional layer between the VFS and the native file system, and thus creates some overhead that varies strongly depending of the operation performed.

1. modifying a regular file for the first time
   This results in a copy of the original file beeing created in the storage directory, that is then modified. Overhead depends on the size of the modified file.
2. Reading from files, creating new files, modifying already modified files
   These operations are passed directly through to the lower native layer, and only impose an overhead of 1-2%.

Further information

This section discusses how the mini_fo overlay file system can be used in embedded systems. More general information is available at the mini_fo project page: http://www.denx.de/wiki/Know/MiniFOHome.

9.8. The Persistent RAM File system (PRAMFS)

The pramfs file system supports persistent memory devices such as SRAM. Instead of having a block emulation layer over such a memory area and using a normal file system on top of that, pramfs seeks to induce minimal overhead in this situation. Most important in this respect is that the normal block layer caching of the Linux kernel is circumvented in pramfs.

9.8.1. Mount Parameters

The most important parameters for normal usage are

• physaddr: The physical address of the static memory.
• init: When given, it will initialize the file system to that size.

9.8.2. Example

We will show a sample usage of pramfs in this section using normal DRAM on a board with at least 256MB of memory. For pramfs we reserve the upper 32MB by appending mem=224M to the kernel command line.

First off we generate some testdata on a persistent file system (/tmp) to demonstrate that pramfs survives a reboot (of course with power always applied to keep the DRAM refreshed):

bash-3.00# dd if=/dev/urandom bs=1M count=8 of=/tmp/testdata
8+0 records in
8+0 records out
bash-3.00# 

Next we mount the 32MB that we reserved and initialize it to be 32MB in size and copy the testfile. A final compare shows that the copy was indeed successful so we can reboot:

bash-3.00# mount -t pramfs -o physaddr=0xe000000,init=0x2000000 none /mnt
bash-3.00# cp /tmp/testdata /mnt
bash-3.00# cmp /tmp/testdata /mnt/testdata
bash-3.00# reboot

Having rebooted (using mem=224M on the kernel command line again of course) we mount the file system but this time without the init parameter because it is preinitialized. We then check the contents again:

bash-3.00# mount -t pramfs -o physaddr=0xe000000 none /mnt
bash-3.00# ls /mnt
testdata
bash-3.00# cmp /tmp/testdata /mnt/testdata
bash-3.00#

10. Debugging

The purpose of this document is not to provide an introduction into programming and debugging in general. We assume that you know how to use the GNU debugger gdb and probably it's graphical frontends like ddd. We also assume that you have access to adequate tools for your work, i. e. a BDI2000 BDM/JTAG debugger. The following discussion assumes that the host name of your BDI2000 is bdi.

Please note that there are several limitations in earlier versions of GDB. The version of GDB as distributed with the ELDK contains several bug fixes and extensions. If you find that your GDB behaves differently, have a look at the GDB sources and patches that come with the ELDK source.

10.1. Debugging of U-Boot

When U-Boot starts it is running from ROM space. Running from flash would make it nearly impossible to read from flash while executing code from flash not to speak of updating the U-Boot image in flash itself. To be able to do just that, U-Boot relocates itself to RAM. We therefore have two phases with different program addresses. The following sections show how to debug U-Boot in both phases.

10.1.1. Debugging of U-Boot Before Relocation

Before relocation, the addresses in the ELF file can be used without any problems, so debugging U-Boot in this phase with the BDI2000 is quite easy:

bash[0]$ ${CROSS_COMPILE}gdb u-boot
GNU gdb 5.1.1
Copyright 2002 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "--host=i386-redhat-linux --target=ppc-linux"...

(gdb) target remote bdi:2001
Remote debugging using bdi:2001
0xfffffffc in ?? ()
(gdb) b cpu_init_f
Breakpoint 1 at 0xfffd3310: file cpu_init.c, line 136.
(gdb) c
Continuing.
 
Breakpoint 1, cpu_init_f () at cpu_init.c:136
136             asm volatile("  bl      0f"             ::: "lr");
(gdb) s
137             asm volatile("0:        mflr    3"              ::: "r3");
(gdb)
138             asm volatile("  addi    4, 0, 14"       ::: "r4");
(gdb)

cpu_init_f is the first C function called from the code in start.C.

10.1.2. Debugging of U-Boot After Relocation

For debugging U-Boot after relocation we need to know the address to which U-Boot relocates itself to. When no exotic features like PRAM are used, this address usually is <MAXMEM> - CONFIG_SYS_MONITOR_LEN. In our example with 16MB RAM and CONFIG_SYS_MONITOR_LEN = 192KB this yields the address 0x1000000 - 0x30000 = 0xFD0000.

In other cases, check the source code, and apply some common sense. For example, on Power Architecture® we use "r2" to hold a pointer to the "global data" structure ("struct global_data"); this structure contains a field

unsigned long   relocaddr;      /* Start address of U-Boot in RAM */

(gdb) print/x ((gd_t *)$r2)->relocaddr

(gdb) symbol-file
Discard symbol table from `/home/dzu/denx/cvs-trees/u-boot/u-boot'? (y or n) y
No symbol file now.
(gdb) add-symbol-file u-boot 0xfd0000
add symbol table from file "u-boot" at
        .text_addr = 0xfd0000
(y or n) y
Reading symbols from u-boot...done.
(gdb) b board_init_r
Breakpoint 2 at 0xfd99ac: file board.c, line 533.
(gdb) c
Continuing.

Breakpoint 2, board_init_r (id=0xfbb1f0, dest_addr=16495088) at board.c:533
533     {
(gdb)

The simple example above relocates the symbols of only one section, .text. Other sections of the executable image (like .data, .bss, etc.) are not relocated and this prevents gdb from accessing static and global variables by name. See more sophisticated examples in section 10.3. GDB Startup File and Utility Scripts.

10.2. Linux Kernel Debugging

10.2.1. Linux Kernel and Statically Linked Device Drivers

10.2.2. Dynamically Loaded Device Drivers (Modules)

First start GDB in the root directory of your Linux kernel, using the vmlinux kernel image as file to debug:

bash$ cd <linux-root>
bash$ ${CROSS_COMPILE}gdb vmlinux 
GNU gdb 5.1.1
Copyright 2002 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "--host=i386-redhat-linux --target=ppc-linux".
(gdb)

Now attach to the target and start execution with the commands:

(gdb) target remote bdi:2001
Remote debugging using bdi:2001
0x00000100 in ?? ()
(gdb) c
Continuing.

Now the target should boot Linux as usual. Next you need to load your kernel module on the target:

bash# insmod -m ex_sw.o
Sections:       Size      Address   Align
.this           00000060  cf030000  2**2
.text           000002f4  cf030060  2**2
.rodata         00000134  cf030354  2**2
.data           00000000  cf030488  2**0
.sdata          0000000c  cf030488  2**2
.kstrtab        00000085  cf030494  2**0
.bss            00000000  cf030519  2**0
.sbss           00000008  cf03051c  2**2
...

The option -m prints out the addresses of the various code and data segments ( .text, .data, .sdata, .bss, .sbss ) after relocation. GDB needs these addresses to know where all the symbols are located. We now interrupt GDB to load the symbol table of the module as follows:

(gdb) ^C
Program received signal SIGSTOP, Stopped (signal).
...
(gdb) add-symbol-file <path-to-module-dir>/ex_sw.o 0xcf030060\
 -s .rodata 0xcf030354\
 -s .data   0xcf030488\
 -s .sdata  0xcf030488\
 -s .bss    0xcf030519\
 -s .sbss   0xcf03051c
add symbol table from file "<path-to-module-dir>/ex_sw.o" at
        .text_addr = 0xcf030060
        .rodata_addr = 0xcf030354
        .data_addr = 0xcf030488
        .sdata_addr = 0xcf030488
        .bss_addr = 0xcf030519
        .sbss_addr = 0xcf03051c
(y or n) y
Reading symbols from <path-to-module-dir>/ex_sw.o...done.

Now you can list the source code of the module, set break points or inspect variables as usual:

(gdb) l fun
61      static RT_TASK *thread;
62
63      static int cpu_used[NR_RT_CPUS];
64
65      static void fun(int t)
66      {
67              unsigned int loops = LOOPS;
68              while(loops--) {
69                      cpu_used[hard_cpu_id()]++;
70                      rt_leds_set_mask(1,t);
(gdb)
(gdb) b ex_sw.c:69
Breakpoint 1 at 0xcf03007c: file ex_sw.c, line 69.
(gdb) c
Continuing.
Breakpoint 1, fun (t=1) at ex_sw.c:69
69                      cpu_used[hard_cpu_id()]++;
(gdb) p ntasks
$1 = 16
(gdb) p stack_size
$2 = 3000

The next section demonstrates a way to automate the symbol table loading procedure.

10.2.3. GDB Macros to Simplify Module Loading

The following GDB macros and scripts help you to load kernel modules into GDB in a half-automatic way. It assumes, that the module on the target has been installed with the command:

bash# insmod -m my_module.o > my_module.o.map

In your $HOME directory you need the scripts add-symbol-file.sh and the GDB startup file .gdbinit, which are listed in 10.3. GDB Startup File and Utility Scripts below.

Now you can include the symbol definition into GDB with:

bash$ ${CROSS_COMPILE}gdb vmlinux 
GNU gdb 5.1.1
Copyright 2002 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "--host=i386-redhat-linux --target=ppc-linux".
0x00000100 in ?? ()
c
Continuing.
^C
Program received signal SIGSTOP, Stopped (signal).
0xcf02a91c in ?? ()
(gdb) add-module rtai4/examples/sw/ex_sw.o
add symbol table from file "/HHL/8xx/target/home/wolf/rtai4/examples/sw/ex_sw.o" at
        .text_addr = 0xcf030060
        .rodata_addr = 0xcf030340
        .data_addr = 0xcf030464
        .sdata_addr = 0xcf030464
        .bss_addr = 0xcf0304f5
        .sbss_addr = 0xcf0304f8
(gdb) b ex_sw.c:69
Breakpoint 1 at 0xcf03007c: file ex_sw.c, line 69.
(gdb) c
Continuing.

Breakpoint 1, fun (t=0x1) at ex_sw.c:69
69                      cpu_used[hard_cpu_id()]++;
(gdb) p/d loops
$2 = 999986939
(gdb) p t
$3 = 0x1
(gdb) d b
Delete all breakpoints? (y or n) y
(gdb) c
Continuing.

10.3. GDB Startup File and Utility Scripts

In addition to the add-module macro, the followin example GDB startup file contains a few other useful settings and macros, which you may want to adjust to your local environment:

set output-radix 16

target remote bdi:2001

define reset
        detach
        target remote bdi:2001
end

define add-module
        shell ~/add-symbol-file.sh $arg0
        source ~/add-symbol-file.gdb
end
document add-module
        Usage: add-module <module>

        Do add-symbol-file for module <module> automatically.
        Note: A map file with the extension ".map" must have
        been created with "insmod -m <module> > <module>.map"
        in advance.
end

The following shell script ~/add-symbol-file.sh is used to run the GDB add-symbol-file command automatically:

#!/bin/sh
#
# Constructs the GDB "add-symbol-file" command string
# from the map file of the specified kernel module.

add_sect() {
    ADDR=`awk '/^'$1' / {print $3}' $MAPFILE`
    if [ "$ADDR" != "" ]; then
        echo "-s $1 0x`awk '/^'$1' / {print $3}' $MAPFILE`"
    fi
}

[ $# == 1 ] && [ -r "$1" ] || { echo "Usage: $0 <module>" >&2 ; exit 1 ; }

MAPFILE=$1.map

ARGS="0x`awk '/^.text / {print $3}' $MAPFILE`\
 `add_sect .rodata`\
 `add_sect .data`\
 `add_sect .sdata`\
 `add_sect .bss`\
 `add_sect .sbss`\
"

echo "add-symbol-file $1 $ARGS" > ~/add-symbol-file.gdb

10.4. Tips and Tricks

• To prevent GDB from jumping around in the code when trying to single step, i. e. when it seems as if the code is not executing line by line, you can recompile your code with the following additional compiler options:

  -fno-schedule-insns -fno-schedule-insns2
  

• On some systems (like the MPC8xx or MPC8260) you can only define one hardware breakpoint. Therefore you must delete an existing breakpoint before you can define a new one:

  (gdb) d b
  Delete all breakpoints? (y or n) y
  (gdb) b ex_preempt.c:63
  Breakpoint 2 at 0xcf030080: file ex_preempt.c, line 63.
  

10.5. Application Debugging

10.5.1. Local Debugging

In case there is a native GDB available for your target you can use it for application debugging as usual:

bash$ gcc -Wall -g -o hello hello.c 
bash$ gdb hello
...
(gdb) l
1       #include <stdio.h>
2
3       int main(int argc, char* argv[])
4       {
5               printf ("Hello world\n");
6               return 0;
7       }
(gdb) break 5
Breakpoint 1 at 0x8048466: file hello.c, line 5.
(gdb) run
Starting program: /opt/eldk/ppc_8xx/tmp/hello

Breakpoint 1, main (argc=0x1, argv=0xbffff9f4) at hello.c:5
5               printf ("Hello world\n");
(gdb) c
Continuing.
Hello world

Program exited normally.

10.5.2. Remote Debugging

gdbserver allows you to connect your program with a remote GDB using the "target remote" command. On the target machine, you need to have a copy of the program you want to debug. gdbserver does not need your program's symbol table, so you can strip the program if necessary to save space. GDB on the host system does all the symbol handling. Here is an example:

bash$ ${CROSS_COMPILE}gcc -Wall -g -o hello hello.c
bash$ cp -p hello <directory-shared-with-target>/hello-stripped
bash$ ${CROSS_COMPILE}strip <directory-shared-with-target>/hello-stripped

To use the server, you must tell it how to communicate with GDB, the name of your program, and the arguments for your program. To start a debugging session via network type on the target:

bash$ cd <directory-shared-with-host>
bash$ gdbserver 192.168.1.1:12345 hello-stripped
Process hello-stripped created; pid = 353

And then on the host:

bash$ ${CROSS_COMPILE}gdb hello
...
(gdb) set solib-absolute-prefix /opt/eldk/$CROSS_COMPILE
(gdb) dir /opt/eldk/$CROSS_COMPILE
Source directories searched:
/opt/eldk/$CROSS_COMPILE:$cdir:$cwd
(gdb) target remote 192.168.1.99:12345
Remote debugging using 192.168.1.99:12345
0x30012748 in ?? ()
...
(gdb) l
1       #include <stdio.h>
2
3       int main(int argc, char* argv[])
4       {
5               printf ("Hello world\n");
6               return 0;
7       }
(gdb) break 5
Breakpoint 1 at 0x10000498: file hello.c, line 5.
(gdb) continue
Continuing.

Breakpoint 1, main (argc=1, argv=0x7ffffbe4) at hello.c:5
5               printf ("Hello world\n");
(gdb) p argc
$1 = 1
(gdb) continue
Continuing.

Program exited normally.

If the target program you want to debug is linked against shared libraries, you must tell GDB where the proper target libraries are located. This is done using the set solib-absolute-prefix GDB command. If this command is omitted, then, apparently, GDB loads the host versions of the libraries and gets crazy because of that.

10.6. Debugging with Graphical User Interfaces

It is convenient to use DDD, a Graphical User Interface to GDB, for debugging as it allows to define and execute frequently used commands via buttons. You can start DDD with the command:

bash$ ddd --debugger ${CROSS_COMPILE}gdb &

If DDD is not already installed on your Linux system, have a look at your distribution media.

11. Simple Embedded Linux Framework

12. Books, Mailing Lists, Links, etc.

This section provides references on where to find more information

Contents:

12.1. Application Notes

A collection of Application Notes relevant for embedded computing can be found on the DENX web server.

12.2. Further Reading

12.2.1. License Issues

Articles

• http://www.gnu.org/licenses/gpl-2.0.html: GNU General Public License, version 2
• http://www.gnu.org/copyleft/gpl.html: GNU General Public License
• http://www.softwarefreedom.org/resources/2008/compliance-guide.html: A Practical Guide to GPL Compliance
• http://article.gmane.org/gmane.comp.video.dri.devel/52751 Alan Cox about combining GPL device drivers with closed source user space libraries

12.2.2. Linux kernel

Books

• Karim Yaghmour, Jon Masters, Gilad Ben-Yossef, Philippe Gerum: "Building Embedded Linux Systems 2nd edition",
  Paperback: 462 pages, O'Reilly & Associates; (August 2008); ISBN 10: 0-596-52968-6; ISBN 13: 9780596529680 ISBN 059600222X - IMHO the best book about Embedded Linux so far. An absolute must have.
• Greg Kroah-Hartman: "Linux Kernel in a Nutshell",
  198 pages, O'Reilly ("In Nutshell" series), (December 2006), ISBN 10: 0-596-10079-5; ISBN 13: 9780596100797
  - Tarball of PDF files (3 MB):
  http://www.kernel.org/pub/linux/kernel/people/gregkh/lkn/lkn_pdf.tar.bz2
  - Tarball of DocBook files (1 MB):
  http://www.kernel.org/pub/linux/kernel/people/gregkh/lkn/lkn_xml.tar.bz2
• Craig Hollabaugh: "Embedded Linux: Hardware, Software, and Interfacing",
  Paperback: 432 pages; Addison Wesley Professional; (March 7, 2002); ISBN 0672322269
• Christopher Hallinan: "Embedded Linux Primer: A Practical Real-World Approach",
  576 pages, Prentice Hall, September 2006, ISBN-10: 0-13-167984-8; ISBN-13: 978-0-13-167984-9
• Jonathan Corbet, Alessandro Rubini, Greg Kroah-Hartman: "Linux Device Drivers", 3rd Edition ;
  Paperback: 636 pages; O'Reilly & Associates; 3rd edition (February 2005); ISBN: 0-596-00590-31 - The reference book for writing Linux device drivers. An absolute must have. => Read online
• Jürgen Quade, Eva-Katharina Kunst: "Linux-Treiber entwickeln"; Broschur: 436 pages; dpunkt.verlag, Juni 2004; ISBN 3898642380
  - focused on kernel 2.6, unfortunately German only
  - => Read online
• Sreekrishnan Venkateswaran: "Essential Linux Device Drivers",
  744 pages, Prentice Hall, March 2008, ISBN-10: 0-13-239655-6; ISBN-13: 978-0-13-239655-4
  - => Read online

Articles

• The Linux Kernel - describing most aspects of the Linux Kernel. Probably, the first reference for beginners. Lots of illustrations explaining data structures use and relationships. In short: a must have.
• Linux Kernel Module Programming Guide - Very nice 92 pages GPL book on the topic of modules programming. Lots of examples.
• LWN: Porting device drivers to the 2.6 kernel - Series of articles (37) in Linux Weekly News: http://lwn.net/Articles/driver-porting/
• MIPS Linux Porting Guide: http://linux.junsun.net/porting-howto/porting-howto.html
• Andries Brouwers remarks to the linux kernel: http://www.win.tue.nl/~aeb/linux/lk/lk.html

12.2.3. General Linux / Unix programming

Books

• W. Richard Stevens: "Advanced Programming in the UNIX Environment", Addision Wesley, ISBN 0-201-56317-7
• Eric S. Raymond: "The Art of Unix Programming", Addision Wesley, ISBN 0131429019 => Read online. If you don't want to read the whole book then at least look at the Basics of the Unix philosophy condensing lots of experience into a few rules. This is essential reading.
• David R. Butenhof: "Programming with POSIX Threads", Addision Wesley, ISBN 0-201-63392-2.
• Bradford Nichols, Dick Buttlar and Jacqueline Proulx Farrell: "Pthreads Programming", O'Reilly & Associates
• "Git Community Book"
  See http://book.git-scm.com/ or download the PDF version.

Articles

• The GNU C Library: http://www.linuxselfhelp.com/gnu/glibc/html_chapter/libc_toc.html
  General Linux Programming: http://www.linuxselfhelp.com/cats/programming.html
• Multi-Threaded Programming With POSIX Threads:
  http://users.actcom.co.il/~choo/lupg/tutorials/multi-thread/multi-thread.html
• Brad Hards: The Linux USB Input Subsystem, Part I
  http://www.linuxjournal.com/article/6396
• Brad Hards: Using the Input Subsystem, Part II
  http://www.linuxjournal.com/article/6429
• Ulrich Drepper: Position Independent Binaries: "Text Relocations"
  http://people.redhat.com/drepper/textrelocs.html
• Ulrich Drepper: "How to Write Shared Libraries"
  http://people.redhat.com/drepper/dsohowto.pdf
• Ulrich Drepper: "What Every Programmer Should Know About Memory"
  http://people.redhat.com/drepper/cpumemory.pdf
• David Goldberg: "What Every Computer Scientist Should Know About Floating-Point Arithmetic"
  http://www.physics.ohio-state.edu/~dws/grouplinks/floating_point_math.pdf
• More Ulrich Drepper stuff: http://people.redhat.com/drepper/
• How to optimize DSOs by identifying unused non-exported functions and data.
  http://blog.flameeyes.eu/articles/2008/01/17/today-how-to-identify-unused-exported-functions-and-variables
• A quite complete history of the UNIX family can be found here: http://www.levenez.com/unix/
• Unix Manual, first edition, 3 November 1971: http://cm.bell-labs.com/cm/cs/who/dmr/1stEdman.html
• John Graham-Cumming: Debugging Makefiles
  http://newsletter.embedded.com/cgi-bin4/DM/y/e4Kd0G1ErD0FrY0E3FS0EZ
• Binutils / ld documentation: Linker Scripts
• git ready - learn one git command at a time: http://gitready.com/

Standards:

• POSIX.1-2008, IEEE Std 1003.1™-2008, The Open Group Technical Standard Base Specifications, Issue 7: http://pubs.opengroup.org/onlinepubs/9699919799/mindex.html
• Linux Standard Base: http://refspecs.freestandards.org/lsb.shtml
• Single UNIX Specification, Version 3 (needs registration even for online viewing)
• Single UNIX Specification, Version 2
• PCI Bus Bindings - Standard for Boot Firmware:
  http://playground.sun.com/1275/bindings/pci/pci2_1.pdf
• International standardization working group for the programming language C: http://www.open-std.org/jtc1/sc22/WG14/

12.2.4. Network Programming

Books

• W. Richard Stevens: "TCP/IP Illustrated, Volume 1 - The Protocols", Addision Wesley, ISBN 0-201-63346-9
• Gary R. Wright, W. Richard Stevens: "TCP/IP Illustrated, Volume 2 - The Implementation", Addision Wesley, ISBN 0-201-63354-X
• W. Richard Stevens: "TCP/IP Illustrated, Volume 3 - TCP for Transactions", Addision Wesley, ISBN 0-201-63495-3
• W. Richard Stevens: "UNIX Network Programming, Volume 1 - Networking APIs: Sockets and XTI", 2nd ed., Prentice Hall, ISBN-0-13-490012-X
• W. Richard Stevens: "UNIX Network Programming, Volume 2 - Interprocess Communication", 2nd ed., Prentice Hall, ISBN-0-13-081081-9

Articles

• Linux Networking topics (like NAPI, GSO, VLAN, IPsec etc.):
  http://linux-net.osdl.org/index.php/Main_Page

12.2.5. C++ programming

Books

• Scott Meyers: "Effective C++: 55 Specific Ways to Improve Your Programs and Designs (3rd Edition)", Addison-Wesley, May 20, 2005, ISBN: 0321334876

12.2.6. Java programming

Books

• Joshua Bloch: "Effective Java -- Programming Language Guide", 2001, Addison Wesley, ISBN 0-201-31005-8, 250 pages

12.2.7. Internationalization And Character Sets

Articles

• Wikipedia article on UTF-8: http://en.wikipedia.org/wiki/UTF-8
• "The Absolute Minimum Every Software Developer Absolutely, Positively Must Know About Unicode and Character Sets (No Excuses!)" by Joel Spolsky: http://www.joelonsoftware.com/articles/Unicode.html
• "UTF-8 and Unicode FAQ for Unix/Linux" by Markus Kuhn http://www.cl.cam.ac.uk/~mgk25/unicode.html
• Understanding Encodings: http://foswiki.org/Development/UnderstandingEncodings
  Detailed primer on character sets and encodings, especially in the context of web design

12.2.8. ARM Architecture Programming

12.2.9. Power Architecture® Programming

12.2.10. Embedded Topics

Articles

• Things you always wanted to know about NAND flash but never dared to ask: Micron Application Note
• The ultimate goal of Embedded C++ is to provide embedded systems programmers with a subset of C++ that is easy for the average C programmer to understand and use.
• Our contribution to the Darwin year 2009: Hardware designs that will not replicate: Topic in DENX Wiki

12.3. Mailing Lists

These are some mailing lists of interest. If you are new to mailing lists then please take the time to read at least RFC 1855.

• linux-arm-kernel - Communications among developers and users of Linux on arm boards
• linuxppc-embedded - Communications among developers and users of Linux on embedded Power Architecture® boards
  This mailing list has been merged into the linuxppc-dev mailing list below and thus does not exist anymore.
• linuxppc-dev - Communications among active developers of Linux on 32 bit Power Architecture® plattforms. Not intended for user support.
• u-boot - Support for "U-Boot" Universal Bootloader
• ELDK - Support for DENX Embedded Linux Development Kit

12.4. Links

Linux Kernel Resources:

• The Linux Documentation Project : http://www.tldp.org/
• Generic ("official") Linux Kernel sources:
  git: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=tree
  FTP: ftp://ftp.kernel.org/pub/linux/kernel/v2.6/
• Full git history of Linux: http://thread.gmane.org/gmane.linux.kernel/690811
• Generic kernel sources for Power Architecture™ systems: http://penguinppc.org/dev/kernel.shtml
• DENX kernel sources: http://git.denx.de/?p=linux-denx.git;a=summary
• Cross-Referencing the Linx Kernel: http://lxr.linux.no/source/?a=ppc
• Starting point for Linux based asm (mostly x86): http://linuxassembly.org/

Realtime, Xenomai, RTAI:

• Xenomai Home Page: http://www.xenomai.org/
• Hackbench, a commonly used system stress tool
• Calibrator, determines cache
• RTAI Home Page: http://www.rtai.org/
• DENX RTAI Patches: ftp://ftp.denx.de/pub/RTAI/ sizes at runtime, rendering it another useful system stress tool

U-Boot:

• U-Boot Project Page: http://www.denx.de/wiki/U-Boot/WebHome.
  Note that the old SourceForge page is not maintained anymore.
• DENX U-Boot and Linux Guide: http://www.denx.de/twiki/bin/view/DULG

Cross Development Tools:

• DENX Embedded Linux Development Kit: http://www.denx.de/twiki/bin/view/DULG/ELDK

Miscalleneous or unsorted material:

• BDI2000 List of supported Flash Memories: This document not only lists the currently supported flash chips, but also the required settings in the BDI config file.
• BDI2000 configuration files: ftp://78.31.64.234/bdigdb/config/

12.5. Tools

• http://lxr.linux.no/source/ - Cross-Referencing the Linux Kernel - using a versatile hypertext cross-referencing tool for the Linux Kernel source tree (the Linux Cross-Reference project)
• ftp://ftp.denx.de/pub/tools/backtrace - Decode Stack Backtrace - Perl script to decode the Stack Backtrace printed by the Linux Kernel when it panics
• ftp://ftp.denx.de/pub/tools/clone_tree - "Clone" a Source Tree - Perl script to create a working copy of a source tree (for example the Linux Kernel) which contains mainly symbolic links (and automagically omits "unwanted" files like CVS repository data, etc.)

13. Appendix

13.1. Flat Device Tree

[marex@pollux]$ /*
 * Copyright (C) 2012 Marek Vasut <marex@denx.de>
 *
 * The code contained herein is licensed under the GNU General Public
 * License. You may obtain a copy of the GNU General Public License
 * Version 2 or later at the following locations:
 *
 * http://www.opensource.org/licenses/gpl-license.html
 * http://www.gnu.org/copyleft/gpl.html
 */

/dts-v1/;
/include/ "imx28.dtsi"

/ {
        model = "DENX M28EVK";
        compatible = "denx,m28evk", "fsl,imx28";

        memory {
                reg = <0x40000000 0x08000000>;
        };

        apb@80000000 {
                apbh@80000000 {
                        gpmi-nand@8000c000 {
                                #address-cells = <1>;
                                #size-cells = <1>;
                                pinctrl-names = "default";
                                pinctrl-0 = <&gpmi_pins_a &gpmi_status_cfg>;
                                status = "okay";

                                partition@0 {
                                        label = "bootloader";
                                        reg = <0x00000000 0x00300000>;
                                        read-only;
                                };

                                partition@1 {
                                        label = "environment";
                                        reg = <0x00300000 0x00080000>;
                                };

                                partition@2 {
                                        label = "redundant-environment";
                                        reg = <0x00380000 0x00080000>;
                                };

                                partition@3 {
                                        label = "kernel";
                                        reg = <0x00400000 0x00400000>;
                                };

                                partition@4 {
                                        label = "filesystem";
                                        reg = <0x00800000 0x0f800000>;
                                };
                        };

                        ssp0: ssp@80010000 {
                                compatible = "fsl,imx28-mmc";
                                pinctrl-names = "default";
                                pinctrl-0 = <&mmc0_8bit_pins_a
                                             &mmc0_cd_cfg
                                             &mmc0_sck_cfg>;
                                bus-width = <8>;
                                wp-gpios = <&gpio3 10 0>;
                                vmmc-supply = <&reg_vddio_sd0>;
                                status = "okay";
                        };

                        ssp2: ssp@80014000 {
                                #address-cells = <1>;
                                #size-cells = <0>;
                                compatible = "fsl,imx28-spi";
                                pinctrl-names = "default";
                                pinctrl-0 = <&spi2_pins_a>;
                                status = "okay";

                                flash: m25p80@0 {
                                        #address-cells = <1>;
                                        #size-cells = <1>;
                                        compatible = "m25p80";
                                        spi-max-frequency = <40000000>;
                                        m25p,fast-read;
                                        reg = <0>;
                                };
                        };

                        pinctrl@80018000 {
                                pinctrl-names = "default";
                                pinctrl-0 = <&hog_pins_a>;

                                hog_pins_a: hog@0 {
                                        reg = <0>;
                                        fsl,pinmux-ids = <
                                                0x31c3 /* MX28_PAD_PWM3__GPIO_3_28 */
                                                0x30a3 /* MX28_PAD_AUART2_CTS__GPIO_3_10 */
                                                0x30b3 /* MX28_PAD_AUART2_RTS__GPIO_3_11 */
                                                0x30c3 /* MX28_PAD_AUART3_RX__GPIO_3_12 */
                                                0x30d3 /* MX28_PAD_AUART3_TX__GPIO_3_13 */
                                        >;
                                        fsl,drive-strength = <0>;
                                        fsl,voltage = <1>;
                                        fsl,pull-up = <0>;
                                };

                                lcdif_pins_m28: lcdif-m28@0 {
                                        reg = <0>;
                                        fsl,pinmux-ids = <
                                                0x11e0 /* MX28_PAD_LCD_DOTCLK__LCD_DOTCLK */
                                                0x11f0 /* MX28_PAD_LCD_ENABLE__LCD_ENABLE */
                                        >;
                                        fsl,drive-strength = <0>;
                                        fsl,voltage = <1>;
                                        fsl,pull-up = <0>;
                                };
                        };

                        lcdif@80030000 {
                                pinctrl-names = "default";
                                pinctrl-0 = <&lcdif_24bit_pins_a
                                             &lcdif_pins_m28>;
                                status = "okay";
                        };

                        can0: can@80032000 {
                                pinctrl-names = "default";
                                pinctrl-0 = <&can0_pins_a>;
                                status = "okay";
                        };

                        can1: can@80034000 {
                                pinctrl-names = "default";
                                pinctrl-0 = <&can1_pins_a>;
                                status = "okay";
                        };
                };

                apbx@80040000 {
                        saif0: saif@80042000 {
                                pinctrl-names = "default";
                                pinctrl-0 = <&saif0_pins_a>;
                                status = "okay";
                        };

                        saif1: saif@80046000 {
                                pinctrl-names = "default";
                                pinctrl-0 = <&saif1_pins_a>;
                                fsl,saif-master = <&saif0>;
                                status = "okay";
                        };

                        i2c0: i2c@80058000 {
                                pinctrl-names = "default";
                                pinctrl-0 = <&i2c0_pins_a>;
                                clock-frequency = <400000>;
                                status = "okay";

                                sgtl5000: codec@0a {
                                        compatible = "fsl,sgtl5000";
                                        reg = <0x0a>;
                                        VDDA-supply = <&reg_3p3v>;
                                        VDDIO-supply = <&reg_3p3v>;

                                };

                                eeprom: eeprom@51 {
                                        compatible = "atmel,24c128";
                                        reg = <0x51>;
                                        pagesize = <32>;
                                };

                                rtc: rtc@68 {
                                        compatible = "stm,mt41t62";
                                        reg = <0x68>;
                                };
                        };

                        lradc@80050000 {
                                status = "okay";
                        };

                        duart: serial@80074000 {
                                pinctrl-names = "default";
                                pinctrl-0 = <&duart_pins_a>;
                                status = "okay";
                        };

                        usbphy0: usbphy@8007c000 {
                                status = "okay";
                        };

                        usbphy1: usbphy@8007e000 {
                                status = "okay";
                        };

                        auart0: serial@8006a000 {
                                pinctrl-names = "default";
                                pinctrl-0 = <&auart0_2pins_a>;
                                status = "okay";
                        };
                };
        };

        ahb@80080000 {
                usb0: usb@80080000 {
                        vbus-supply = <&reg_usb0_vbus>;
                        pinctrl-names = "default";
                        pinctrl-0 = <&usbphy0_pins_a>;
                        status = "okay";
                };

                usb1: usb@80090000 {
                        vbus-supply = <&reg_usb1_vbus>;
                        pinctrl-names = "default";
                        pinctrl-0 = <&usbphy1_pins_a>;
                        status = "okay";
                };

                mac0: ethernet@800f0000 {
                        phy-mode = "rmii";
                        pinctrl-names = "default";
                        pinctrl-0 = <&mac0_pins_a>;
                        status = "okay";
                };

                mac1: ethernet@800f4000 {
                        phy-mode = "rmii";
                        pinctrl-names = "default";
                        pinctrl-0 = <&mac1_pins_a>;
                        status = "okay";
                };
        };

        regulators {
                compatible = "simple-bus";

                reg_3p3v: 3p3v {
                        compatible = "regulator-fixed";
                        regulator-name = "3P3V";
                        regulator-min-microvolt = <3300000>;
                        regulator-max-microvolt = <3300000>;
                        regulator-always-on;
                };

                reg_vddio_sd0: vddio-sd0 {
                        compatible = "regulator-fixed";
                        regulator-name = "vddio-sd0";
                        regulator-min-microvolt = <3300000>;
                        regulator-max-microvolt = <3300000>;
                        gpio = <&gpio3 28 0>;
                };

                reg_usb0_vbus: usb0_vbus {
                        compatible = "regulator-fixed";
                        regulator-name = "usb0_vbus";
                        regulator-min-microvolt = <5000000>;
                        regulator-max-microvolt = <5000000>;
                        gpio = <&gpio3 12 0>;
                };

                reg_usb1_vbus: usb1_vbus {
                        compatible = "regulator-fixed";
                        regulator-name = "usb1_vbus";
                        regulator-min-microvolt = <5000000>;
                        regulator-max-microvolt = <5000000>;
                        gpio = <&gpio3 13 0>;
                };
        };

        sound {
                compatible = "denx,m28evk-sgtl5000",
                             "fsl,mxs-audio-sgtl5000";
                model = "m28evk-sgtl5000";
                saif-controllers = <&saif0 &saif1>;
                audio-codec = <&sgtl5000>;
        };
};

13.2. BDI2000 Configuration file

[marex@pollux]$ [INIT]
;WM32   0x800401E0 0x00000002   ; Restart the chip altogether (HW_CLKCTRL_RESET = CHIP)
;DELAY  1000

;WM32 0x80056050 0x1
;WM32 0x80056000 0x10
;DELAY 100
WREG    CPSR    0x000000D3      ; Configure CPSR

[TARGET]
CPUTYPE         ARM926E
CLOCK           0 8                     ;JTAG clock (0=Adaptive, 1=8MHz, 2=4MHz, 3=2MHz)
;CLOCK          4                       ;JTAG clock (0=Adaptive, 1=8MHz, 2=4MHz, 3=2MHz)
WAKEUP          500             ;because of slow rising reset line
RESET           HARD 500        ;because of heavy capacitive load on reset line
TRST            PUSHPULL
ENDIAN          LITTLE          ;memory model (LITTLE | BIG)
BREAKMODE       HARD            ;SOFT or HARD, ARM / Thumb break code
SCANSUCC        0 0             ; 1 4 when the ETMBUF after the ARM926 core is enabled via TESTMODE pin
STARTUP         RESET           ; the ARM 926 core is stop after reset
;BDIMODE                LOADONLY

[HOST]
IP              192.168.1.1
PROMPT          M28>

[FLASH]
; only nand and this is not supported directly from the bdi

[REGS]
FILE            BDI2000/reg926e.def

14. FAQ - Frequently Asked Questions

This is a collection of questions which came up repeatedly. Give me more feedback and I will add more stuff here.

The items are categorized whether they concern U-Boot itself, the Linux kernel or the SELF framework.

14.1. ELDK

14.1.1. ELDK Installation under FreeBSD

Question:

How can I install ELDK on a FreeBSD system?

Answer:

[Thanks to Rafal Jaworowski for these detailed instructions.] This is a short tutorial how to host ELDK on FreeBSD 5.x and 6.x. The procedure described below was tested on 5.2.1, 5.3 and 6-current releases; we assume the reader is equipped with the ELDK 3.x CDROM or ISO image for installation, and is familiar with FreeBSD basic administration tasks like ports/packages installation.

1. Prerequisites:
   1. Install linux_base
      
      The first step is to install the Linux compatibility layer from ports /usr/ports/emulators/linux_base/ or packages ftp://ftp.freebsd.org/pub/FreeBSD/ports/i386/packages/emulators/
      
      Please make sure to install version 7.1_5 (linux_base-7.1_5.tbz) or later; in particular, version 6.1.5 which can also be found in the ports tree does not work properly!
      
      The compatibility layer is activated by

      # kldload linux
      

   2. Install bash
      
      Since ELDK and Linux build scripts are organised around bash while FreeBSD does not have it in base, this shell needs to be installed either from ports /usr/ports/shells/bash2/ or packages collection ftp://ftp.freebsd.org/pub/FreeBSD/ports/i386/packages/shells/
      
      The installation puts the bash binary in /usr/local/bin. It is a good idea to create a symlink in /bin so that hash bang from scripts (#!/bin/bash) works without modifications:

      # cd /bin 
      # ln -s /usr/local/bin/bash
      

2. Prepare ELDK
   
   This step is only needed for ELDK release 3.1 and older versions.
   
   Copy the install files from the CDROM or ISO image to a writable location. Brand the ELDK installer as Linux ELF file:

   # cd <elkd_install_dir>
   # brandelf -t Linux ./install
   

   Note: The following workaround might be a good alternative for the tedious copying of the installation CDROM to a writable location and manual branding: you can set a fallback branding in FreeBSD - when the loader cannot recognise the ELF brand it will switch to the last resort defined.

   # sysctl -w kern.elf32.fallback_brand=3
   kern.elf32.fallback_brand: -1 -> 3
   

   With this setting, the normal ELDK CDROM images should work.
3. Install ELDK normally as described in 3.5.3. Initial Installation
4. Set envrionment variables and PATH as needed for ELDK (in bash); for example:

   bash$ export CROSS_COMPILE=ppc_8xx-
   bash$ export PATH=${PATH}:/opt/eldk/bin:/opt/eldk/usr/bin
   

5. Hints for building U-Boot:
   
   FreeBSD normally uses BSD-style 'make' in base, but in order to compile U-Boot 'gmake' (GNU make) has to be used; this is installed as part of the 'linux_base' package (see above).
   
   U-Boot should build according to standard ELDK instructions, for example:

   bash$ cd /opt/eldk/ppc_8xx/usr/src/u-boot-1.1.2
   bash$ gmake TQM823L_config
   bash$ gmake all
   

6. Hints for building Linux:
   
   There are three issues with the Makefile in the Linux kernel source tree:
   • GNU make has to be used.
   • The 'expr' utility in FreeBSD base behaves differently from the version than is used in Linux so we need to modify the Makefile to explicitly use the Linux version (which is part of the Linux compatibility package). This is best achieved with defining "EXPR = /compat/linux/usr/bin/expr" somewhere at =Makefile='s beginning and replacing all references to 'expr' with the variable ${EXPR).
   • Some build steps (like when running 'scripts/mkdep' can generate very long arguments lists (especially is the Linux kernel tree is in a directory with long absolute filenames). A solution is to use xargs to split such long commands into several with shorter argument lists.
     
     The Linux kernel can then be built following the standard instructions, for example:

     bash$ cd /opt/eldk/ppc_8xx/usr/src/linux-2.4.25/
     bash$ gmake mrproper
     bash$ gmake TQM823L_config
     bash$ gmake oldconfig
     bash$ gmake dep
     bash$ gmake -j6 uImage
     

ELDK Installation Hangs

Question:

I try to install the ELDK on a Linux PC, and the installation hangs. It starts fine, but then it freezes like this:

...
Preparing...                ########################################### [100%]
   1:db4-devel-ppc_4xx      ########################################### [100%]
Preparing...                ########################################### [100%]
   1:db4-utils-ppc_4xx      ########################################### [100%]
Preparing...                ########################################### [100%]
   1:glib2-ppc_4xx          ########################################### [100%]
Preparing...                ########################################### [100%]
   1:glib2-devel-ppc_4xx    ########################################### [100%]
Preparing...                ########################################### [100%]

<hangs here>

Answer:

This is almost certainly a FUTEX problem. To verify this, please wait until the process grinds to a halt, then use ps to find the pid of the "rpm" process that was started by the "install" program (use "ps -axf" which gives you a nice hierarchy, look for the "install" process, then for "rpm") and then attach to it with "strace -p". Most probably you will see the something like this:

# strace -p 21197
Process 21197 attached - interrupt to quit
futex(0x96fe17c, FUTEX_WAIT_PRIVATE, 1, NULL

i. e. the process is hanging in a futex call.
We have seen this more than once with differing Linux systems, but unfortunately we don't know a clean and reliable way to fix it yet. We suspect that it is a kernel/libc combination problem because it usually went away usually after changing the exact used kernel version.
The only workaround we can recommend so far is to update your host system and install more recent versions of the Linux kernel and/or the glibc C library (assuming such are available for your Linux distribution; if not, falling back to a previous kernel version may help, too).
Note: This is only needed for the installer, the problem does not happen with the regular use of the ELDK.

14.1.3. .gvfs: Permission Denied

Question:

When trying to install the ELDK, I get error messages like this for each and every package that gets installed:

Preparing...   ################### 100%
1: rpm...   ################### 100%
Error: Failed to stat /home/wd/.gvfs: Permission Denied

This happens even though I run the installer as root.

Answer:

Even though flagged as an error, these messages are harmless warnings that can be safely ignored. Before the RPM tool starts to install a package, it checks if there is sufficient space for it in the file system. Unfortunately it is dumb and checks all mounted file systems for space, but the permissions of the ".gvfs" directory (the mount point for the Gnome Virtual File System) do not permit this.

Note:

Actually the messages are not printed despite the fact that you are running as root, but because you run as root. You have permissions to check the "$HOME/.gvfs" directory, while root gets an error:

$ df -h /home/wd/.gvfs
Filesystem            Size  Used Avail Use% Mounted on
gvfs-fuse-daemon         0     0     0   -  /home/wd/.gvfs
$ sudo df -h /home/wd/.gvfs
df: `/home/wd/.gvfs': Permission denied
df: no file systems processed

14.1.4. Installation on Local Harddisk

Question:

I have a local harddisk drive connected to my target board. Can I install the ELDK on it and run it like a standard Linux distribution?

Answer:

Yes, this is possible. It requires only minor adjustments. The following example assumes you are using a SCSI disk drive, but the same can be done with standard SATA or PATA drives, too:

1. Boot the target with root file system over NFS.
2. Create the necessary partitions on your disk drive: you need at last a swap partition and a file system partition.

   bash-3.00# fdisk -l 
   
   Disk /dev/sda: 36.9 GB, 36951490048 bytes
   64 heads, 32 sectors/track, 35239 cylinders
   Units = cylinders of 2048 * 512 = 1048576 bytes
   
      Device Boot      Start         End      Blocks   Id  System
   /dev/sda1               1         978     1001456   82  Linux swap / Solaris
   /dev/sda2             979       12423    11719680   83  Linux
   /dev/sda3           12424       23868    11719680   83  Linux
   /dev/sda4           23869       35239    11643904   83  Linux
   

3. Format the partititons:

   bash-3.00# mkswap /dev/sda1
   bash-3.00# mke2fs -j -m1 /dev/sda2
   

4. Mount the file system:

   bash-3.00# mount /dev/sda2 /mnt
   

5. Copy the content of the (NFS) root file system into the mounted file system:

   bash-3.00# tar --one-file-system -c -f - / | ( cd /mnt ; tar xpf - )
   

6. Adjust /etc/fstab for the disk file system:

   bash-3.00# vi /mnt/etc/fstab
   bash-3.00# cat /mnt/etc/fstab
   /dev/sda2       /               ext3    defaults        1 1
   /dev/sda1       swap            swap    defaults        0 0
   proc            /proc           proc    defaults        0 0
   sysfs           /sys            sysfs   defaults        0 0
   

7. Adjust /etc/rc.sysinit for running from local disk; remove the following comments:

   bash-3.00# diff -u /mnt/etc/rc.sysinit.ORIG /mnt/etc/rc.sysinit
   --- /mnt/etc/rc.sysinit.ORIG    2007-01-21 04:37:00.000000000 +0100
   +++ /mnt/etc/rc.sysinit 2007-03-02 10:58:22.000000000 +0100
   @@ -460,9 +460,9 @@
    
    # Remount the root filesystem read-write.
    update_boot_stage RCmountfs
   -#state=`LC_ALL=C awk '/ \/ / && ($3 !~ /rootfs/) { print $4 }' /proc/mounts`
   -#[ "$state" != "rw" -a "$READONLY" != "yes" ] && \
   -#  action $"Remounting root filesystem in read-write mode: " mount -n -o remount,rw /
   +state=`LC_ALL=C awk '/ \/ / && ($3 !~ /rootfs/) { print $4 }' /proc/mounts`
   +[ "$state" != "rw" -a "$READONLY" != "yes" ] && \
   +  action $"Remounting root filesystem in read-write mode: " mount -n -o remount,rw /
    
    # Clean up SELinux labels
    if [ -n "$SELINUX" ]; then
   

8. Unmount disk:

   bash-3.00# umount /mnt
   

9. Reboot, and adjust boot arguments to use disk partition as root file system

   => setenv diskargs setenv bootargs root=/dev/sda2 ro
   => setenv net_disk 'tftp ${loadaddr} ${bootfile};run diskargs addip addcons;bootm'
   => saveenv
   

10. Boot with these settings

    => run net_disk
    

14.1.5. System Include Files Missing

Question:

when installing ELDK on Ubuntu 6.06 dapper drake I get the following error messages....

Preparing...                ########################################### [100%]
   1:kernel-source-ppc_6xx  ########################################### [100%]
Configuring kernel...
scripts/basic/fixdep.c:107:23: error: sys/types.h: No such file or directory
scripts/basic/fixdep.c:108:22: error: sys/stat.h: No such file or directory
scripts/basic/fixdep.c:109:22: error: sys/mman.h: No such file or directory
...

Answer:

The Linux installation on your host is missing essential files that are needed to perform software development and use a C compiler. On Ubuntu, check for example if you miss a "libc6-dev" package. The specific package name differs from distribution to distribution; on Fedora, you need for example the "glibc-headers" package.

If you want to work with a Linux kernel you will probably also need other packages.

14.1.6. patch: command not found

Question:

When installing ELDK on Ubuntu I get the following error message:

...prepare-kernel.sh: line 376: patch: command not found

Answer:

The error message contains clear hints for the solution: the "patch" command cannot be found on your system, so most probably it has not been installed yet. Please try:

$ sudo apt-get install patch

14.1.7. ELDK Include Files Missing

Question:

After configuring and compiling a Linux kernel in the kernel source tree that comes with the ELDK, I cannot compile user space programs any more - I get error messages because many #include file like <errno.h> etc. are missing.
This is with ELDK 4.0 or 4.1.

Answer:

This problem is caused by the way how the ELDK is packaged. At the moment, the ELDK kernel headers are not packed into a separate "kernel-headers" RPM to avoid duplication, because the kernel source tree is always installed. Instead, the ELDK "kernel-headers" package is just a set of symlinks. This worked fine in the past, but fails with the new support for ARCH=powerpc systems.
The next version of the ELDK will contain a real kernel-headers RPM, which will fix this problem.
As a workaround on current systems, you can install the real kernel include files into the "include/asm", "include/linux" and "include/mtd" directories.
To do this, the following commands can be used:

bash$ <eldkroot>/bin/rpm -e kernel-headers-ppc_<target>
bash$ cd <eldkroot>/ppc_<target>
bash$ rm usr/include/asm
bash$ tar -xvzf kernel-headers-powerpc.tar.gz

The tarball mentioned above can be downloaded here. It contains the include files that get installed by running the "make ARCH=powerpc headers_install" command in the Linux kernel tree.

This problem is fixed in ELDK 4.2 and later releases.

14.1.8. Using the ELDK on a 64 bit platform

As the ELDK is compiled for 32-bit host systems, a compatibility layer is required on 64-bit systems. This package is usually called ia32-libs. So on a Debian or Ubuntu system a

sudo apt-get install ia32-libs

should be enough to make the ELDK work.

On the U-Boot mailing list, it was reported that for a 64 bit Fedora 11 the following should be enough:

sudo yum -y install glibc.i686 zlib.i686

14.1.9. How can I check if Floating Point support is working?

Question:

The floating point performance of my P2020 QorIQ processor is really poor. I am not using the ELDK, but a tool chain from FOOBAR. Can this be a problem? What can I do to verify this?

Answer:

The P20xx QorIQ processors use an e500v2 core which does not include a normal Floating Point Unit (FPU), but instead a Signal Processing Engine (SPE Version 2). You can run FP calculations on the SPE, but there are no special FP registers available as on a normal FPU, so General Purpose Registers must be used for passing of FP operands. While this is still much faster than pure soft-float emulation, it is missing the advantages and the speed of a full-blown, separate standard FPU with a full FP register set.

Also, your tool chain needs to be aware of this feature, and must contain support for it. Eventually special compiler options are needed - check your documentation.

With the ELDK, the needed settings are automatically pre-set when you just chose the correct target architecture packages, cf. 3.4. Supported Target Architectures

To test what your tool chain is doing, you best compile a smal test program and check the generated code. The following examples were done with ELDK 4.2 for Power Architecture® targets:

1. Test Program:

   $ cat fp_test.c
   double foo (double x, double y)
   {
           double z;
   
           z = (x + y) / (x * y);
   
           return z;
   }
   

2. Build for normal FPU support (using the ppc_6xx target architecture):

   $ export CROSS_COMPILE=ppc_6xx-
   $ ppc_6xx-gcc -S -O fp_test.c
   

   Check results:

   $ cat fp_test.s
           .file   "fp_test.c"
           .section        ".text"
           .align 2
           .globl foo
           .type   foo, @function
   foo:
           fadd 0,1,2
           fmul 1,1,2
           fdiv 1,0,1
           blr
           .size   foo, .-foo
           .ident  "GCC: (GNU) 4.2.2"
           .section        .note.GNU-stack,"",@progbits
   

   The use of floating point machine instructions ("fadd", "fmul", "fdiv") and the fact that no additional register use is needed is a clear indication that full support for the hardware FPU is available in this configuration.
3. Build for soft-float emulation (using the ppc_8xx target architecure):

   $ export CROSS_COMPILE=ppc_8xx-
   $ ppc_8xx-gcc -S -O fp_test.c 
   

   $ cat fp_test.s
           .file   "fp_test.c"
           .globl __adddf3
           .globl __muldf3
           .globl __divdf3
           .section        ".text"
           .align 2
           .globl foo
           .type   foo, @function
   foo:
           stwu 1,-48(1)
           mflr 0
           stw 24,16(1)
           stw 25,20(1)
           stw 26,24(1)
           stw 27,28(1)
           stw 28,32(1)
           stw 29,36(1)
           stw 0,52(1)
           mr 28,3
           mr 29,4
           mr 26,5
           mr 27,6
           bl __adddf3
           mr 24,3
           mr 25,4
           mr 3,28
           mr 4,29
           mr 5,26
           mr 6,27
           bl __muldf3
           mr 5,3
           mr 6,4
           mr 3,24
           mr 4,25
           bl __divdf3
           lwz 0,52(1)
           mtlr 0
           lwz 24,16(1)
           lwz 25,20(1)
           lwz 26,24(1)
           lwz 27,28(1)
           lwz 28,32(1)
           lwz 29,36(1)
           addi 1,1,48
           blr
           .size   foo, .-foo
           .ident  "GCC: (GNU) 4.2.2"
           .section        .note.GNU-stack,"",@progbits
   

   The fact that the compiler is calling helper functions (__adddf3, __muldf3, __divdf3) combined with heavy use of the General Purpose Registers is a clear indication for software-emulated FP support - and explains why this is so slow compared to a real FPU.
4. Build for SPE v2 support (as needed for example for a P2020 QorIQ processor, using the ppc_85xxDP target architecture):

   $ export CROSS_COMPILE=ppc_85xxDP-
   $ ppc_85xxDP-gcc -S -O fp_test.c 
   

   $ cat fp_test.s
           .file   "fp_test.c"
           .section        ".text"
           .align 2
           .globl foo
           .type   foo, @function
   foo:
           stwu 1,-48(1)
           stw 3,8(1)
           stw 4,12(1)
           stw 5,16(1)
           stw 6,20(1)
           evmergelo 0,3,4
           evmergelo 9,5,6
           efdadd 11,0,9
           efdmul 0,0,9
           efddiv 11,11,0
           evstdd 11,24(1)
           evmergehi 9,11,11
           mr 10,11
           stw 9,32(1)
           stw 10,36(1)
           mr 3,9
           mr 4,10
           addi 1,1,48
           blr
           .size   foo, .-foo
           .ident  "GCC: (GNU) 4.2.2"
           .section        .note.GNU-stack,"",@progbits
   

   Here we can see moderate use of General Purpos Registers combined with the use of SPE machine instructions (evmergelo, efdadd, efdmul, efddiv, evstdd, evmergehi) which proves that the compiler really generates code that supports the SPE.

14.1.10. ELDK 2.x Installation Aborts

Question:

I tried to install ELDK version 2.x on a SuSE 8.2 / SuSE 9 / RedHat-9 Linux host but failed - it terminated without installing any packages. Why?

Answer:

Newer Linux distributions use libraries that are incompatible to those used by the ELDK's installation tools. This problem was fixed in later releases of the ELDK (version 3.0 and later). It is therefore recommended to use a more recent version of the ELDK. If you really want to install an old version, the following back-port is available:

Please download the file ftp://ftp.denx.de/pub/tmp/ELDK-update-2.2.0.tar.bz2

Then change into the source tree with the ELDK files and perform the following operations:

bash$ rm RPMS/rpm-4.0.3-1.03b_2.i386.rpm \
    RPMS/rpm-build-4.0.3-1.03b_2.i386.rpm \
    RPMS/rpm-devel-4.0.3-1.03b_2.i386.rpm \
    tools/usr/lib/rpm/rpmpopt-4.0.3
bash$ tar jxf /tmp/ELDK-update-2.2.0.tar.bz2

Then build the ISO image as documented, and try again.

14.1.11. Enable SSH Access

Question:

How can I enable SSH access to the target system when running with the ELDK file system mounted over NFS?

Answer:

The ELDK includes the dropbear SSH server and client packages (see http://matt.ucc.asn.au/dropbear/dropbear.html). To enable SSH access, you must first generate the SSH host keys for RSA and DSS:

# dropbearkey -t rsa -f /etc/dropbear/dropbear_rsa_host_key
Will output 1024 bit rsa secret key to '/etc/dropbear/dropbear_rsa_host_key'
Generating key, this may take a while...
Public key portion is:
ssh-rsa ...
Fingerprint: md5 ...
# dropbearkey -t dss -f /etc/dropbear/dropbear_dss_host_key
Will output 1024 bit dss secret key to '/etc/dropbear/dropbear_dss_host_key'
Generating key, this may take a while...
Public key portion is:
ssh-dss ...
Fingerprint: md5 ...

Then you can start the dropbear daemon:

# dropbear

Now you should be able to access the target system through SSH.

14.2. U-Boot

14.2.1. Can U-Boot be configured such that it can be started in RAM?

Question:

I don't want to erase my flash memory because I'm not sure if my new U-Boot image will work. Is it possible to configure U-Boot such that I can load it into RAM instead of flash, and start it from my old boot loader?

Answer:

No. (Unless you're using a Blackfin processor, but you probably aren't.)

Question:

But I've been told it is possible??

Answer:

Well, yes. Of course this is possible. This is software, so everything is possible. But it is difficult, unsupported, and fraught with peril. You are on your own if you choose to do it. And it will not help you to solve your problem.

Question:

Why?

Answer:

U-Boot expects to see a virgin CPU, i. e. the CPU state must match what you see if the processor starts executing the first instructions when it comes out of reset. If you want to start U-Boot from another boot loader, you must disable a lot of code, i. e. all initialization parts that already have been performed by this other boot loader, like setting up the memory controller, initializing the SDRAM, initializing the serial port, setting up a stack frame etc. Also you must disable the relocation to RAM and adjust the link addresses etc.

This requires a lot of experience with U-Boot, and the fact that you had to ask if this can be done means that you are not in a position to do this.

The code you have to disable contains the most critical parts in U-Boot, i. e. these are the areas where 99% or more of all errors are located when you port U-Boot to a new hardware. In the result, your RAM image may work, but in the end you will need a full image to program the flash memory with it, and then you will have to enable all this highly critical and completely untested code.

You see? You cannot use a RAM version of U-Boot to avoid testing a flash version, so you can save all this effort and just burn your image to flash.

Question:

So how can I test an U-Boot image and recover my system if it doesn't work?

Answer:

Attach a BDI2000 (or any appropriate JTAG ICE) to your board, burn the image to flash, and debug it in its natural environment, i.e. U-Boot being the boot loader of the system and taking control over the CPU right as it comes out of reset. If something goes wrong, erase the flash and program a new image. This is a routine job using a BDI2000.

14.2.2. Relocation cannot be done when using -mrelocatable

Question:

I use ELDK version 3.0. When I build U-Boot I get error messages like this:

{standard input}: Assembler messages:
{standard input}:4998: Error: Relocation cannot be done when using -mrelocatable
...

Answer:

ELDK 3.0 uses GCC-3.2.2; your U-Boot sources are too old for this compiler. GCC-3.x requires a few adaptions which were added in later versions of U-Boot. Use for example the source tree (1.0.2) which is included with the ELDK, or download the latest version from CVS.

14.2.3. Source object has EABI version 4, but target has EABI version 0

Question:

When trying to build U-Boot with an EABI compliant tool chain, I get such error messages:

arm-ld: ERROR: Source object ... has EABI version 4, but target ... has EABI version 0

What does that mean, and how can I fix that?

Answer:

"EABI version 0" means the "apcs-gnu" ABI, while "EABI version 4" is the "aapcs-linux" ABI, aka "gnueabi".
All U-Boot ARM sources are built with "-mapcs-gnu" option set in "cpu/arm/config.mk", while libgcc.a modules are built in "gnueabi" format, which is for example the ARM GCC default in ELDK Release 4.2.
So the real problem is compatibility between toolchain ABI and U-Boot ARM ABI. In the Linux kernel there is a special kernel config option for EABI-enabled tool chains (CONFIG_AEABI), which enables special pieces of code in ARM assembler modules. We could follow this approach, reworking existing assembler sources and respective config.mk files in U-Boot.
Alternatively, the tool chain could provide a separate version of libgcc.a built with old ABI. This could be done using the multilib approach. The advantage here is that no U-boot changes will be required.

14.2.4. U-Boot crashes after relocation to RAM

Question:

I have ported U-Boot to a custom board. It starts OK, but crashes or hangs after relocating itself to RAM. Why?

Answer:

Your SDRAM initialization is bad, and the system crashes when it tries to fetch instructions from RAM. Note that simple read and write accesses may still work, it's the burst mode that is failing. This only shows up when caches are enabled because cache is the primary (or only) user of burst operations in U-Boot. In Linux, burst accesses may also result from DMA. For example, it is typical that a system may crash under heavy network load if the Ethernet controller uses DMA to memory.

It is NOT sufficient to program the memory controller of your CPU; each SDRAM chip also requires a specific initialization sequence which you must adhere to to the letter - check with the chip manufacturer's manual.

It has been observed that some operating systems like pSOS+ or VxWorks do not stress the memory subsystem as much as Linux or other UNIX systems like LynxOS do, so just because your board appears to work running another OS does not mean it is 100% OK.

Standard memory tests are not effective in identifying this type of problem because they do not cause stressful cache burst read/write operations.

With this caveat in mind, reportedly this program has found memory problems before: http://pyropus.ca/software/memtester/

Argument:

But my board ran fine with bootloader XYZ and/or operating system ABC.

Answer:

Double-check your configuration that you claim runs properly...

1. Are you sure the SDRAM is initialized using the same init sequence and values?
2. Are you sure the memory controlling registers are set the same?
3. Are you sure your other configuration uses caches and/or DMA? If it doesn't, it isn't a valid comparison.

14.2.5. Warning - bad CRC, using default environment

Question:

I have ported U-Boot to a custom board. It seems to boot OK, but it prints:

*** Warning - bad CRC, using default environment

Why?

Answer:

Most probably everything is OK. The message is printed because the flash sector or ERPROM containing the environment variables has never been initialized yet. The message will go away as soon as you save the envrionment variables using the saveenv command.

14.2.6. Net: No ethernet found

Question:

I have ported U-Boot to a custom board. It seems to boot OK, but it prints:

Net:   No ethernet found.

Why?

Answer:

Some network drivers (especially on ppc4xx) respond with such a message when no valid MAC address has been defined. Please make sure that a valid MAC address has been defined in the environment (using the "setenv" command). Then store the environment (using the "saveenv" command). After the next reboot, the Ethernet interface should be available.

14.2.7. Wrong debug symbols after relocation

Question:

I want to debug U-Boot after relocation to RAM, but it doesn't work since all the symbols are at wrong addresses now.

Answer:

To debug parts of U-Boot that are running from ROM/flash, i. e. before relocation, just use a command like "powerpc-linux-gdb uboot" as usual.

For parts of U-Boot that run from RAM, i. e. after relocation, use "powerpc-linux-gdb" without arguments, and use the add-symbol-file command in GDB to load the symbol table at the relocation address in RAM. The only problem is that you need to know that address, which depends on RAM size, length reserved for U-Boot, size of "protected RAM" area, etc. If in doubt, enable DEBUG mode when building U-Boot so it prints the address to the console.

Hint: I use definitions like these in my .gdbinit file:

define rom
        symbol-file
        file u-boot
end
                                                                                
define ram
        symbol-file
        add-symbol-file u-boot  0x01fe0000
end

Note: when you want to switch modes during one debug session (i. e. without restarting GDB) you can "delete" the current symbol information by using the symbol-file command without arguments, and then either using "symbol-file u-boot" for code before relocation, or "add-symbol-file u-boot _offset_" for code after relocation.

14.2.8. Decoding U-Boot Crash Dumps

When you are porting U-Boot to new hardware, or implementing extensions, you might run into situations where U-Boot crashes and prints a register dump and a stack trace, for example like this:

Bus Fault @ 0x00f8d70c, fixup 0x00000000
Machine check in kernel mode.
Caused by (from msr): regs 00f52cf8 Unknown values in msr
NIP: 00F8D70C XER: 0000005F LR: 00F8D6F4 REGS: 00f52cf8 TRAP: 0200 DAR: F9F68C00
MSR: 00009002 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 00

GPR00: 00016ACC 00F52DE8 00000000 F9F68C00 00FA38EC 00000001 F9F68BF8 0000000B 
GPR08: 00000002 00F55470 00000000 00F52D94 44004024 00000000 00FA2F00 C0F75000 
GPR16: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 
GPR24: 00000000 00FA38EC 00F553C0 00F55480 00000000 00F52F80 00FA41C0 00000001 
Call backtrace: 
00000000 00F8F998 00F8FA88 00F8FAF8 00F90B5C 00F90CF8 00F8385C 
00F79E6C 00F773B0 
machine check

To find out what happened, you can try to decode the stack backtrace (the list of addresses printed after the "Call backtrace:" line. The backtrace tool can be used for this purpose. However, there is a little problem: the addresses printed for the stack backtrace are after relocation of the U-Boot code to RAM; to use the backtrace tool you need to know U-Boot's address offset (the difference between the start address of U-Boot in flash and its relocation address in RAM).

The easiest way to find out the relocation address is to enable debugging for the U-Boot source file lib_*/board.c - U-Boot will then print some debug messages

...
Now running in RAM - U-Boot at: 00f75000
...

Now you have to calculate the address offset between your link address (The value of the TEXT_BASE definition in your board/?/config.mk file). In our case this value is 0x40000000, so the address offset is 0x40000000 - 0x00f75000 = 0x3f08b000

Now we use the backtrace script with the System.map file in the U-Boot source tree and this address offset:

-> backtrace System.map 0x3f08b000
Reading symbols from System.map
Using Address Offset 0x3f08b000
0x3f08b000 -- unknown address
0x4001a998 -- 0x4001a8d0 + 0x00c8   free_pipe
0x4001aa88 -- 0x4001aa2c + 0x005c   free_pipe_list
0x4001aaf8 -- 0x4001aad0 + 0x0028   run_list
0x4001bb5c -- 0x4001ba68 + 0x00f4   parse_stream_outer
0x4001bcf8 -- 0x4001bcd8 + 0x0020   parse_file_outer
0x4000e85c -- 0x4000e6f8 + 0x0164   main_loop
0x40004e6c -- 0x40004b9c + 0x02d0   board_init_r
0x400023b0 -- 0x400023b0 + 0x0000   trap_init

In this case the last "good" entry on the stack was in free_pipe...

14.2.9. Porting Problem: cannot move location counter backwards

Question:

I'm trying to port U-Boot to a new board and the linker throws an error message like this:

board/<your_board>/u-boot.lds:75 cannot move location counter backwards (from 00000000b0008010 to 00000000b0008000)

Answer:

Check your linker script board/your_board/u-boot.lds which controls how the object files are linked together to build the U-Boot image.

It looks as if your board uses an "embedded" environment, i. e. the flash sector containing the environment variables is surrounded by code. The u-boot.lds tries to collect as many as possible code in the first part, making the gap between this first part and the environment sector as small as possible. Everything that does not fit is then placed in the second part, after the environment sector.

Some your modifications caused the code that was put in this first part to grow, so that the linker finds that it would have to overwrite space that is already used.

Try commenting out one (or more) line(s) before the line containing the "common/environment.o" statement. [ "lib_generic/zlib.o" is usually a good candidate for testing as it's big ]. Once you get U-Boot linked, you can check in the u-boot.map file how big the gap is, and which object files could be used to fill it up again.

14.2.10. U-Boot Doesn't Run after Upgrading my Compiler

Question:

I encountered a big problem that U-Boot 1.1.4 compiled by ELDK 4.1 for MPC82xx crashed.

But if I build it using gcc-3.4.6 based cross tools, U-Boot on my board boots correctly.

The same U-Boot code built by ELDK 4.1 (gcc-4.0) failed, nothing occurs on the serial port.

Answer:

This is often a missing volatile attribute on shared variable references, particularly hardware registers. Newer compiler versions optimize more aggressively, making missing volatile attributes visible.

If you use -O0 (no optimization) does it fix the problem?
If it does, it most likely is an optimization/volatile issue. The hard part is figuring out where. Device handling and board-specific code is the place to start.

14.2.11. How Can I Reduce The Image Size?

Question:

I am trying to reduce the size of the u-boot.bin file so that it fits into 256 KB. I disabled all the drivers that I didn't need but the binary size is still 512KB, it seems to be a hard number coded in somewhere. Where can the image size be altered from?

Answer:

Some processors have a fixed reset vector address at 0xFFFFFFFC, so the U-Boot image has to include that address, i. e. it covers the full range from the start address to the end of the 32 bit address space. In such a case, the start address must be changed - check the setting of TEXT_BASE in your board/<name>/config.mk file.

14.2.12. Erasing Flash Fails

Question:

I tried to erase the flash memory like

erase 40050000 40050100

It fails. What am I doing wrong?

Answer:

Remember that flash memory cannot be erased in arbitrary areas, but only in so called "erase regions" or "sectors". If you have U-Boot running you can use the flinfo (Flash information, short fli) command to print information about the flash memory on your board, for instance:

=> fli
                                                                                
Bank # 1: AMD AM29LV160B (16 Mbit, bottom boot sect)
  Size: 4 MB in 35 Sectors
  Sector Start Addresses:
    40000000 (RO) 40008000 (RO) 4000C000 (RO) 40010000 (RO) 40020000 (RO)
    40040000      40060000      40080000      400A0000      400C0000
    400E0000      40100000      40120000      40140000      40160000
    40180000      401A0000      401C0000      401E0000      40200000
    40220000      40240000      40260000      40280000      402A0000
    402C0000      402E0000      40300000      40320000      40340000
    40360000      40380000      403A0000      403C0000      403E0000

In the example above, the area 40050000 ... 40050100 lies right in the middle of a erase unit (40040000 ... 4005FFFF), so you cannot erase it without erasing the whole sector, i. e. you have to type

=> erase 40040000 4005FFFF

Also note that there are some sectors marked as read-only ((RO)); you cannot erase or overwrite these sectors without un-protecting the sectors first (see the U-Boot protect command).

14.2.13. Ethernet Does Not Work

Question:

Ethernet does not work on my board.

Answer:

Maybe you forgot to set a MAC address? Check if the "ethaddr" environment variable is defined, and if it has a sane value. If there are more than one Ethernet interfaces on your board, you may also have to check the MAC addresses for these, i. e. check the "eth1addr", "eth2addr", etc. variables, too.

Question:

I have configured a MAC address of 01:02:03:04:05:06, and I can see that an ARP packet is sent by U-Boot, and that an ARP reply is sent by the server, but U-Boot never receives any packets. What's wrong?

Answer:

You have chosen a MAC address which, according to the ANSI/IEEE 802-1990 standard, has the multicast bit set. Under normal conditions a network interface discards such packets, and this is what U-Boot is doing. This is not a bug, but correct behaviour.

Please use only valid MAC addresses that were assigned to you.

For bring-up testing in the lab you can also use so-called locally administered ethernet addresses. These are addresses that have the 2nd LSB in the most significant byte of MAC address set. The gen_eth_addr tool that comes with U-Boot (see "tools/gen_eth_addr" ) can be used to generate random addresses from this pool.

14.2.14. Where Can I Get a Valid MAC Address from?

Question:

Where can I get a valid MAC address from?

Answer:

You have to buy a block of 4096 MAC addresses (IAB = Individual Address Block) or a block of 16M MAC addresses (OUI = Organizationally Unique Identifier, also referred to as 'company id') from IEEE Registration Authority. The current cost of an IAB is $550.00, the cost of an OUI is $1,650.00. See http://standards.ieee.org/regauth/oui/index.shtml

You can buy Eproms containing MAC addresses from: Maxim or Microchip.

You can set the "locally administered" bit to make your own MAC address (no guarantee of uniqueness, but pretty good odds if you don't do something dumb). Ref: Wikipedia

Universally administered and locally administered addresses are distinguished by setting the second least significant bit of the most significant byte of the address. If the bit is 0, the address is universally administered. If it is 1, the address is locally administered. The bit is 0 in all OUIs. For example, 02-00-00-00-00-01. The most significant byte is 02h. The binary is 00000010 and the second least significant bit is 1. Therefore, it is a locally administered address.

In U-Boot, you can use the "gen_eth_addr" tool to generate a random "locally administered" MAC address. Here are the needed commands:

$ make tools/gen_eth_addr
cc     tools/gen_eth_addr.c   -o tools/gen_eth_addr
$ tools/gen_eth_addr
ba:d0:4a:9c:4e:ce

14.2.15. Why do I get TFTP timeouts?

Question 1:: When trying to download a file from the TFTP server I always get timeouts like these:

...
Loading: #######T ##################################T###################T ####T ##T #
       ###T #T #########T ########T #############T ##T #############T ########T #############T
       #####T ###T ######T #######T #######T #############T ##T ##############T ###########
       ###########
done

Answer 1:: Most probably you have a full duplex/half duplex problem. Verify that U-Boot is setting the ethernet interface on your board to the proper duplex mode (full/half). I'm guessing your board is half duplex but your switch is full (typical of a switch ;-).

The switch sends traffic to your board while your board is transmitting... that is a collision (late collision at that) to your board but is OK to the switch. This doesn't happen nearly as much with a direct link to your PC since then you have a dedicated link without much asynchronous traffic.

The software (U-Boot/Linux) needs to poll the PHY chip for duplex mode and then (re)configure the MAC chip (separate or built into the CPU) to match. If the poll isn't happening or has a bug, you have problems like described above.

Question 2:: When I use tftp, there are some problems. My terminal always displays "Loading: T T T T T T T T T T T T T T T T T T T T". The whole information as follows:

U-Boot 1.1.4_XT (Jun  6 2006 - 17:36:18)
U-Boot code: 0C300000 -> 0C31AD70  BSS: -> 0C31EF98
RAM Configuration:
Bank #0: 0c000000  8 MB
Bank #1: 0c800000  8 MB
Flash:  2 MB
*** Warning - bad CRC, using default environment
In:    serial
Out:   serial
Err:   serial
Hit any key to stop autoboot:  0
XT=>  help tftp
tftpboot [loadAddress] [bootfilename]
XT=>  tftpboot 0x0c700000 image.bin
TFTP from server 192.168.0.23; our IP address is 192.168.0.70
Filename 'image.bin'.
Load address: 0xc700000
Loading: T T T T T T T T T T T T T T T T T T T T
Retry count exceeded; starting again
TFTP from server 192.168.0.23; our IP address is 192.168.0.70

Answer 2:: (1) Verify your TFTP server is working. On a machine (not the TFTP server nor your development board) use tftp to read the target file.

$ tftp 192.168.0.23 get image.bin

14.2.16. Why is my Ethernet operation not reliable?

Question:

My ethernet connection is not working reliable. On one switch it works fine, but on another one it doesn't.

Question:

I always see transmit errors or timeouts for the first packet of a download, but then it works well.

Question:

I cannot mount the Linux root file system over NFS; especially not with recent Linux kernel versions (older kernel versions work better). Specifying "proto=tcp" as mount option greatly improves the situation.

Answer:

There are many possible explanations for such problems. After eliminating the obvious sources (like broken cables etc.) you should check the configuration of your Ethernet PHY. One common cause of problems is if your PHY is hard configured in duplex mode (for example 100baseTX Full Duplex or 10baseT Full Duplex). If such a setup is combined with a autonegotiating switch, then trouble is ahead.

Jerry Van Baren explained this as follows:

Ignoring the configuration where both ends are (presumably correctly) 
manually configured, you end up with five cases, two of them 
misconfigured and WRONG:
1) Autonegotiation     <-> autonegotiation - reliable.
2) 10bT half duplex    <-> autonegotiation - reliable.
3) 100bT half duplex   <-> autonegotiation - reliable.
4) 10bT *FULL* duplex  <-> autonegotiation - *UNreliable*.
5) 100bT *FULL* duplex <-> autonegotiation - *UNreliable*.

The problem that I've observed is that the *humans* (the weak links) 
that do the manual configuration don't understand that "parallel 
detection" *must be* half duplex by definition in the spec (it is hard 
to define a reliable algorithm to detect full duplex capability so the 
spec writers punted).  As a result, the human invariably picks "full 
duplex" because everybody knows full duplex is better... and end up as 
case (4) or (5).  They inadvertently end up with a slower unreliable 
link (lots of "collisions" resulting in runt packets) rather than the 
faster better link they thought they were picking (d'oh!).  The really 
bad thing is that the network works fine in testing on an isolated LAN 
with no traffic and absolutely craps its pants when it hits the real
world.

That is my reasoning behind my statement that we can generally ignore 
the autonegotiation <-> fixed configuration case because the odds of it 
working properly are poor anyway.

Rule:

Always try to set up your PHY for autonegotiation.
If you must use some fixed setting, then set it to half duplex mode.
If you really must use a fixed full-duplex setting, then you absolutley must make sure that the link partner is configured exactly the same.

See also:

Wikipedia: Autonegotiation and Wikipedia: Duplex mismatch