%poky; ] > This chapter describes fundamental procedures such as creating layers, adding new software packages, extending or customizing images, porting work to new hardware (adding a new machine), and so forth. You will find that the procedures documented here occur often in the development cycle using the Yocto Project.

The OpenEmbedded build system supports organizing Metadata into multiple layers. Layers allow you to isolate different types of customizations from each other. You might find it tempting to keep everything in one layer when working on a single project. However, the more modular your Metadata, the easier it is to cope with future changes. To illustrate how layers are used to keep things modular, consider machine customizations. These types of customizations typically reside in a special layer, rather than a general layer, called a Board Support Package (BSP) Layer. Furthermore, the machine customizations should be isolated from recipes and Metadata that support a new GUI environment, for example. This situation gives you a couple of layers: one for the machine configurations, and one for the GUI environment. It is important to understand, however, that the BSP layer can still make machine-specific additions to recipes within the GUI environment layer without polluting the GUI layer itself with those machine-specific changes. You can accomplish this through a recipe that is a BitBake append (.bbappend) file, which is described later in this section. For general information on BSP layer structure, see the Board Support Packages (BSP) - Developer's Guide.

The Source Directory contains both general layers and BSP layers right out of the box. You can easily identify layers that ship with a Yocto Project release in the Source Directory by their folder names. Folders that represent layers typically have names that begin with the string meta-. It is not a requirement that a layer name begin with the prefix meta-, but it is a commonly accepted standard in the Yocto Project community. For example, when you set up the Source Directory structure, you will see several layers: meta, meta-skeleton, meta-selftest, meta-poky, and meta-yocto-bsp. Each of these folders represents a distinct layer. As another example, if you set up a local copy of the meta-intel Git repository and then explore the folder of that general layer, you will discover many Intel-specific BSP layers inside. For more information on BSP layers, see the "BSP Layers" section in the Yocto Project Board Support Package (BSP) Developer's Guide.

It is very easy to create your own layers to use with the OpenEmbedded build system. The Yocto Project ships with scripts that speed up creating general layers and BSP layers. This section describes the steps you perform by hand to create a layer so that you can better understand them. For information about the layer-creation scripts, see the "Creating a New BSP Layer Using the yocto-bsp Script" section in the Yocto Project Board Support Package (BSP) Developer's Guide and the "Creating a General Layer Using the bitbake-layers Script" section further down in this manual. Follow these general steps to create your layer without the aid of a script: Check Existing Layers: Before creating a new layer, you should be sure someone has not already created a layer containing the Metadata you need. You can see the OpenEmbedded Metadata Index for a list of layers from the OpenEmbedded community that can be used in the Yocto Project. Create a Directory: Create the directory for your layer. While not strictly required, prepend the name of the folder with the string meta-. For example: meta-mylayer meta-GUI_xyz meta-mymachine Create a Layer Configuration File: Inside your new layer folder, you need to create a conf/layer.conf file. It is easiest to take an existing layer configuration file and copy that to your layer's conf directory and then modify the file as needed. The meta-yocto-bsp/conf/layer.conf file demonstrates the required syntax: # We have a conf and classes directory, add to BBPATH BBPATH .= ":${LAYERDIR}" # We have recipes-* directories, add to BBFILES BBFILES += "${LAYERDIR}/recipes-*/*/*.bb \ ${LAYERDIR}/recipes-*/*/*.bbappend" BBFILE_COLLECTIONS += "yoctobsp" BBFILE_PATTERN_yoctobsp = "^${LAYERDIR}/" BBFILE_PRIORITY_yoctobsp = "5" LAYERVERSION_yoctobsp = "3" Here is an explanation of the example: The configuration and classes directory is appended to BBPATH. All non-distro layers, which include all BSP layers, are expected to append the layer directory to the BBPATH. On the other hand, distro layers, such as meta-poky, can choose to enforce their own precedence over BBPATH. For an example of that syntax, see the layer.conf file for the meta-poky layer. The recipes for the layers are appended to BBFILES. The BBFILE_COLLECTIONS variable is then appended with the layer name. The BBFILE_PATTERN variable is set to a regular expression and is used to match files from BBFILES into a particular layer. In this case, LAYERDIR is used to make BBFILE_PATTERN match within the layer's path. The BBFILE_PRIORITY variable then assigns a priority to the layer. Applying priorities is useful in situations where the same recipe might appear in multiple layers and allows you to choose the layer that takes precedence. The LAYERVERSION variable optionally specifies the version of a layer as a single number. Note the use of the LAYERDIR variable, which expands to the directory of the current layer. Through the use of the BBPATH variable, BitBake locates class files (.bbclass), configuration files, and files that are included with include and require statements. For these cases, BitBake uses the first file that matches the name found in BBPATH. This is similar to the way the PATH variable is used for binaries. It is recommended, therefore, that you use unique class and configuration filenames in your custom layer. Add Content: Depending on the type of layer, add the content. If the layer adds support for a machine, add the machine configuration in a conf/machine/ file within the layer. If the layer adds distro policy, add the distro configuration in a conf/distro/ file within the layer. If the layer introduces new recipes, put the recipes you need in recipes-* subdirectories within the layer. In order to be compliant with the Yocto Project, a layer must contain a README file. Optionally Test for Compatibility: If you want permission to use the Yocto Project Compatibility logo with your layer or application that uses your layer, perform the steps to apply for compatibility. See the "Making Sure Your Layer is Compatible With Yocto Project" section for more information.

To create layers that are easier to maintain and that will not impact builds for other machines, you should consider the information in the following list: Avoid "Overlaying" Entire Recipes from Other Layers in Your Configuration: In other words, do not copy an entire recipe into your layer and then modify it. Rather, use an append file (.bbappend) to override only those parts of the original recipe you need to modify. Avoid Duplicating Include Files: Use append files (.bbappend) for each recipe that uses an include file. Or, if you are introducing a new recipe that requires the included file, use the path relative to the original layer directory to refer to the file. For example, use require recipes-core/package/file.inc instead of require file.inc. If you're finding you have to overlay the include file, it could indicate a deficiency in the include file in the layer to which it originally belongs. If this is the case, you should try to address that deficiency instead of overlaying the include file. For example, you could address this by getting the maintainer of the include file to add a variable or variables to make it easy to override the parts needing to be overridden. Structure Your Layers: Proper use of overrides within append files and placement of machine-specific files within your layer can ensure that a build is not using the wrong Metadata and negatively impacting a build for a different machine. Following are some examples: Modify Variables to Support a Different Machine: Suppose you have a layer named meta-one that adds support for building machine "one". To do so, you use an append file named base-files.bbappend and create a dependency on "foo" by altering the DEPENDS variable: DEPENDS = "foo" The dependency is created during any build that includes the layer meta-one. However, you might not want this dependency for all machines. For example, suppose you are building for machine "two" but your bblayers.conf file has the meta-one layer included. During the build, the base-files for machine "two" will also have the dependency on foo. To make sure your changes apply only when building machine "one", use a machine override with the DEPENDS statement: DEPENDS_one = "foo" You should follow the same strategy when using _append and _prepend operations: DEPENDS_append_one = " foo" DEPENDS_prepend_one = "foo " As an actual example, here's a line from the recipe for gnutls, which adds dependencies on "argp-standalone" when building with the musl C library: DEPENDS_append_libc-musl = " argp-standalone" Avoiding "+=" and "=+" and using machine-specific _append and _prepend operations is recommended as well. Place Machine-Specific Files in Machine-Specific Locations: When you have a base recipe, such as base-files.bb, that contains a SRC_URI statement to a file, you can use an append file to cause the build to use your own version of the file. For example, an append file in your layer at meta-one/recipes-core/base-files/base-files.bbappend could extend FILESPATH using FILESEXTRAPATHS as follows: FILESEXTRAPATHS_prepend := "${THISDIR}/${BPN}:" The build for machine "one" will pick up your machine-specific file as long as you have the file in meta-one/recipes-core/base-files/base-files/. However, if you are building for a different machine and the bblayers.conf file includes the meta-one layer and the location of your machine-specific file is the first location where that file is found according to FILESPATH, builds for all machines will also use that machine-specific file. You can make sure that a machine-specific file is used for a particular machine by putting the file in a subdirectory specific to the machine. For example, rather than placing the file in meta-one/recipes-core/base-files/base-files/ as shown above, put it in meta-one/recipes-core/base-files/base-files/one/. Not only does this make sure the file is used only when building for machine "one", but the build process locates the file more quickly. In summary, you need to place all files referenced from SRC_URI in a machine-specific subdirectory within the layer in order to restrict those files to machine-specific builds. Perform Steps to Apply for Yocto Project Compatibility: If you want permission to use the Yocto Project Compatibility logo with your layer or application that uses your layer, perform the steps to apply for compatibility. See the "Making Sure Your Layer is Compatible With Yocto Project" section for more information. Follow the Layer Naming Convention: Store custom layers in a Git repository that use the meta-layer_name format. Group Your Layers Locally: Clone your repository alongside other cloned meta directories from the Source Directory.

When you create a layer used with the Yocto Project, it is advantageous to make sure that the layer interacts well with existing Yocto Project layers (i.e. the layer is compatible with the Yocto Project). Ensuring compatibility makes the layer easy to be consumed by others in the Yocto Project community and could allow you permission to use the Yocto Project Compatible Logo. Only Yocto Project member organizations are permitted to use the Yocto Project Compatible Logo. The logo is not available for general use. For information on how to become a Yocto Project member organization, see the Yocto Project Website. The Yocto Project Compatibility Program consists of a layer application process that requests permission to use the Yocto Project Compatibility Logo for your layer and application. The process consists of two parts: Successfully passing a script (yocto-check-layer) that when run against your layer, tests it against constraints based on experiences of how layers have worked in the real world and where pitfalls have been found. Getting a "PASS" result from the script is required for successful compatibility registration. Completion of an application acceptance form, which you can find at . To be granted permission to use the logo, you need to satisfy the following: Be able to check the box indicating that you got a "PASS" when running the script against your layer. Answer "Yes" to the questions on the form or have an acceptable explanation for any questions answered "No". You need to be a Yocto Project Member Organization. The remainder of this section presents information on the registration form and on the yocto-check-layer script.

Use the form to apply for your layer's approval. Upon successful application, you can use the Yocto Project Compatibility Logo with your layer and the application that uses your layer. To access the form, use this link: . Follow the instructions on the form to complete your application. The application consists of the following sections: Contact Information: Provide your contact information as the fields require. Along with your information, provide the released versions of the Yocto Project for which your layer is compatible. Acceptance Criteria: Provide "Yes" or "No" answers for each of the items in the checklist. Space exists at the bottom of the form for any explanations for items for which you answered "No". Recommendations: Provide answers for the questions regarding Linux kernel use and build success.

The yocto-check-layer script provides you a way to assess how compatible your layer is with the Yocto Project. You should run this script prior to using the form to apply for compatibility as described in the previous section. You need to achieve a "PASS" result in order to have your application form successfully processed. The script divides tests into three areas: COMMON, BSD, and DISTRO. For example, given a distribution layer (DISTRO), the layer must pass both the COMMON and DISTRO related tests. Furthermore, if your layer is a BSP layer, the layer must pass the COMMON and BSP set of tests. To execute the script, enter the following commands from your build directory: $ source oe-init-build-env $ yocto-check-layer your_layer_directory Be sure to provide the actual directory for your layer as part of the command. Entering the command causes the script to determine the type of layer and then to execute a set of specific tests against the layer. The following list overviews the test: common.test_readme: Tests if a README file exists in the layer and the file is not empty. common.test_parse: Tests to make sure that BitBake can parse the files without error (i.e. bitbake -p). common.test_show_environment: Tests that the global or per-recipe environment is in order without errors (i.e. bitbake -e). common.test_signatures: Tests to be sure that BSP and DISTRO layers do not come with recipes that change signatures. bsp.test_bsp_defines_machines: Tests if a BSP layer has machine configurations. bsp.test_bsp_no_set_machine: Tests to ensure a BSP layer does not set the machine when the layer is added. distro.test_distro_defines_distros: Tests if a DISTRO layer has distro configurations. distro.test_distro_no_set_distro: Tests to ensure a DISTRO layer does not set the distribution when the layer is added.

Before the OpenEmbedded build system can use your new layer, you need to enable it. To enable your layer, simply add your layer's path to the BBLAYERS variable in your conf/bblayers.conf file, which is found in the Build Directory. The following example shows how to enable a layer named meta-mylayer: LCONF_VERSION = "6" BBPATH = "${TOPDIR}" BBFILES ?= "" BBLAYERS ?= " \ $HOME/poky/meta \ $HOME/poky/meta-poky \ $HOME/poky/meta-yocto-bsp \ $HOME/poky/meta-mylayer \ " BitBake parses each conf/layer.conf file as specified in the BBLAYERS variable within the conf/bblayers.conf file. During the processing of each conf/layer.conf file, BitBake adds the recipes, classes and configurations contained within the particular layer to the source directory.

A recipe that appends Metadata to another recipe is called a BitBake append file. A BitBake append file uses the .bbappend file type suffix, while the corresponding recipe to which Metadata is being appended uses the .bb file type suffix. You can use a .bbappend file in your layer to make additions or changes to the content of another layer's recipe without having to copy the other layer's recipe into your layer. Your .bbappend file resides in your layer, while the main .bb recipe file to which you are appending Metadata resides in a different layer. Being able to append information to an existing recipe not only avoids duplication, but also automatically applies recipe changes from a different layer into your layer. If you were copying recipes, you would have to manually merge changes as they occur. When you create an append file, you must use the same root name as the corresponding recipe file. For example, the append file someapp_&DISTRO;.bbappend must apply to someapp_&DISTRO;.bb. This means the original recipe and append file names are version number-specific. If the corresponding recipe is renamed to update to a newer version, you must also rename and possibly update the corresponding .bbappend as well. During the build process, BitBake displays an error on starting if it detects a .bbappend file that does not have a corresponding recipe with a matching name. See the BB_DANGLINGAPPENDS_WARNONLY variable for information on how to handle this error. As an example, consider the main formfactor recipe and a corresponding formfactor append file both from the Source Directory. Here is the main formfactor recipe, which is named formfactor_0.0.bb and located in the "meta" layer at meta/recipes-bsp/formfactor: SUMMARY = "Device formfactor information" SECTION = "base" LICENSE = "MIT" LIC_FILES_CHKSUM = "file://${COREBASE}/meta/COPYING.MIT;md5=3da9cfbcb788c80a0384361b4de20420" PR = "r45" SRC_URI = "file://config file://machconfig" S = "${WORKDIR}" PACKAGE_ARCH = "${MACHINE_ARCH}" INHIBIT_DEFAULT_DEPS = "1" do_install() { # Install file only if it has contents install -d ${D}${sysconfdir}/formfactor/ install -m 0644 ${S}/config ${D}${sysconfdir}/formfactor/ if [ -s "${S}/machconfig" ]; then install -m 0644 ${S}/machconfig ${D}${sysconfdir}/formfactor/ fi } In the main recipe, note the SRC_URI variable, which tells the OpenEmbedded build system where to find files during the build. Following is the append file, which is named formfactor_0.0.bbappend and is from the Raspberry Pi BSP Layer named meta-raspberrypi. The file is in the layer at recipes-bsp/formfactor: FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}:" By default, the build system uses the FILESPATH variable to locate files. This append file extends the locations by setting the FILESEXTRAPATHS variable. Setting this variable in the .bbappend file is the most reliable and recommended method for adding directories to the search path used by the build system to find files. The statement in this example extends the directories to include ${THISDIR}/${PN}, which resolves to a directory named formfactor in the same directory in which the append file resides (i.e. meta-raspberrypi/recipes-bsp/formfactor. This implies that you must have the supporting directory structure set up that will contain any files or patches you will be including from the layer. Using the immediate expansion assignment operator := is important because of the reference to THISDIR. The trailing colon character is important as it ensures that items in the list remain colon-separated. BitBake automatically defines the THISDIR variable. You should never set this variable yourself. Using "_prepend" as part of the FILESEXTRAPATHS ensures your path will be searched prior to other paths in the final list. Also, not all append files add extra files. Many append files simply exist to add build options (e.g. systemd). For these cases, your append file would not even use the FILESEXTRAPATHS statement.

Each layer is assigned a priority value. Priority values control which layer takes precedence if there are recipe files with the same name in multiple layers. For these cases, the recipe file from the layer with a higher priority number takes precedence. Priority values also affect the order in which multiple .bbappend files for the same recipe are applied. You can either specify the priority manually, or allow the build system to calculate it based on the layer's dependencies. To specify the layer's priority manually, use the BBFILE_PRIORITY variable. For example: BBFILE_PRIORITY_mylayer = "1" It is possible for a recipe with a lower version number PV in a layer that has a higher priority to take precedence. Also, the layer priority does not currently affect the precedence order of .conf or .bbclass files. Future versions of BitBake might address this.

You can use the BitBake layer management tool to provide a view into the structure of recipes across a multi-layer project. Being able to generate output that reports on configured layers with their paths and priorities and on .bbappend files and their applicable recipes can help to reveal potential problems. Use the following form when running the layer management tool. $ bitbake-layers command [arguments] The following list describes the available commands: help: Displays general help or help on a specified command. show-layers: Shows the current configured layers. show-recipes: Lists available recipes and the layers that provide them. show-overlayed: Lists overlayed recipes. A recipe is overlayed when a recipe with the same name exists in another layer that has a higher layer priority. show-appends: Lists .bbappend files and the recipe files to which they apply. show-cross-depends: Lists dependency relationships between recipes that cross layer boundaries. add-layer: Adds a layer to bblayers.conf. remove-layer: Removes a layer from bblayers.conf flatten: Flattens the layer configuration into a separate output directory. Flattening your layer configuration builds a "flattened" directory that contains the contents of all layers, with any overlayed recipes removed and any .bbappend files appended to the corresponding recipes. You might have to perform some manual cleanup of the flattened layer as follows: Non-recipe files (such as patches) are overwritten. The flatten command shows a warning for these files. Anything beyond the normal layer setup has been added to the layer.conf file. Only the lowest priority layer's layer.conf is used. Overridden and appended items from .bbappend files need to be cleaned up. The contents of each .bbappend end up in the flattened recipe. However, if there are appended or changed variable values, you need to tidy these up yourself. Consider the following example. Here, the bitbake-layers command adds the line #### bbappended ... so that you know where the following lines originate: ... DESCRIPTION = "A useful utility" ... EXTRA_OECONF = "--enable-something" ... #### bbappended from meta-anotherlayer #### DESCRIPTION = "Customized utility" EXTRA_OECONF += "--enable-somethingelse" Ideally, you would tidy up these utilities as follows: ... DESCRIPTION = "Customized utility" ... EXTRA_OECONF = "--enable-something --enable-somethingelse" ... layerindex-fetch: Fetches a layer from a layer index, along with its dependent layers, and adds the layers to the conf/bblayers.conf file. layerindex-show-depends: Finds layer dependencies from the layer index. create-layer: Creates a basic layer.

The bitbake-layers script with the create-layer subcommand simplifies creating a new general layer. For information on BSP layers, see the "BSP Layers" section in the Yocto Project Board Specific (BSP) Developer's Guide. The default mode of the script's operation with this subcommand is to create a layer with the following: A layer priority of 6. A conf subdirectory that contains a layer.conf file. A recipes-example subdirectory that contains a further subdirectory named example, which contains an example.bb recipe file. A COPYING.MIT, which is the license statement for the layer. The script assumes you want to use the MIT license, which is typical for most layers, for the contents of the layer itself. A README file, which is a file describing the contents of your new layer. In its simplest form, you can use the following command form to create a layer. The command creates a layer whose name corresponds to your_layer_name in the current directory: $ bitbake-layers create-layer your_layer_name If you want to set the priority of the layer to other than the default value of "6", you can either use the ‐‐priority option or you can edit the BBFILE_PRIORITY value in the conf/layer.conf after the script creates it. Furthermore, if you want to give the example recipe file some name other than the default, you can use the ‐‐example-recipe-name option. The easiest way to see how the bitbake-layers create-layer command works is to experiment with the script. You can also read the usage information by entering the following: $ bitbake-layers create-layer --help NOTE: Starting bitbake server... usage: bitbake-layers create-layer [-h] [--priority PRIORITY] [--example-recipe-name EXAMPLERECIPE] layerdir Create a basic layer positional arguments: layerdir Layer directory to create optional arguments: -h, --help show this help message and exit --priority PRIORITY, -p PRIORITY Layer directory to create --example-recipe-name EXAMPLERECIPE, -e EXAMPLERECIPE Filename of the example recipe Once you create your general layer, you must add it to your bblayers.conf file. You can add your layer by using the bitbake-layers add-layer command: $ bitbake-layers add-layer your_layer_name Here is an example where a layer named meta-scottrif is added and then the layers are shown using the bitbake-layers show-layers command: $ bitbake-layers add-layer meta-scottrif NOTE: Starting bitbake server... Loading cache: 100% |############################################| Time: 0:00:00 Loaded 1275 entries from dependency cache. Parsing recipes: 100% |##########################################| Time: 0:00:00 Parsing of 819 .bb files complete (817 cached, 2 parsed). 1276 targets, 44 skipped, 0 masked, 0 errors. $ bitbake-layers show-layers NOTE: Starting bitbake server... layer path priority ========================================================================== meta /home/scottrif/poky/meta 5 meta-poky /home/scottrif/poky/meta-poky 5 meta-yocto-bsp /home/scottrif/poky/meta-yocto-bsp 5 meta-mylayer /home/scottrif/meta-mylayer 6 workspace /home/scottrif/poky/build/workspace 99 meta-scottrif /home/scottrif/poky/build/meta-scottrif 6 Adding the layer to this file enables the build system to locate the layer during the build.

You can customize images to satisfy particular requirements. This section describes several methods and provides guidelines for each.

Probably the easiest way to customize an image is to add a package by way of the local.conf configuration file. Because it is limited to local use, this method generally only allows you to add packages and is not as flexible as creating your own customized image. When you add packages using local variables this way, you need to realize that these variable changes are in effect for every build and consequently affect all images, which might not be what you require. To add a package to your image using the local configuration file, use the IMAGE_INSTALL variable with the _append operator: IMAGE_INSTALL_append = " strace" Use of the syntax is important - specifically, the space between the quote and the package name, which is strace in this example. This space is required since the _append operator does not add the space. Furthermore, you must use _append instead of the += operator if you want to avoid ordering issues. The reason for this is because doing so unconditionally appends to the variable and avoids ordering problems due to the variable being set in image recipes and .bbclass files with operators like ?=. Using _append ensures the operation takes affect. As shown in its simplest use, IMAGE_INSTALL_append affects all images. It is possible to extend the syntax so that the variable applies to a specific image only. Here is an example: IMAGE_INSTALL_append_pn-core-image-minimal = " strace" This example adds strace to the core-image-minimal image only. You can add packages using a similar approach through the CORE_IMAGE_EXTRA_INSTALL variable. If you use this variable, only core-image-* images are affected.

Another method for customizing your image is to enable or disable high-level image features by using the IMAGE_FEATURES and EXTRA_IMAGE_FEATURES variables. Although the functions for both variables are nearly equivalent, best practices dictate using IMAGE_FEATURES from within a recipe and using EXTRA_IMAGE_FEATURES from within your local.conf file, which is found in the Build Directory. To understand how these features work, the best reference is meta/classes/core-image.bbclass. This class lists out the available IMAGE_FEATURES of which most map to package groups while some, such as debug-tweaks and read-only-rootfs, resolve as general configuration settings. In summary, the file looks at the contents of the IMAGE_FEATURES variable and then maps or configures the feature accordingly. Based on this information, the build system automatically adds the appropriate packages or configurations to the IMAGE_INSTALL variable. Effectively, you are enabling extra features by extending the class or creating a custom class for use with specialized image .bb files. Use the EXTRA_IMAGE_FEATURES variable from within your local configuration file. Using a separate area from which to enable features with this variable helps you avoid overwriting the features in the image recipe that are enabled with IMAGE_FEATURES. The value of EXTRA_IMAGE_FEATURES is added to IMAGE_FEATURES within meta/conf/bitbake.conf. To illustrate how you can use these variables to modify your image, consider an example that selects the SSH server. The Yocto Project ships with two SSH servers you can use with your images: Dropbear and OpenSSH. Dropbear is a minimal SSH server appropriate for resource-constrained environments, while OpenSSH is a well-known standard SSH server implementation. By default, the core-image-sato image is configured to use Dropbear. The core-image-full-cmdline and core-image-lsb images both include OpenSSH. The core-image-minimal image does not contain an SSH server. You can customize your image and change these defaults. Edit the IMAGE_FEATURES variable in your recipe or use the EXTRA_IMAGE_FEATURES in your local.conf file so that it configures the image you are working with to include ssh-server-dropbear or ssh-server-openssh. See the "Images" section in the Yocto Project Reference Manual for a complete list of image features that ship with the Yocto Project.

You can also customize an image by creating a custom recipe that defines additional software as part of the image. The following example shows the form for the two lines you need: IMAGE_INSTALL = "packagegroup-core-x11-base package1 package2" inherit core-image Defining the software using a custom recipe gives you total control over the contents of the image. It is important to use the correct names of packages in the IMAGE_INSTALL variable. You must use the OpenEmbedded notation and not the Debian notation for the names (e.g. glibc-dev instead of libc6-dev). The other method for creating a custom image is to base it on an existing image. For example, if you want to create an image based on core-image-sato but add the additional package strace to the image, copy the meta/recipes-sato/images/core-image-sato.bb to a new .bb and add the following line to the end of the copy: IMAGE_INSTALL += "strace"

For complex custom images, the best approach for customizing an image is to create a custom package group recipe that is used to build the image or images. A good example of a package group recipe is meta/recipes-core/packagegroups/packagegroup-base.bb. If you examine that recipe, you see that the PACKAGES variable lists the package group packages to produce. The inherit packagegroup statement sets appropriate default values and automatically adds -dev, -dbg, and -ptest complementary packages for each package specified in the PACKAGES statement. The inherit packages should be located near the top of the recipe, certainly before the PACKAGES statement. For each package you specify in PACKAGES, you can use RDEPENDS and RRECOMMENDS entries to provide a list of packages the parent task package should contain. You can see examples of these further down in the packagegroup-base.bb recipe. Here is a short, fabricated example showing the same basic pieces: DESCRIPTION = "My Custom Package Groups" inherit packagegroup PACKAGES = "\ packagegroup-custom-apps \ packagegroup-custom-tools \ " RDEPENDS_packagegroup-custom-apps = "\ dropbear \ portmap \ psplash" RDEPENDS_packagegroup-custom-tools = "\ oprofile \ oprofileui-server \ lttng-tools" RRECOMMENDS_packagegroup-custom-tools = "\ kernel-module-oprofile" In the previous example, two package group packages are created with their dependencies and their recommended package dependencies listed: packagegroup-custom-apps, and packagegroup-custom-tools. To build an image using these package group packages, you need to add packagegroup-custom-apps and/or packagegroup-custom-tools to IMAGE_INSTALL. For other forms of image dependencies see the other areas of this section.

By default, the configured hostname (i.e. /etc/hostname) in an image is the same as the machine name. For example, if MACHINE equals "qemux86", the configured hostname written to /etc/hostname is "qemux86". You can customize this name by altering the value of the "hostname" variable in the base-files recipe using either an append file or a configuration file. Use the following in an append file: hostname="myhostname" Use the following in a configuration file: hostname_pn-base-files = "myhostname" Changing the default value of the variable "hostname" can be useful in certain situations. For example, suppose you need to do extensive testing on an image and you would like to easily identify the image under test from existing images with typical default hostnames. In this situation, you could change the default hostname to "testme", which results in all the images using the name "testme". Once testing is complete and you do not need to rebuild the image for test any longer, you can easily reset the default hostname. Another point of interest is that if you unset the variable, the image will have no default hostname in the filesystem. Here is an example that unsets the variable in a configuration file: hostname_pn-base-files = "" Having no default hostname in the filesystem is suitable for environments that use dynamic hostnames such as virtual machines.

Recipes (.bb files) are fundamental components in the Yocto Project environment. Each software component built by the OpenEmbedded build system requires a recipe to define the component. This section describes how to create, write, and test a new recipe. For information on variables that are useful for recipes and for information about recipe naming issues, see the "Required" section of the Yocto Project Reference Manual.

The following figure shows the basic process for creating a new recipe. The remainder of the section provides details for the steps.

You can always write a recipe from scratch. However, three choices exist that can help you quickly get a start on a new recipe: devtool add: A command that assists in creating a recipe and an environment conducive to development. recipetool create: A command provided by the Yocto Project that automates creation of a base recipe based on the source files. Existing Recipes: Location and modification of an existing recipe that is similar in function to the recipe you need. For information on recipe syntax, see the "Recipe Syntax" section in the Yocto Project Reference Manual.

The devtool add command uses the same logic for auto-creating the recipe as recipetool create, which is listed below. Additionally, however, devtool add sets up an environment that makes it easy for you to patch the source and to make changes to the recipe as is often necessary when adding a recipe to build a new piece of software to be included in a build. You can find a complete description of the devtool add command in the "A Closer Look at devtool add" section in the Yocto Project Application Development and the Extensible Software Development Kit (eSDK) manual.

recipetool create automates creation of a base recipe given a set of source code files. As long as you can extract or point to the source files, the tool will construct a recipe and automatically configure all pre-build information into the recipe. For example, suppose you have an application that builds using Autotools. Creating the base recipe using recipetool results in a recipe that has the pre-build dependencies, license requirements, and checksums configured. To run the tool, you just need to be in your Build Directory and have sourced the build environment setup script (i.e. oe-init-build-env). Here is the basic recipetool syntax: Running recipetool -h or recipetool create -h produces the Python-generated help, which presented differently than what follows here. recipetool -h recipetool create [-h] recipetool [-d] [-q] [--color auto | always | never ] create -o OUTFILE [-m] [-x EXTERNALSRC] source -d Enables debug output. -q Outputs only errors (quiet mode). --color Colorizes the output automatically, always, or never. -h Displays Python generated syntax for recipetool. create Causes recipetool to create a base recipe. The create command is further defined with these options: -o OUTFILE Specifies the full path and filename for the generated recipe. -m Causes the recipe to be machine-specific rather than architecture-specific (default). -x EXTERNALSRC Fetches and extracts source files from source and places them in EXTERNALSRC. source must be a URL. -h Displays Python-generated syntax for create. source Specifies the source code on which to base the recipe. Running recipetool create -o OUTFILE creates the base recipe and locates it properly in the layer that contains your source files. Following are some syntax examples: Use this syntax to generate a recipe based on source. Once generated, the recipe resides in the existing source code layer: recipetool create -o OUTFILE source Use this syntax to generate a recipe using code that you extract from source. The extracted code is placed in its own layer defined by EXTERNALSRC. recipetool create -o OUTFILE -x EXTERNALSRC source Use this syntax to generate a recipe based on source. The options direct recipetool to generate debugging information. Once generated, the recipe resides in the existing source code layer: recipetool create -d -o OUTFILE source

Before writing a recipe from scratch, it is often useful to discover whether someone else has already written one that meets (or comes close to meeting) your needs. The Yocto Project and OpenEmbedded communities maintain many recipes that might be candidates for what you are doing. You can find a good central index of these recipes in the OpenEmbedded metadata index. Working from an existing recipe or a skeleton recipe is the best way to get started. Here are some points on both methods: Locate and modify a recipe that is close to what you want to do: This method works when you are familiar with the current recipe space. The method does not work so well for those new to the Yocto Project or writing recipes. Some risks associated with this method are using a recipe that has areas totally unrelated to what you are trying to accomplish with your recipe, not recognizing areas of the recipe that you might have to add from scratch, and so forth. All these risks stem from unfamiliarity with the existing recipe space. Use and modify the following skeleton recipe: If for some reason you do not want to use recipetool and you cannot find an existing recipe that is close to meeting your needs, you can use the following structure to provide the fundamental areas of a new recipe. DESCRIPTION = "" HOMEPAGE = "" LICENSE = "" SECTION = "" DEPENDS = "" LIC_FILES_CHKSUM = "" SRC_URI = ""

Once you have your base recipe, you should put it in your own layer and name it appropriately. Locating it correctly ensures that the OpenEmbedded build system can find it when you use BitBake to process the recipe. Storing Your Recipe: The OpenEmbedded build system locates your recipe through the layer's conf/layer.conf file and the BBFILES variable. This variable sets up a path from which the build system can locate recipes. Here is the typical use: BBFILES += "${LAYERDIR}/recipes-*/*/*.bb \ ${LAYERDIR}/recipes-*/*/*.bbappend" Consequently, you need to be sure you locate your new recipe inside your layer such that it can be found. You can find more information on how layers are structured in the "Understanding and Creating Layers" section. Naming Your Recipe: When you name your recipe, you need to follow this naming convention: basename_version.bb Use lower-cased characters and do not include the reserved suffixes -native, -cross, -initial, or -dev casually (i.e. do not use them as part of your recipe name unless the string applies). Here are some examples: cups_1.7.0.bb gawk_4.0.2.bb irssi_0.8.16-rc1.bb

Creating a new recipe is usually an iterative process that requires using BitBake to process the recipe multiple times in order to progressively discover and add information to the recipe file. Assuming you have sourced the build environment setup script (i.e. &OE_INIT_FILE;) and you are in the Build Directory, use BitBake to process your recipe. All you need to provide is the basename of the recipe as described in the previous section: $ bitbake basename During the build, the OpenEmbedded build system creates a temporary work directory for each recipe (${WORKDIR}) where it keeps extracted source files, log files, intermediate compilation and packaging files, and so forth. The path to the per-recipe temporary work directory depends on the context in which it is being built. The quickest way to find this path is to have BitBake return it by running the following: $ bitbake -e basename | grep ^WORKDIR= As an example, assume a Source Directory top-level folder named poky, a default Build Directory at poky/build, and a qemux86-poky-linux machine target system. Furthermore, suppose your recipe is named foo_1.3.0.bb. In this case, the work directory the build system uses to build the package would be as follows: poky/build/tmp/work/qemux86-poky-linux/foo/1.3.0-r0 Inside this directory you can find sub-directories such as image, packages-split, and temp. After the build, you can examine these to determine how well the build went. You can find log files for each task in the recipe's temp directory (e.g. poky/build/tmp/work/qemux86-poky-linux/foo/1.3.0-r0/temp). Log files are named log.taskname (e.g. log.do_configure, log.do_fetch, and log.do_compile). You can find more information about the build process in "The Yocto Project Development Environment" chapter of the Yocto Project Reference Manual.

The first thing your recipe must do is specify how to fetch the source files. Fetching is controlled mainly through the SRC_URI variable. Your recipe must have a SRC_URI variable that points to where the source is located. For a graphical representation of source locations, see the "Sources" section in the Yocto Project Reference Manual. The do_fetch task uses the prefix of each entry in the SRC_URI variable value to determine which fetcher to use to get your source files. It is the SRC_URI variable that triggers the fetcher. The do_patch task uses the variable after source is fetched to apply patches. The OpenEmbedded build system uses FILESOVERRIDES for scanning directory locations for local files in SRC_URI. The SRC_URI variable in your recipe must define each unique location for your source files. It is good practice to not hard-code pathnames in an URL used in SRC_URI. Rather than hard-code these paths, use ${PV}, which causes the fetch process to use the version specified in the recipe filename. Specifying the version in this manner means that upgrading the recipe to a future version is as simple as renaming the recipe to match the new version. Here is a simple example from the meta/recipes-devtools/cdrtools/cdrtools-native_3.01a20.bb recipe where the source comes from a single tarball. Notice the use of the PV variable: SRC_URI = "ftp://ftp.berlios.de/pub/cdrecord/alpha/cdrtools-${PV}.tar.bz2" Files mentioned in SRC_URI whose names end in a typical archive extension (e.g. .tar, .tar.gz, .tar.bz2, .zip, and so forth), are automatically extracted during the do_unpack task. For another example that specifies these types of files, see the "Autotooled Package" section. Another way of specifying source is from an SCM. For Git repositories, you must specify SRCREV and you should specify PV to include the revision with SRCPV. Here is an example from the recipe meta/recipes-kernel/blktrace/blktrace_git.bb: SRCREV = "d6918c8832793b4205ed3bfede78c2f915c23385" PR = "r6" PV = "1.0.5+git${SRCPV}" SRC_URI = "git://git.kernel.dk/blktrace.git \ file://ldflags.patch" If your SRC_URI statement includes URLs pointing to individual files fetched from a remote server other than a version control system, BitBake attempts to verify the files against checksums defined in your recipe to ensure they have not been tampered with or otherwise modified since the recipe was written. Two checksums are used: SRC_URI[md5sum] and SRC_URI[sha256sum]. If your SRC_URI variable points to more than a single URL (excluding SCM URLs), you need to provide the md5 and sha256 checksums for each URL. For these cases, you provide a name for each URL as part of the SRC_URI and then reference that name in the subsequent checksum statements. Here is an example: SRC_URI = "${DEBIAN_MIRROR}/main/a/apmd/apmd_3.2.2.orig.tar.gz;name=tarball \ ${DEBIAN_MIRROR}/main/a/apmd/apmd_${PV}.diff.gz;name=patch" SRC_URI[tarball.md5sum] = "b1e6309e8331e0f4e6efd311c2d97fa8" SRC_URI[tarball.sha256sum] = "7f7d9f60b7766b852881d40b8ff91d8e39fccb0d1d913102a5c75a2dbb52332d" SRC_URI[patch.md5sum] = "57e1b689264ea80f78353519eece0c92" SRC_URI[patch.sha256sum] = "7905ff96be93d725544d0040e425c42f9c05580db3c272f11cff75b9aa89d430" Proper values for md5 and sha256 checksums might be available with other signatures on the download page for the upstream source (e.g. md5, sha1, sha256, GPG, and so forth). Because the OpenEmbedded build system only deals with sha256sum and md5sum, you should verify all the signatures you find by hand. If no SRC_URI checksums are specified when you attempt to build the recipe, or you provide an incorrect checksum, the build will produce an error for each missing or incorrect checksum. As part of the error message, the build system provides the checksum string corresponding to the fetched file. Once you have the correct checksums, you can copy and paste them into your recipe and then run the build again to continue. As mentioned, if the upstream source provides signatures for verifying the downloaded source code, you should verify those manually before setting the checksum values in the recipe and continuing with the build. This final example is a bit more complicated and is from the meta/recipes-sato/rxvt-unicode/rxvt-unicode_9.20.bb recipe. The example's SRC_URI statement identifies multiple files as the source files for the recipe: a tarball, a patch file, a desktop file, and an icon. SRC_URI = "http://dist.schmorp.de/rxvt-unicode/Attic/rxvt-unicode-${PV}.tar.bz2 \ file://xwc.patch \ file://rxvt.desktop \ file://rxvt.png" When you specify local files using the file:// URI protocol, the build system fetches files from the local machine. The path is relative to the FILESPATH variable and searches specific directories in a certain order: ${BP}, ${BPN}, and files. The directories are assumed to be subdirectories of the directory in which the recipe or append file resides. For another example that specifies these types of files, see the "Single .c File Package (Hello World!)" section. The previous example also specifies a patch file. Patch files are files whose names usually end in .patch or .diff but can end with compressed suffixes such as diff.gz and patch.bz2, for example. The build system automatically applies patches as described in the "Patching Code" section.

During the build, the do_unpack task unpacks the source with ${S} pointing to where it is unpacked. If you are fetching your source files from an upstream source archived tarball and the tarball's internal structure matches the common convention of a top-level subdirectory named ${BPN}-${PV}, then you do not need to set S. However, if SRC_URI specifies to fetch source from an archive that does not use this convention, or from an SCM like Git or Subversion, your recipe needs to define S. If processing your recipe using BitBake successfully unpacks the source files, you need to be sure that the directory pointed to by ${S} matches the structure of the source.

Sometimes it is necessary to patch code after it has been fetched. Any files mentioned in SRC_URI whose names end in .patch or .diff or compressed versions of these suffixes (e.g. diff.gz are treated as patches. The do_patch task automatically applies these patches. The build system should be able to apply patches with the "-p1" option (i.e. one directory level in the path will be stripped off). If your patch needs to have more directory levels stripped off, specify the number of levels using the "striplevel" option in the SRC_URI entry for the patch. Alternatively, if your patch needs to be applied in a specific subdirectory that is not specified in the patch file, use the "patchdir" option in the entry. As with all local files referenced in SRC_URI using file://, you should place patch files in a directory next to the recipe either named the same as the base name of the recipe (BP and BPN) or "files".

Your recipe needs to have both the LICENSE and LIC_FILES_CHKSUM variables: LICENSE: This variable specifies the license for the software. If you do not know the license under which the software you are building is distributed, you should go to the source code and look for that information. Typical files containing this information include COPYING, LICENSE, and README files. You could also find the information near the top of a source file. For example, given a piece of software licensed under the GNU General Public License version 2, you would set LICENSE as follows: LICENSE = "GPLv2" The licenses you specify within LICENSE can have any name as long as you do not use spaces, since spaces are used as separators between license names. For standard licenses, use the names of the files in meta/files/common-licenses/ or the SPDXLICENSEMAP flag names defined in meta/conf/licenses.conf. LIC_FILES_CHKSUM: The OpenEmbedded build system uses this variable to make sure the license text has not changed. If it has, the build produces an error and it affords you the chance to figure it out and correct the problem. You need to specify all applicable licensing files for the software. At the end of the configuration step, the build process will compare the checksums of the files to be sure the text has not changed. Any differences result in an error with the message containing the current checksum. For more explanation and examples of how to set the LIC_FILES_CHKSUM variable, see the "Tracking License Changes" section in the Yocto Project Reference Manual. To determine the correct checksum string, you can list the appropriate files in the LIC_FILES_CHKSUM variable with incorrect md5 strings, attempt to build the software, and then note the resulting error messages that will report the correct md5 strings. See the "Fetching Code" section for additional information. Here is an example that assumes the software has a COPYING file: LIC_FILES_CHKSUM = "file://COPYING;md5=xxx" When you try to build the software, the build system will produce an error and give you the correct string that you can substitute into the recipe file for a subsequent build.

Most software packages have a short list of other packages that they require, which are called dependencies. These dependencies fall into two main categories: build-time dependencies, which are required when the software is built; and runtime dependencies, which are required to be installed on the target in order for the software to run. Within a recipe, you specify build-time dependencies using the DEPENDS variable. Although nuances exist, items specified in DEPENDS should be names of other recipes. It is important that you specify all build-time dependencies explicitly. If you do not, due to the parallel nature of BitBake's execution, you can end up with a race condition where the dependency is present for one task of a recipe (e.g. do_configure) and then gone when the next task runs (e.g. do_compile). Another consideration is that configure scripts might automatically check for optional dependencies and enable corresponding functionality if those dependencies are found. This behavior means that to ensure deterministic results and thus avoid more race conditions, you need to either explicitly specify these dependencies as well, or tell the configure script explicitly to disable the functionality. If you wish to make a recipe that is more generally useful (e.g. publish the recipe in a layer for others to use), instead of hard-disabling the functionality, you can use the PACKAGECONFIG variable to allow functionality and the corresponding dependencies to be enabled and disabled easily by other users of the recipe. Similar to build-time dependencies, you specify runtime dependencies through a variable - RDEPENDS, which is package-specific. All variables that are package-specific need to have the name of the package added to the end as an override. Since the main package for a recipe has the same name as the recipe, and the recipe's name can be found through the ${PN} variable, then you specify the dependencies for the main package by setting RDEPENDS_${PN}. If the package were named ${PN}-tools, then you would set RDEPENDS_${PN}-tools, and so forth. Some runtime dependencies will be set automatically at packaging time. These dependencies include any shared library dependencies (i.e. if a package "example" contains "libexample" and another package "mypackage" contains a binary that links to "libexample" then the OpenEmbedded build system will automatically add a runtime dependency to "mypackage" on "example"). See the "Automatically Added Runtime Dependencies" in the Yocto Project Reference Manual for further details.

Most software provides some means of setting build-time configuration options before compilation. Typically, setting these options is accomplished by running a configure script with some options, or by modifying a build configuration file. As of Yocto Project Release 1.7, some of the core recipes that package binary configuration scripts now disable the scripts due to the scripts previously requiring error-prone path substitution. The OpenEmbedded build system uses pkg-config now, which is much more robust. You can find a list of the *-config scripts that are disabled list in the "Binary Configuration Scripts Disabled" section in the Yocto Project Reference Manual. A major part of build-time configuration is about checking for build-time dependencies and possibly enabling optional functionality as a result. You need to specify any build-time dependencies for the software you are building in your recipe's DEPENDS value, in terms of other recipes that satisfy those dependencies. You can often find build-time or runtime dependencies described in the software's documentation. The following list provides configuration items of note based on how your software is built: Autotools: If your source files have a configure.ac file, then your software is built using Autotools. If this is the case, you just need to worry about modifying the configuration. When using Autotools, your recipe needs to inherit the autotools class and your recipe does not have to contain a do_configure task. However, you might still want to make some adjustments. For example, you can set EXTRA_OECONF or PACKAGECONFIG_CONFARGS to pass any needed configure options that are specific to the recipe. CMake: If your source files have a CMakeLists.txt file, then your software is built using CMake. If this is the case, you just need to worry about modifying the configuration. When you use CMake, your recipe needs to inherit the cmake class and your recipe does not have to contain a do_configure task. You can make some adjustments by setting EXTRA_OECMAKE to pass any needed configure options that are specific to the recipe. Other: If your source files do not have a configure.ac or CMakeLists.txt file, then your software is built using some method other than Autotools or CMake. If this is the case, you normally need to provide a do_configure task in your recipe unless, of course, there is nothing to configure. Even if your software is not being built by Autotools or CMake, you still might not need to deal with any configuration issues. You need to determine if configuration is even a required step. You might need to modify a Makefile or some configuration file used for the build to specify necessary build options. Or, perhaps you might need to run a provided, custom configure script with the appropriate options. For the case involving a custom configure script, you would run ./configure --help and look for the options you need to set. Once configuration succeeds, it is always good practice to look at the log.do_configure file to ensure that the appropriate options have been enabled and no additional build-time dependencies need to be added to DEPENDS. For example, if the configure script reports that it found something not mentioned in DEPENDS, or that it did not find something that it needed for some desired optional functionality, then you would need to add those to DEPENDS. Looking at the log might also reveal items being checked for, enabled, or both that you do not want, or items not being found that are in DEPENDS, in which case you would need to look at passing extra options to the configure script as needed. For reference information on configure options specific to the software you are building, you can consult the output of the ./configure --help command within ${S} or consult the software's upstream documentation.

If your recipe builds an application that needs to communicate with some device or needs an API into a custom kernel, you will need to provide appropriate header files. Under no circumstances should you ever modify the existing meta/recipes-kernel/linux-libc-headers/linux-libc-headers.inc file. These headers are used to build libc and must not be compromised with custom or machine-specific header information. If you customize libc through modified headers all other applications that use libc thus become affected. Never copy and customize the libc header file (i.e. meta/recipes-kernel/linux-libc-headers/linux-libc-headers.inc). The correct way to interface to a device or custom kernel is to use a separate package that provides the additional headers for the driver or other unique interfaces. When doing so, your application also becomes responsible for creating a dependency on that specific provider. Consider the following: Never modify linux-libc-headers.inc. Consider that file to be part of the libc system, and not something you use to access the kernel directly. You should access libc through specific libc calls. Applications that must talk directly to devices should either provide necessary headers themselves, or establish a dependency on a special headers package that is specific to that driver. For example, suppose you want to modify an existing header that adds I/O control or network support. If the modifications are used by a small number programs, providing a unique version of a header is easy and has little impact. When doing so, bear in mind the guidelines in the previous list. If for some reason your changes need to modify the behavior of the libc, and subsequently all other applications on the system, use a .bbappend to modify the linux-kernel-headers.inc file. However, take care to not make the changes machine specific. Consider a case where your kernel is older and you need an older libc ABI. The headers installed by your recipe should still be a standard mainline kernel, not your own custom one. When you use custom kernel headers you need to get them from STAGING_KERNEL_DIR, which is the directory with kernel headers that are required to build out-of-tree modules. Your recipe will also need the following: do_configure[depends] += "virtual/kernel:do_shared_workdir"

During a build, the do_compile task happens after source is fetched, unpacked, and configured. If the recipe passes through do_compile successfully, nothing needs to be done. However, if the compile step fails, you need to diagnose the failure. Here are some common issues that cause failures. For cases where improper paths are detected for configuration files or for when libraries/headers cannot be found, be sure you are using the more robust pkg-config. See the note in section "Configuring the Recipe" for additional information. Parallel build failures: These failures manifest themselves as intermittent errors, or errors reporting that a file or directory that should be created by some other part of the build process could not be found. This type of failure can occur even if, upon inspection, the file or directory does exist after the build has failed, because that part of the build process happened in the wrong order. To fix the problem, you need to either satisfy the missing dependency in the Makefile or whatever script produced the Makefile, or (as a workaround) set PARALLEL_MAKE to an empty string: PARALLEL_MAKE = "" For information on parallel Makefile issues, see the "Debugging Parallel Make Races" section. Improper host path usage: This failure applies to recipes building for the target or nativesdk only. The failure occurs when the compilation process uses improper headers, libraries, or other files from the host system when cross-compiling for the target. To fix the problem, examine the log.do_compile file to identify the host paths being used (e.g. /usr/include, /usr/lib, and so forth) and then either add configure options, apply a patch, or do both. Failure to find required libraries/headers: If a build-time dependency is missing because it has not been declared in DEPENDS, or because the dependency exists but the path used by the build process to find the file is incorrect and the configure step did not detect it, the compilation process could fail. For either of these failures, the compilation process notes that files could not be found. In these cases, you need to go back and add additional options to the configure script as well as possibly add additional build-time dependencies to DEPENDS. Occasionally, it is necessary to apply a patch to the source to ensure the correct paths are used. If you need to specify paths to find files staged into the sysroot from other recipes, use the variables that the OpenEmbedded build system provides (e.g. STAGING_BINDIR, STAGING_INCDIR, STAGING_DATADIR, and so forth).

During do_install, the task copies the built files along with their hierarchy to locations that would mirror their locations on the target device. The installation process copies files from the ${S}, ${B}, and ${WORKDIR} directories to the ${D} directory to create the structure as it should appear on the target system. How your software is built affects what you must do to be sure your software is installed correctly. The following list describes what you must do for installation depending on the type of build system used by the software being built: Autotools and CMake: If the software your recipe is building uses Autotools or CMake, the OpenEmbedded build system understands how to install the software. Consequently, you do not have to have a do_install task as part of your recipe. You just need to make sure the install portion of the build completes with no issues. However, if you wish to install additional files not already being installed by make install, you should do this using a do_install_append function using the install command as described in the "Manual" bulleted item later in this list. Other (using make install): You need to define a do_install function in your recipe. The function should call oe_runmake install and will likely need to pass in the destination directory as well. How you pass that path is dependent on how the Makefile being run is written (e.g. DESTDIR=${D}, PREFIX=${D}, INSTALLROOT=${D}, and so forth). For an example recipe using make install, see the "Makefile-Based Package" section. Manual: You need to define a do_install function in your recipe. The function must first use install -d to create the directories under ${D}. Once the directories exist, your function can use install to manually install the built software into the directories. You can find more information on install at . For the scenarios that do not use Autotools or CMake, you need to track the installation and diagnose and fix any issues until everything installs correctly. You need to look in the default location of ${D}, which is ${WORKDIR}/image, to be sure your files have been installed correctly. During the installation process, you might need to modify some of the installed files to suit the target layout. For example, you might need to replace hard-coded paths in an initscript with values of variables provided by the build system, such as replacing /usr/bin/ with ${bindir}. If you do perform such modifications during do_install, be sure to modify the destination file after copying rather than before copying. Modifying after copying ensures that the build system can re-execute do_install if needed. oe_runmake install, which can be run directly or can be run indirectly by the autotools and cmake classes, runs make install in parallel. Sometimes, a Makefile can have missing dependencies between targets that can result in race conditions. If you experience intermittent failures during do_install, you might be able to work around them by disabling parallel Makefile installs by adding the following to the recipe: PARALLEL_MAKEINST = "" See PARALLEL_MAKEINST for additional information.

If you want to install a service, which is a process that usually starts on boot and runs in the background, then you must include some additional definitions in your recipe. If you are adding services and the service initialization script or the service file itself is not installed, you must provide for that installation in your recipe using a do_install_append function. If your recipe already has a do_install function, update the function near its end rather than adding an additional do_install_append function. When you create the installation for your services, you need to accomplish what is normally done by make install. In other words, make sure your installation arranges the output similar to how it is arranged on the target system. The OpenEmbedded build system provides support for starting services two different ways: SysVinit: SysVinit is a system and service manager that manages the init system used to control the very basic functions of your system. The init program is the first program started by the Linux kernel when the system boots. Init then controls the startup, running and shutdown of all other programs. To enable a service using SysVinit, your recipe needs to inherit the update-rc.d class. The class helps facilitate safely installing the package on the target. You will need to set the INITSCRIPT_PACKAGES, INITSCRIPT_NAME, and INITSCRIPT_PARAMS variables within your recipe. systemd: System Management Daemon (systemd) was designed to replace SysVinit and to provide enhanced management of services. For more information on systemd, see the systemd homepage at . To enable a service using systemd, your recipe needs to inherit the systemd class. See the systemd.bbclass file located in your Source Directory. section for more information.

Successful packaging is a combination of automated processes performed by the OpenEmbedded build system and some specific steps you need to take. The following list describes the process: Splitting Files: The do_package task splits the files produced by the recipe into logical components. Even software that produces a single binary might still have debug symbols, documentation, and other logical components that should be split out. The do_package task ensures that files are split up and packaged correctly. Running QA Checks: The insane class adds a step to the package generation process so that output quality assurance checks are generated by the OpenEmbedded build system. This step performs a range of checks to be sure the build's output is free of common problems that show up during runtime. For information on these checks, see the insane class and the "QA Error and Warning Messages" chapter in the Yocto Project Reference Manual. Hand-Checking Your Packages: After you build your software, you need to be sure your packages are correct. Examine the ${WORKDIR}/packages-split directory and make sure files are where you expect them to be. If you discover problems, you can set PACKAGES, FILES, do_install(_append), and so forth as needed. Splitting an Application into Multiple Packages: If you need to split an application into several packages, see the "Splitting an Application into Multiple Packages" section for an example. Installing a Post-Installation Script: For an example showing how to install a post-installation script, see the "Post-Installation Scripts" section. Marking Package Architecture: Depending on what your recipe is building and how it is configured, it might be important to mark the packages produced as being specific to a particular machine, or to mark them as not being specific to a particular machine or architecture at all. By default, packages apply to any machine with the same architecture as the target machine. When a recipe produces packages that are machine-specific (e.g. the MACHINE value is passed into the configure script or a patch is applied only for a particular machine), you should mark them as such by adding the following to the recipe: PACKAGE_ARCH = "${MACHINE_ARCH}" On the other hand, if the recipe produces packages that do not contain anything specific to the target machine or architecture at all (e.g. recipes that simply package script files or configuration files), you should use the allarch class to do this for you by adding this to your recipe: inherit allarch Ensuring that the package architecture is correct is not critical while you are doing the first few builds of your recipe. However, it is important in order to ensure that your recipe rebuilds (or does not rebuild) appropriately in response to changes in configuration, and to ensure that you get the appropriate packages installed on the target machine, particularly if you run separate builds for more than one target machine.

Recipes often need to use files provided by other recipes on the build host. For example, an application linking to a common library needs access to the library itself and its associated headers. The way this access is accomplished is by populating a sysroot with files. Each recipe has two sysroots in its work directory, one for target files (recipe-sysroot) and one for files that are native to the build host (recipe-sysroot-native). You could find the term "staging" used within the Yocto project regarding files populating sysroots (e.g. the STAGING_DIR variable). Recipes should never populate the sysroot directly (i.e. write files into sysroot). Instead, files should be installed into standard locations during the do_install task within the ${D} directory. The reason for this limitation is that almost all files that populate the sysroot are cataloged in manifests in order to ensure the files can be removed later when a recipe is either modified or removed. Thus, the sysroot is able to remain free from stale files. A subset of the files installed by the do_install task are used by the do_populate_sysroot task as defined by the the SYSROOT_DIRS variable to automatically populate the sysroot. It is possible to modify the list of directories that populate the sysroot. The following example shows how you could add the /opt directory to the list of directories within a recipe: SYSROOT_DIRS += "/opt" For a more complete description of the do_populate_sysroot task and its associated functions, see the staging class.

Sometimes the name of a recipe can lead to versioning problems when the recipe is upgraded to a final release. For example, consider the irssi_0.8.16-rc1.bb recipe file in the list of example recipes in the "Storing and Naming the Recipe" section. This recipe is at a release candidate stage (i.e. "rc1"). When the recipe is released, the recipe filename becomes irssi_0.8.16.bb. The version change from 0.8.16-rc1 to 0.8.16 is seen as a decrease by the build system and package managers, so the resulting packages will not correctly trigger an upgrade. In order to ensure the versions compare properly, the recommended convention is to set PV within the recipe to "previous_version+current_version". You can use an additional variable so that you can use the current version elsewhere. Here is an example: REALPV = "0.8.16-rc1" PV = "0.8.15+${REALPV}"

Post-installation scripts run immediately after installing a package on the target or during image creation when a package is included in an image. To add a post-installation script to a package, add a pkg_postinst_PACKAGENAME() function to the recipe file (.bb) and replace PACKAGENAME with the name of the package you want to attach to the postinst script. To apply the post-installation script to the main package for the recipe, which is usually what is required, specify ${PN} in place of PACKAGENAME. A post-installation function has the following structure: pkg_postinst_PACKAGENAME() { # Commands to carry out } The script defined in the post-installation function is called when the root filesystem is created. If the script succeeds, the package is marked as installed. If the script fails, the package is marked as unpacked and the script is executed when the image boots again. Any RPM post-installation script that runs on the target should return a 0 exit code. RPM does not allow non-zero exit codes for these scripts, and the RPM package manager will cause the package to fail installation on the target. Sometimes it is necessary for the execution of a post-installation script to be delayed until the first boot. For example, the script might need to be executed on the device itself. To delay script execution until boot time, use the following structure in the post-installation script: pkg_postinst_PACKAGENAME() { if [ x"$D" = "x" ]; then # Actions to carry out on the device go here else exit 1 fi } The previous example delays execution until the image boots again because the environment variable D points to the directory containing the image when the root filesystem is created at build time but is unset when executed on the first boot. If you have recipes that use pkg_postinst scripts and they require the use of non-standard native tools that have dependencies during rootfs construction, you need to use the PACKAGE_WRITE_DEPS variable in your recipe to list these tools. If you do not use this variable, the tools might be missing and execution of the post-installation script is deferred until first boot. Deferring the script to first boot is undesirable and for read-only rootfs impossible. Equivalent support for pre-install, pre-uninstall, and post-uninstall scripts exist by way of pkg_preinst, pkg_prerm, and pkg_postrm, respectively. These scrips work in exactly the same way as does pkg_postinst with the exception that they run at different times. Also, because of when they run, they are not applicable to being run at image creation time like pkg_postinst.

The final step for completing your recipe is to be sure that the software you built runs correctly. To accomplish runtime testing, add the build's output packages to your image and test them on the target. For information on how to customize your image by adding specific packages, see the "Customizing Images" section.

To help summarize how to write a recipe, this section provides some examples given various scenarios: Recipes that use local files Using an Autotooled package Using a Makefile-based package Splitting an application into multiple packages Adding binaries to an image

Building an application from a single file that is stored locally (e.g. under files) requires a recipe that has the file listed in the SRC_URI variable. Additionally, you need to manually write the do_compile and do_install tasks. The S variable defines the directory containing the source code, which is set to WORKDIR in this case - the directory BitBake uses for the build. SUMMARY = "Simple helloworld application" SECTION = "examples" LICENSE = "MIT" LIC_FILES_CHKSUM = "file://${COMMON_LICENSE_DIR}/MIT;md5=0835ade698e0bcf8506ecda2f7b4f302" SRC_URI = "file://helloworld.c" S = "${WORKDIR}" do_compile() { ${CC} helloworld.c -o helloworld } do_install() { install -d ${D}${bindir} install -m 0755 helloworld ${D}${bindir} } By default, the helloworld, helloworld-dbg, and helloworld-dev packages are built. For information on how to customize the packaging process, see the "Splitting an Application into Multiple Packages" section.

Applications that use Autotools such as autoconf and automake require a recipe that has a source archive listed in SRC_URI and also inherit the autotools class, which contains the definitions of all the steps needed to build an Autotool-based application. The result of the build is automatically packaged. And, if the application uses NLS for localization, packages with local information are generated (one package per language). Following is one example: (hello_2.3.bb) SUMMARY = "GNU Helloworld application" SECTION = "examples" LICENSE = "GPLv2+" LIC_FILES_CHKSUM = "file://COPYING;md5=751419260aa954499f7abaabaa882bbe" SRC_URI = "${GNU_MIRROR}/hello/hello-${PV}.tar.gz" inherit autotools gettext The variable LIC_FILES_CHKSUM is used to track source license changes as described in the "Tracking License Changes" section. You can quickly create Autotool-based recipes in a manner similar to the previous example.

Applications that use GNU make also require a recipe that has the source archive listed in SRC_URI. You do not need to add a do_compile step since by default BitBake starts the make command to compile the application. If you need additional make options, you should store them in the EXTRA_OEMAKE or PACKAGECONFIG_CONFARGS variables. BitBake passes these options into the GNU make invocation. Note that a do_install task is still required. Otherwise, BitBake runs an empty do_install task by default. Some applications might require extra parameters to be passed to the compiler. For example, the application might need an additional header path. You can accomplish this by adding to the CFLAGS variable. The following example shows this: CFLAGS_prepend = "-I ${S}/include " In the following example, mtd-utils is a makefile-based package: SUMMARY = "Tools for managing memory technology devices" SECTION = "base" DEPENDS = "zlib lzo e2fsprogs util-linux" HOMEPAGE = "http://www.linux-mtd.infradead.org/" LICENSE = "GPLv2+" LIC_FILES_CHKSUM = "file://COPYING;md5=0636e73ff0215e8d672dc4c32c317bb3 \ file://include/common.h;beginline=1;endline=17;md5=ba05b07912a44ea2bf81ce409380049c" # Use the latest version at 26 Oct, 2013 SRCREV = "9f107132a6a073cce37434ca9cda6917dd8d866b" SRC_URI = "git://git.infradead.org/mtd-utils.git \ file://add-exclusion-to-mkfs-jffs2-git-2.patch \ " PV = "1.5.1+git${SRCPV}" S = "${WORKDIR}/git" EXTRA_OEMAKE = "'CC=${CC}' 'RANLIB=${RANLIB}' 'AR=${AR}' 'CFLAGS=${CFLAGS} -I${S}/include -DWITHOUT_XATTR' 'BUILDDIR=${S}'" do_install () { oe_runmake install DESTDIR=${D} SBINDIR=${sbindir} MANDIR=${mandir} INCLUDEDIR=${includedir} } PACKAGES =+ "mtd-utils-jffs2 mtd-utils-ubifs mtd-utils-misc" FILES_mtd-utils-jffs2 = "${sbindir}/mkfs.jffs2 ${sbindir}/jffs2dump ${sbindir}/jffs2reader ${sbindir}/sumtool" FILES_mtd-utils-ubifs = "${sbindir}/mkfs.ubifs ${sbindir}/ubi*" FILES_mtd-utils-misc = "${sbindir}/nftl* ${sbindir}/ftl* ${sbindir}/rfd* ${sbindir}/doc* ${sbindir}/serve_image ${sbindir}/recv_image" PARALLEL_MAKE = "" BBCLASSEXTEND = "native"

You can use the variables PACKAGES and FILES to split an application into multiple packages. Following is an example that uses the libxpm recipe. By default, this recipe generates a single package that contains the library along with a few binaries. You can modify the recipe to split the binaries into separate packages: require xorg-lib-common.inc SUMMARY = "Xpm: X Pixmap extension library" LICENSE = "BSD" LIC_FILES_CHKSUM = "file://COPYING;md5=51f4270b012ecd4ab1a164f5f4ed6cf7" DEPENDS += "libxext libsm libxt" PE = "1" XORG_PN = "libXpm" PACKAGES =+ "sxpm cxpm" FILES_cxpm = "${bindir}/cxpm" FILES_sxpm = "${bindir}/sxpm" In the previous example, we want to ship the sxpm and cxpm binaries in separate packages. Since bindir would be packaged into the main PN package by default, we prepend the PACKAGES variable so additional package names are added to the start of list. This results in the extra FILES_* variables then containing information that define which files and directories go into which packages. Files included by earlier packages are skipped by latter packages. Thus, the main PN package does not include the above listed files.

Sometimes, you need to add pre-compiled binaries to an image. For example, suppose that binaries for proprietary code exist, which are created by a particular division of a company. Your part of the company needs to use those binaries as part of an image that you are building using the OpenEmbedded build system. Since you only have the binaries and not the source code, you cannot use a typical recipe that expects to fetch the source specified in SRC_URI and then compile it. One method is to package the binaries and then install them as part of the image. Generally, it is not a good idea to package binaries since, among other things, it can hinder the ability to reproduce builds and could lead to compatibility problems with ABI in the future. However, sometimes you have no choice. The easiest solution is to create a recipe that uses the bin_package class and to be sure that you are using default locations for build artifacts. In most cases, the bin_package class handles "skipping" the configure and compile steps as well as sets things up to grab packages from the appropriate area. In particular, this class sets noexec on both the do_configure and do_compile tasks, sets FILES_${PN} to "/" so that it picks up all files, and sets up a do_install task, which effectively copies all files from ${S} to ${D}. The bin_package class works well when the files extracted into ${S} are already laid out in the way they should be laid out on the target. For more information on these variables, see the FILES, PN, S, and D variables in the Yocto Project Reference Manual's variable glossary. Using DEPENDS is a good idea even for components distributed in binary form, and is often necessary for shared libraries. For a shared library, listing the library dependencies in DEPENDS makes sure that the libraries are available in the staging sysroot when other recipes link against the library, which might be necessary for successful linking. Using DEPENDS also allows runtime dependencies between packages to be added automatically. See the "Automatically Added Runtime Dependencies" section in the Yocto Project Reference Manual for more information. If you cannot use the bin_package class, you need to be sure you are doing the following: Create a recipe where the do_configure and do_compile tasks do nothing: It is usually sufficient to just not define these tasks in the recipe, because the default implementations do nothing unless a Makefile is found in ${S}. If ${S} might contain a Makefile, or if you inherit some class that replaces do_configure and do_compile with custom versions, then you can use the [noexec] flag to turn the tasks into no-ops, as follows: do_configure[noexec] = "1" do_compile[noexec] = "1" Unlike deleting the tasks, using the flag preserves the dependency chain from the do_fetch, do_unpack, and do_patch tasks to the do_install task. Make sure your do_install task installs the binaries appropriately. Ensure that you set up FILES (usually FILES_${PN}) to point to the files you have installed, which of course depends on where you have installed them and whether those files are in different locations than the defaults.

When writing recipes, it is good to conform to existing style guidelines. The OpenEmbedded Styleguide wiki page provides rough guidelines for preferred recipe style. It is common for existing recipes to deviate a bit from this style. However, aiming for at least a consistent style is a good idea. Some practices, such as omitting spaces around = operators in assignments or ordering recipe components in an erratic way, are widely seen as poor style.

Adding a new machine to the Yocto Project is a straightforward process. This section describes how to add machines that are similar to those that the Yocto Project already supports. Although well within the capabilities of the Yocto Project, adding a totally new architecture might require changes to gcc/glibc and to the site information, which is beyond the scope of this manual. For a complete example that shows how to add a new machine, see the "Creating a New BSP Layer Using the yocto-bsp Script" section in the Yocto Project Board Support Package (BSP) Developer's Guide.

To add a new machine, you need to add a new machine configuration file to the layer's conf/machine directory. This configuration file provides details about the device you are adding. The OpenEmbedded build system uses the root name of the machine configuration file to reference the new machine. For example, given a machine configuration file named crownbay.conf, the build system recognizes the machine as "crownbay". The most important variables you must set in your machine configuration file or include from a lower-level configuration file are as follows: TARGET_ARCH (e.g. "arm") PREFERRED_PROVIDER_virtual/kernel MACHINE_FEATURES (e.g. "apm screen wifi") You might also need these variables: SERIAL_CONSOLES (e.g. "115200;ttyS0 115200;ttyS1") KERNEL_IMAGETYPE (e.g. "zImage") IMAGE_FSTYPES (e.g. "tar.gz jffs2") You can find full details on these variables in the reference section. You can leverage existing machine .conf files from meta-yocto-bsp/conf/machine/.

The OpenEmbedded build system needs to be able to build a kernel for the machine. You need to either create a new kernel recipe for this machine, or extend an existing kernel recipe. You can find several kernel recipe examples in the Source Directory at meta/recipes-kernel/linux that you can use as references. If you are creating a new kernel recipe, normal recipe-writing rules apply for setting up a SRC_URI. Thus, you need to specify any necessary patches and set S to point at the source code. You need to create a do_configure task that configures the unpacked kernel with a defconfig file. You can do this by using a make defconfig command or, more commonly, by copying in a suitable defconfig file and then running make oldconfig. By making use of inherit kernel and potentially some of the linux-*.inc files, most other functionality is centralized and the defaults of the class normally work well. If you are extending an existing kernel recipe, it is usually a matter of adding a suitable defconfig file. The file needs to be added into a location similar to defconfig files used for other machines in a given kernel recipe. A possible way to do this is by listing the file in the SRC_URI and adding the machine to the expression in COMPATIBLE_MACHINE: COMPATIBLE_MACHINE = '(qemux86|qemumips)' For more information on defconfig files, see the "Changing the Configuration" section in the Yocto Project Linux Kernel Development Manual.

A formfactor configuration file provides information about the target hardware for which the image is being built and information that the build system cannot obtain from other sources such as the kernel. Some examples of information contained in a formfactor configuration file include framebuffer orientation, whether or not the system has a keyboard, the positioning of the keyboard in relation to the screen, and the screen resolution. The build system uses reasonable defaults in most cases. However, if customization is necessary, you need to create a machconfig file in the meta/recipes-bsp/formfactor/files directory. This directory contains directories for specific machines such as qemuarm and qemux86. For information about the settings available and the defaults, see the meta/recipes-bsp/formfactor/files/config file found in the same area. Following is an example for "qemuarm" machine: HAVE_TOUCHSCREEN=1 HAVE_KEYBOARD=1 DISPLAY_CAN_ROTATE=0 DISPLAY_ORIENTATION=0 #DISPLAY_WIDTH_PIXELS=640 #DISPLAY_HEIGHT_PIXELS=480 #DISPLAY_BPP=16 DISPLAY_DPI=150 DISPLAY_SUBPIXEL_ORDER=vrgb

You might find it helpful during development to modify the temporary source code used by recipes to build packages. For example, suppose you are developing a patch and you need to experiment a bit to figure out your solution. After you have initially built the package, you can iteratively tweak the source code, which is located in the Build Directory, and then you can force a re-compile and quickly test your altered code. Once you settle on a solution, you can then preserve your changes in the form of patches. During a build, the unpacked temporary source code used by recipes to build packages is available in the Build Directory as defined by the S variable. Below is the default value for the S variable as defined in the meta/conf/bitbake.conf configuration file in the Source Directory: S = "${WORKDIR}/${BP}" You should be aware that many recipes override the S variable. For example, recipes that fetch their source from Git usually set S to ${WORKDIR}/git. The BP represents the base recipe name, which consists of the name and version: BP = "${BPN}-${PV}" The path to the work directory for the recipe (WORKDIR) is defined as follows: ${TMPDIR}/work/${MULTIMACH_TARGET_SYS}/${PN}/${EXTENDPE}${PV}-${PR} The actual directory depends on several things: TMPDIR: The top-level build output directory. MULTIMACH_TARGET_SYS: The target system identifier. PN: The recipe name. EXTENDPE: The epoch - (if PE is not specified, which is usually the case for most recipes, then EXTENDPE is blank). PV: The recipe version. PR: The recipe revision. As an example, assume a Source Directory top-level folder named poky, a default Build Directory at poky/build, and a qemux86-poky-linux machine target system. Furthermore, suppose your recipe is named foo_1.3.0.bb. In this case, the work directory the build system uses to build the package would be as follows: poky/build/tmp/work/qemux86-poky-linux/foo/1.3.0-r0

Quilt is a powerful tool that allows you to capture source code changes without having a clean source tree. This section outlines the typical workflow you can use to modify source code, test changes, and then preserve the changes in the form of a patch all using Quilt. With regard to preserving changes to source files, if you clean a recipe or have rm_work enabled, the devtool workflow as described in the Yocto Project Application Development and the Extensible Software Development Kit (eSDK) manual is a safer development flow than the flow that uses Quilt. Follow these general steps: Find the Source Code: Temporary source code used by the OpenEmbedded build system is kept in the Build Directory. See the "Finding Temporary Source Code" section to learn how to locate the directory that has the temporary source code for a particular package. Change Your Working Directory: You need to be in the directory that has the temporary source code. That directory is defined by the S variable. Create a New Patch: Before modifying source code, you need to create a new patch. To create a new patch file, use quilt new as below: $ quilt new my_changes.patch Notify Quilt and Add Files: After creating the patch, you need to notify Quilt about the files you plan to edit. You notify Quilt by adding the files to the patch you just created: $ quilt add file1.c file2.c file3.c Edit the Files: Make your changes in the source code to the files you added to the patch. Test Your Changes: Once you have modified the source code, the easiest way to test your changes is by calling the do_compile task as shown in the following example: $ bitbake -c compile -f package The -f or --force option forces the specified task to execute. If you find problems with your code, you can just keep editing and re-testing iteratively until things work as expected. All the modifications you make to the temporary source code disappear once you run the do_clean or do_cleanall tasks using BitBake (i.e. bitbake -c clean package and bitbake -c cleanall package). Modifications will also disappear if you use the rm_work feature as described in the "Building Images" section of the Yocto Project Quick Start. Generate the Patch: Once your changes work as expected, you need to use Quilt to generate the final patch that contains all your modifications. $ quilt refresh At this point, the my_changes.patch file has all your edits made to the file1.c, file2.c, and file3.c files. You can find the resulting patch file in the patches/ subdirectory of the source (S) directory. Copy the Patch File: For simplicity, copy the patch file into a directory named files, which you can create in the same directory that holds the recipe (.bb) file or the append (.bbappend) file. Placing the patch here guarantees that the OpenEmbedded build system will find the patch. Next, add the patch into the SRC_URI of the recipe. Here is an example: SRC_URI += "file://my_changes.patch"

When debugging certain commands or even when just editing packages, devshell can be a useful tool. When you invoke devshell, all tasks up to and including do_patch are run for the specified target. Then, a new terminal is opened and you are placed in ${S}, the source directory. In the new terminal, all the OpenEmbedded build-related environment variables are still defined so you can use commands such as configure and make. The commands execute just as if the OpenEmbedded build system were executing them. Consequently, working this way can be helpful when debugging a build or preparing software to be used with the OpenEmbedded build system. Following is an example that uses devshell on a target named matchbox-desktop: $ bitbake matchbox-desktop -c devshell This command spawns a terminal with a shell prompt within the OpenEmbedded build environment. The OE_TERMINAL variable controls what type of shell is opened. For spawned terminals, the following occurs: The PATH variable includes the cross-toolchain. The pkgconfig variables find the correct .pc files. The configure command finds the Yocto Project site files as well as any other necessary files. Within this environment, you can run configure or compile commands as if they were being run by the OpenEmbedded build system itself. As noted earlier, the working directory also automatically changes to the Source Directory (S). To manually run a specific task using devshell, run the corresponding run.* script in the ${WORKDIR}/temp directory (e.g., run.do_configure.pid). If a task's script does not exist, which would be the case if the task was skipped by way of the sstate cache, you can create the task by first running it outside of the devshell: $ bitbake -c task Execution of a task's run.* script and BitBake's execution of a task are identical. In other words, running the script re-runs the task just as it would be run using the bitbake -c command. Any run.* file that does not have a .pid extension is a symbolic link (symlink) to the most recent version of that file. Remember, that the devshell is a mechanism that allows you to get into the BitBake task execution environment. And as such, all commands must be called just as BitBake would call them. That means you need to provide the appropriate options for cross-compilation and so forth as applicable. When you are finished using devshell, exit the shell or close the terminal window. It is worth remembering that when using devshell you need to use the full compiler name such as arm-poky-linux-gnueabi-gcc instead of just using gcc. The same applies to other applications such as binutils, libtool and so forth. BitBake sets up environment variables such as CC to assist applications, such as make to find the correct tools. It is also worth noting that devshell still works over X11 forwarding and similar situations.

Similar to working within a development shell as described in the previous section, you can also spawn and work within an interactive Python development shell. When debugging certain commands or even when just editing packages, devpyshell can be a useful tool. When you invoke devpyshell, all tasks up to and including do_patch are run for the specified target. Then a new terminal is opened. Additionally, key Python objects and code are available in the same way they are to BitBake tasks, in particular, the data store 'd'. So, commands such as the following are useful when exploring the data store and running functions: pydevshell> d.getVar("STAGING_DIR", True) '/media/build1/poky/build/tmp/sysroots' pydevshell> d.getVar("STAGING_DIR", False) '${TMPDIR}/sysroots' pydevshell> d.setVar("FOO", "bar") pydevshell> d.getVar("FOO", True) 'bar' pydevshell> d.delVar("FOO") pydevshell> d.getVar("FOO", True) pydevshell> bb.build.exec_func("do_unpack", d) pydevshell> The commands execute just as if the OpenEmbedded build system were executing them. Consequently, working this way can be helpful when debugging a build or preparing software to be used with the OpenEmbedded build system. Following is an example that uses devpyshell on a target named matchbox-desktop: $ bitbake matchbox-desktop -c devpyshell This command spawns a terminal and places you in an interactive Python interpreter within the OpenEmbedded build environment. The OE_TERMINAL variable controls what type of shell is opened. When you are finished using devpyshell, you can exit the shell either by using Ctrl+d or closing the terminal window.

Bitbake also has functionality that allows you to build multiple targets at the same time, where each target uses a different configuration. In order to accomplish this, you setup each of the configurations you need to use in parallel by placing the configuration files in your current build directory alongside the usual local.conf file. Follow these guidelines to create an environment that supports multiple configurations: Create Configuration Files: You need to create a single configuration file for each configuration for which you want to add support. These files would contain lines such as the following: MACHINE = "A" The files would contain any other variables that can be set and built in the same directory. You can change the TMPDIR to not conflict. Furthermore, the configuration file must be located in the current build directory in a directory named multiconfig under the build's conf directory where local.conf resides. The reason for this restriction is because the BBPATH variable is not constructed until the layers are parsed. Consequently, using the configuration file as a pre-configuration file is not possible unless it is located in the current working directory. Add the BitBake Multi-Config Variable to you Local Configuration File: Use the BBMULTICONFIG variable in your conf/local.conf configuration file to specify each separate configuration. For example, the following line tells BitBake it should load conf/multiconfig/configA.conf, conf/multiconfig/configB.conf, and conf/multiconfig/configC.conf. BBMULTICONFIG = "configA configB configC" Launch BitBake: Use the following BitBake command form to launch the build: $ bitbake [multiconfig:multiconfigname:]target [[[multiconfig:multiconfigname:]target] ... ] Following is an example that supports building a minimal image for configuration A alongside a standard core-image-sato, which takes its configuration from local.conf: $ bitbake multiconfig:configA:core-image-minimal core-image-sato Support for multiple configurations in this current release of the Yocto Project (&DISTRO_NAME; &DISTRO;) has some known issues: No inter-multi-configuration dependencies exist. Shared State (sstate) optimizations do not exist. Consequently, if the build uses the same object twice in, for example, two different TMPDIR directories, the build will either load from an existing sstate cache at the start or build the object twice.

Libraries are an integral part of your system. This section describes some common practices you might find helpful when working with libraries to build your system: How to include static library files How to use the Multilib feature to combine multiple versions of library files into a single image How to install multiple versions of the same library in parallel on the same system

If you are building a library and the library offers static linking, you can control which static library files (*.a files) get included in the built library. The PACKAGES and FILES_* variables in the meta/conf/bitbake.conf configuration file define how files installed by the do_install task are packaged. By default, the PACKAGES variable includes ${PN}-staticdev, which represents all static library files. Some previously released versions of the Yocto Project defined the static library files through ${PN}-dev. Following is part of the BitBake configuration file, where you can see how the static library files are defined: PACKAGE_BEFORE_PN ?= "" PACKAGES = "${PN}-dbg ${PN}-staticdev ${PN}-dev ${PN}-doc ${PN}-locale ${PACKAGE_BEFORE_PN} ${PN}" PACKAGES_DYNAMIC = "^${PN}-locale-.*" FILES = "" FILES_${PN} = "${bindir}/* ${sbindir}/* ${libexecdir}/* ${libdir}/lib*${SOLIBS} \ ${sysconfdir} ${sharedstatedir} ${localstatedir} \ ${base_bindir}/* ${base_sbindir}/* \ ${base_libdir}/*${SOLIBS} \ ${base_prefix}/lib/udev/rules.d ${prefix}/lib/udev/rules.d \ ${datadir}/${BPN} ${libdir}/${BPN}/* \ ${datadir}/pixmaps ${datadir}/applications \ ${datadir}/idl ${datadir}/omf ${datadir}/sounds \ ${libdir}/bonobo/servers" FILES_${PN}-bin = "${bindir}/* ${sbindir}/*" FILES_${PN}-doc = "${docdir} ${mandir} ${infodir} ${datadir}/gtk-doc \ ${datadir}/gnome/help" SECTION_${PN}-doc = "doc" FILES_SOLIBSDEV ?= "${base_libdir}/lib*${SOLIBSDEV} ${libdir}/lib*${SOLIBSDEV}" FILES_${PN}-dev = "${includedir} ${FILES_SOLIBSDEV} ${libdir}/*.la \ ${libdir}/*.o ${libdir}/pkgconfig ${datadir}/pkgconfig \ ${datadir}/aclocal ${base_libdir}/*.o \ ${libdir}/${BPN}/*.la ${base_libdir}/*.la" SECTION_${PN}-dev = "devel" ALLOW_EMPTY_${PN}-dev = "1" RDEPENDS_${PN}-dev = "${PN} (= ${EXTENDPKGV})" FILES_${PN}-staticdev = "${libdir}/*.a ${base_libdir}/*.a ${libdir}/${BPN}/*.a" SECTION_${PN}-staticdev = "devel" RDEPENDS_${PN}-staticdev = "${PN}-dev (= ${EXTENDPKGV})"

The build system offers the ability to build libraries with different target optimizations or architecture formats and combine these together into one system image. You can link different binaries in the image against the different libraries as needed for specific use cases. This feature is called "Multilib." An example would be where you have most of a system compiled in 32-bit mode using 32-bit libraries, but you have something large, like a database engine, that needs to be a 64-bit application and uses 64-bit libraries. Multilib allows you to get the best of both 32-bit and 64-bit libraries. While the Multilib feature is most commonly used for 32 and 64-bit differences, the approach the build system uses facilitates different target optimizations. You could compile some binaries to use one set of libraries and other binaries to use a different set of libraries. The libraries could differ in architecture, compiler options, or other optimizations. Several examples exist in the meta-skeleton layer found in the Source Directory: conf/multilib-example.conf configuration file conf/multilib-example2.conf configuration file recipes-multilib/images/core-image-multilib-example.bb recipe

User-specific requirements drive the Multilib feature. Consequently, there is no one "out-of-the-box" configuration that likely exists to meet your needs. In order to enable Multilib, you first need to ensure your recipe is extended to support multiple libraries. Many standard recipes are already extended and support multiple libraries. You can check in the meta/conf/multilib.conf configuration file in the Source Directory to see how this is done using the BBCLASSEXTEND variable. Eventually, all recipes will be covered and this list will not be needed. For the most part, the Multilib class extension works automatically to extend the package name from ${PN} to ${MLPREFIX}${PN}, where MLPREFIX is the particular multilib (e.g. "lib32-" or "lib64-"). Standard variables such as DEPENDS, RDEPENDS, RPROVIDES, RRECOMMENDS, PACKAGES, and PACKAGES_DYNAMIC are automatically extended by the system. If you are extending any manual code in the recipe, you can use the ${MLPREFIX} variable to ensure those names are extended correctly. This automatic extension code resides in multilib.bbclass.

After you have set up the recipes, you need to define the actual combination of multiple libraries you want to build. You accomplish this through your local.conf configuration file in the Build Directory. An example configuration would be as follows: MACHINE = "qemux86-64" require conf/multilib.conf MULTILIBS = "multilib:lib32" DEFAULTTUNE_virtclass-multilib-lib32 = "x86" IMAGE_INSTALL_append = " lib32-glib-2.0" This example enables an additional library named lib32 alongside the normal target packages. When combining these "lib32" alternatives, the example uses "x86" for tuning. For information on this particular tuning, see meta/conf/machine/include/ia32/arch-ia32.inc. The example then includes lib32-glib-2.0 in all the images, which illustrates one method of including a multiple library dependency. You can use a normal image build to include this dependency, for example: $ bitbake core-image-sato You can also build Multilib packages specifically with a command like this: $ bitbake lib32-glib-2.0

Generic implementation details as well as details that are specific to package management systems exist. Following are implementation details that exist regardless of the package management system: The typical convention used for the class extension code as used by Multilib assumes that all package names specified in PACKAGES that contain ${PN} have ${PN} at the start of the name. When that convention is not followed and ${PN} appears at the middle or the end of a name, problems occur. The TARGET_VENDOR value under Multilib will be extended to "-vendormlmultilib" (e.g. "-pokymllib32" for a "lib32" Multilib with Poky). The reason for this slightly unwieldy contraction is that any "-" characters in the vendor string presently break Autoconf's config.sub, and other separators are problematic for different reasons. For the RPM Package Management System, the following implementation details exist: A unique architecture is defined for the Multilib packages, along with creating a unique deploy folder under tmp/deploy/rpm in the Build Directory. For example, consider lib32 in a qemux86-64 image. The possible architectures in the system are "all", "qemux86_64", "lib32_qemux86_64", and "lib32_x86". The ${MLPREFIX} variable is stripped from ${PN} during RPM packaging. The naming for a normal RPM package and a Multilib RPM package in a qemux86-64 system resolves to something similar to bash-4.1-r2.x86_64.rpm and bash-4.1.r2.lib32_x86.rpm, respectively. When installing a Multilib image, the RPM backend first installs the base image and then installs the Multilib libraries. The build system relies on RPM to resolve the identical files in the two (or more) Multilib packages. For the IPK Package Management System, the following implementation details exist: The ${MLPREFIX} is not stripped from ${PN} during IPK packaging. The naming for a normal RPM package and a Multilib IPK package in a qemux86-64 system resolves to something like bash_4.1-r2.x86_64.ipk and lib32-bash_4.1-rw_x86.ipk, respectively. The IPK deploy folder is not modified with ${MLPREFIX} because packages with and without the Multilib feature can exist in the same folder due to the ${PN} differences. IPK defines a sanity check for Multilib installation using certain rules for file comparison, overridden, etc.

Situations can exist where you need to install and use multiple versions of the same library on the same system at the same time. These situations almost always exist when a library API changes and you have multiple pieces of software that depend on the separate versions of the library. To accommodate these situations, you can install multiple versions of the same library in parallel on the same system. The process is straightforward as long as the libraries use proper versioning. With properly versioned libraries, all you need to do to individually specify the libraries is create separate, appropriately named recipes where the PN part of the name includes a portion that differentiates each library version (e.g.the major part of the version number). Thus, instead of having a single recipe that loads one version of a library (e.g. clutter), you provide multiple recipes that result in different versions of the libraries you want. As an example, the following two recipes would allow the two separate versions of the clutter library to co-exist on the same system: clutter-1.6_1.6.20.bb clutter-1.8_1.8.4.bb Additionally, if you have other recipes that depend on a given library, you need to use the DEPENDS variable to create the dependency. Continuing with the same example, if you want to have a recipe depend on the 1.8 version of the clutter library, use the following in your recipe: DEPENDS = "clutter-1.8"

GObject introspection is the standard mechanism for accessing GObject-based software from runtime environments. GObject is a feature of the GLib library that provides an object framework for the GNOME desktop and related software. GObject Introspection adds information to GObject that allows objects created within it to be represented across different programming languages. If you want to construct GStreamer pipelines using Python, or control UPnP infrastructure using Javascript and GUPnP, GObject introspection is the only way to do it. This section describes the Yocto Project support for generating and packaging GObject introspection data. GObject introspection data is a description of the API provided by libraries built on top of GLib framework, and, in particular, that framework's GObject mechanism. GObject Introspection Repository (GIR) files go to -dev packages, typelib files go to main packages as they are packaged together with libraries that are introspected. The data is generated when building such a library, by linking the library with a small executable binary that asks the library to describe itself, and then executing the binary and processing its output. Generating this data in a cross-compilation environment is difficult because the library is produced for the target architecture, but its code needs to be executed on the build host. This problem is solved with the OpenEmbedded build system by running the code through QEMU, which allows precisely that. Unfortunately, QEMU does not always work perfectly as mentioned in the xxx section.

Enabling the generation of introspection data (GIR files) in your library package involves the following: Inherit the gobject-introspection class. Make sure introspection is not disabled anywhere in the recipe or from anything the recipe includes. Also, make sure that "gobject-introspection-data" is not in DISTRO_FEATURES_BACKFILL_CONSIDERED and that "qemu-usermode" is not in MACHINE_FEATURES_BACKFILL_CONSIDERED. If either of these conditions exist, nothing will happen. Try to build the recipe. If you encounter build errors that look like something is unable to find .so libraries, check where these libraries are located in the source tree and add the following to the recipe: GIR_EXTRA_LIBS_PATH = "${B}/something/.libs" See recipes in the oe-core repository that use that GIR_EXTRA_LIBS_PATH variable as an example. Look for any other errors, which probably mean that introspection support in a package is not entirely standard, and thus breaks down in a cross-compilation environment. For such cases, custom-made fixes are needed. A good place to ask and receive help in these cases is the Yocto Project mailing lists. Using a library that no longer builds against the latest Yocto Project release and prints introspection related errors is a good candidate for the previous procedure.

You might find that you do not want to generate introspection data. Or, perhaps QEMU does not work on your build host and target architecture combination. If so, you can use either of the following methods to disable GIR file generations: Add the following to your distro configuration: DISTRO_FEATURES_BACKFILL_CONSIDERED = "gobject-introspection-data" Adding this statement disables generating introspection data using QEMU but will still enable building introspection tools and libraries (i.e. building them does not require the use of QEMU). Add the following to your machine configuration: MACHINE_FEATURES_BACKFILL_CONSIDERED = "qemu-usermode" Adding this statement disables the use of QEMU when building packages for your machine. Currently, this feature is used only by introspection recipes and has the same effect as the previously described option. Future releases of the Yocto Project might have other features affected by this option. If you disable introspection data, you can still obtain it through other means such as copying the data from a suitable sysroot, or by generating it on the target hardware. The OpenEmbedded build system does not currently provide specific support for these techniques.

Use the following procedure to test if generating introspection data is working in an image: Make sure that "gobject-introspection-data" is not in DISTRO_FEATURES_BACKFILL_CONSIDERED and that "qemu-usermode" is not in MACHINE_FEATURES_BACKFILL_CONSIDERED. Build core-image-sato. Launch a Terminal and then start Python in the terminal. Enter the following in the terminal: >>> from gi.repository import GLib >>> GLib.get_host_name() For something a little more advanced, enter the following: http://python-gtk-3-tutorial.readthedocs.org/en/latest/introduction.html

The following know issues exist for GObject Introspection Support: qemu-ppc64 immediately crashes. Consequently, you cannot build introspection data on that architecture. x32 is not supported by QEMU. Consequently, introspection data is disabled. musl causes transient GLib binaries to crash on assertion failures. Consequently, generating introspection data is disabled. Because QEMU is not able to run the binaries correctly, introspection is disabled for some specific packages under specific architectures (e.g. gcr, libsecret, and webkit). QEMU usermode might not work properly when running 64-bit binaries under 32-bit host machines. In particular, "qemumips64" is known to not work under i686.

You might want to use an external toolchain as part of your development. If this is the case, the fundamental steps you need to accomplish are as follows: Understand where the installed toolchain resides. For cases where you need to build the external toolchain, you would need to take separate steps to build and install the toolchain. Make sure you add the layer that contains the toolchain to your bblayers.conf file through the BBLAYERS variable. Set the EXTERNAL_TOOLCHAIN variable in your local.conf file to the location in which you installed the toolchain. A good example of an external toolchain used with the Yocto Project is Mentor Graphics Sourcery G++ Toolchain. You can see information on how to use that particular layer in the README file at . You can find further information by reading about the TCMODE variable in the Yocto Project Reference Manual's variable glossary.

Creating an image for a particular hardware target using the OpenEmbedded build system does not necessarily mean you can boot that image as is on your device. Physical devices accept and boot images in various ways depending on the specifics of the device. Usually, information about the hardware can tell you what image format the device requires. Should your device require multiple partitions on an SD card, flash, or an HDD, you can use the OpenEmbedded Image Creator, Wic, to create the properly partitioned image. The wic command generates partitioned images from existing OpenEmbedded build artifacts. Image generation is driven by partitioning commands contained in an Openembedded kickstart file (.wks) specified either directly on the command line or as one of a selection of canned kickstart files as shown with the wic list images command in the "Using an Existing Kickstart File" section. When you apply the command to a given set of build artifacts, the result is an image or set of images that can be directly written onto media and used on a particular system. For a kickstart file reference, see the "OpenEmbedded Kickstart (.wks) Reference" Chapter in the Yocto Project Reference Manual. The wic command and the infrastructure it is based on is by definition incomplete. The purpose of the command is to allow the generation of customized images, and as such, was designed to be completely extensible through a plug-in interface. See the "Wic Plug-Ins Interface" section in the Yocto Project Reference Manual for information on these plug-ins. This section provides some background information on Wic, describes what you need to have in place to run the tool, provides instruction on how to use the Wic utility, and provides several examples.

This section provides some background on the Wic utility. While none of this information is required to use Wic, you might find it interesting. The name "Wic" is derived from OpenEmbedded Image Creator (oeic). The "oe" diphthong in "oeic" was promoted to the letter "w", because "oeic" is both difficult to remember and to pronounce. Wic is loosely based on the Meego Image Creator (mic) framework. The Wic implementation has been heavily modified to make direct use of OpenEmbedded build artifacts instead of package installation and configuration, which are already incorporated within the OpenEmbedded artifacts. Wic is a completely independent standalone utility that initially provides easier-to-use and more flexible replacements for an existing functionality in OE Core's image-live class and mkefidisk.sh script. The difference between Wic and those examples is that with Wic the functionality of those scripts is implemented by a general-purpose partitioning language, which is based on Redhat kickstart syntax.

In order to use the Wic utility with the OpenEmbedded Build system, your system needs to meet the following requirements: The Linux distribution on your development host must support the Yocto Project. See the "Supported Linux Distributions" section in the Yocto Project Reference Manual for the list of distributions that support the Yocto Project. The standard system utilities, such as cp, must be installed on your development host system. You must have sourced the build environment setup script (i.e. &OE_INIT_FILE;) found in the Build Directory. You need to have the build artifacts already available, which typically means that you must have already created an image using the Openembedded build system (e.g. core-image-minimal). While it might seem redundant to generate an image in order to create an image using Wic, the current version of Wic requires the artifacts in the form generated by the OpenEmbedded build system. You must build several native tools, which are built to run on the build system: $ bitbake parted-native dosfstools-native mtools-native Include "wic" as part of the IMAGE_FSTYPES variable. Include the name of the wic kickstart file as part of the WKS_FILE variable

You can get general help for the wic command by entering the wic command by itself or by entering the command with a help argument as follows: $ wic -h $ wic --help Currently, Wic supports seven commands: cp, create, help, list, ls, rm, and write. You can get help for these commands as follows with command being one of the supported commands: $ wic help command You can also get detailed help on a number of topics from the help system. The output of wic --help displays a list of available help topics under a "Help topics" heading. You can have the help system display the help text for a given topic by prefacing the topic with wic help: $ wic help help_topic You can find out more about the images Wic creates using the existing kickstart files with the following form of the command: $ wic list image help For image, you can provide any of the following: beaglebone mpc8315e-rdb genericx86 edgerouter qemux86-directdisk directdisk-gpt mkefidisk directdisk systemd-bootdisk mkhybridiso sdimage-bootpart directdisk-multi-rootfs directdisk-bootloader-config

You can use Wic in two different modes, depending on how much control you need for specifying the Openembedded build artifacts that are used for creating the image: Raw and Cooked: Raw Mode: You explicitly specify build artifacts through wic command-line arguments. Cooked Mode: The current MACHINE setting and image name are used to automatically locate and provide the build artifacts. You just supply a kickstart file and the name of the image from which to use artifacts. Regardless of the mode you use, you need to have the build artifacts ready and available.

Running Wic in raw mode allows you to specify all the partitions through the wic command line. The primary use for raw mode is if you have built your kernel outside of the Yocto Project Build Directory. In other words, you can point to arbitrary kernel, root filesystem locations, and so forth. Contrast this behavior with cooked mode where Wic looks in the Build Directory (e.g. tmp/deploy/images/machine). The general form of the wic command in raw mode is: $ wic create wks_file options ... Where: wks_file: An OpenEmbedded kickstart file. You can provide your own custom file or use a file from a set of existing files as described by further options. optional arguments: -h, --help show this help message and exit -o OUTDIR, --outdir OUTDIR name of directory to create image in -e IMAGE_NAME, --image-name IMAGE_NAME name of the image to use the artifacts from e.g. core- image-sato -r ROOTFS_DIR, --rootfs-dir ROOTFS_DIR path to the /rootfs dir to use as the .wks rootfs source -b BOOTIMG_DIR, --bootimg-dir BOOTIMG_DIR path to the dir containing the boot artifacts (e.g. /EFI or /syslinux dirs) to use as the .wks bootimg source -k KERNEL_DIR, --kernel-dir KERNEL_DIR path to the dir containing the kernel to use in the .wks bootimg -n NATIVE_SYSROOT, --native-sysroot NATIVE_SYSROOT path to the native sysroot containing the tools to use to build the image -s, --skip-build-check skip the build check -f, --build-rootfs build rootfs -c {gzip,bzip2,xz}, --compress-with {gzip,bzip2,xz} compress image with specified compressor -m, --bmap generate .bmap --no-fstab-update Do not change fstab file. -v VARS_DIR, --vars VARS_DIR directory with <image>.env files that store bitbake variables -D, --debug output debug information You do not need root privileges to run Wic. In fact, you should not run as root when using the utility.

Running Wic in cooked mode leverages off artifacts in Build Directory. In other words, you do not have to specify kernel or root filesystem locations as part of the command. All you need to provide is a kickstart file and the name of the image from which to use artifacts by using the "-e" option. Wic looks in the Build Directory (e.g. tmp/deploy/images/machine) for artifacts. The general form of the wic command using Cooked Mode is as follows: $ wic create wks_file -e IMAGE_NAME Where: wks_file: An OpenEmbedded kickstart file. You can provide your own custom file or use a file from a set of existing files provided with the Yocto Project release. required argument: -e IMAGE_NAME, --image-name IMAGE_NAME name of the image to use the artifacts from e.g. core- image-sato

If you do not want to create your own kickstart file, you can use an existing file provided by the Wic installation. As shipped, kickstart files can be found in the Yocto Project Source Repositories in the following two locations: poky/meta-yocto-bsp/wic poky/scripts/lib/wic/canned-wks Use the following command to list the available kickstart files: $ wic list images beaglebone Create SD card image for Beaglebone mpc8315e-rdb Create SD card image for MPC8315E-RDB genericx86 Create an EFI disk image for genericx86* edgerouter Create SD card image for Edgerouter qemux86-directdisk Create a qemu machine 'pcbios' direct disk image directdisk-gpt Create a 'pcbios' direct disk image mkefidisk Create an EFI disk image directdisk Create a 'pcbios' direct disk image systemd-bootdisk Create an EFI disk image with systemd-boot mkhybridiso Create a hybrid ISO image sdimage-bootpart Create SD card image with a boot partition directdisk-multi-rootfs Create multi rootfs image using rootfs plugin directdisk-bootloader-config Create a 'pcbios' direct disk image with custom bootloader config When you use an existing file, you do not have to use the .wks extension. Here is an example in Raw Mode that uses the directdisk file: $ wic create directdisk -r rootfs_dir -b bootimg_dir \ -k kernel_dir -n native_sysroot Here are the actual partition language commands used in the genericx86.wks file to generate an image: # short-description: Create an EFI disk image for genericx86* # long-description: Creates a partitioned EFI disk image for genericx86* machines part /boot --source bootimg-efi --sourceparams="loader=grub-efi" --ondisk sda --label msdos --active --align 1024 part / --source rootfs --ondisk sda --fstype=ext4 --label platform --align 1024 --use-uuid part swap --ondisk sda --size 44 --label swap1 --fstype=swap bootloader --ptable gpt --timeout=5 --append="rootfstype=ext4 console=ttyS0,115200 console=tty0"

This section provides several examples that show how to use the Wic utility. All the examples assume the list of requirements in the "Requirements" section have been met. The examples assume the previously generated image is core-image-minimal.

This example runs in Cooked Mode and uses the mkefidisk kickstart file: $ wic create mkefidisk -e core-image-minimal INFO: Building wic-tools... . . . INFO: The new image(s) can be found here: ./mkefidisk-201710061409-sda.direct The following build artifacts were used to create the image(s): ROOTFS_DIR: /home/scottrif/poky/build/tmp.wic.r4hkds0b/rootfs_copy BOOTIMG_DIR: /home/scottrif/poky/build/tmp/work/qemux86-poky-linux/core-image-minimal/1.0-r0/recipe-sysroot/usr/share KERNEL_DIR: /home/scottrif/poky/build/tmp/deploy/images/qemux86 NATIVE_SYSROOT: /home/scottrif/poky/build/tmp/work/i586-poky-linux/wic-tools/1.0-r0/recipe-sysroot-native INFO: The image(s) were created using OE kickstart file: /home/scottrif/poky/scripts/lib/wic/canned-wks/mkefidisk.wks The previous example shows the easiest way to create an image by running in cooked mode and supplying a kickstart file and the "-e" option to point to the existing build artifacts. Your local.conf file needs to have the MACHINE variable set to the machine you are using, which is "qemux86" in this example. Once the image builds, the output provides image location, artifact use, and kickstart file information. You should always verify the details provided in the output to make sure that the image was indeed created exactly as expected. Continuing with the example, you can now write the image to a USB stick, or whatever media for which you built your image, and boot from the media. You can write the image by using bmaptool or dd: $ oe-run-native bmaptool copy build/mkefidisk-201710061409-sda.direct /dev/sdX or $ sudo dd if=build/mkefidisk-201710061409-sda.direct of=/dev/sdX For more information on how to use the bmaptool to flash a device with an image, see the "Flashing Images Using bmaptool" section.

Because partitioned image creation is driven by the kickstart file, it is easy to affect image creation by changing the parameters in the file. This next example demonstrates that through modification of the directdisk-gpt kickstart file. As mentioned earlier, you can use the command wic list images to show the list of existing kickstart files. The directory in which the directdisk-gpt.wks file resides is scripts/lib/image/canned-wks/, which is located in the Source Directory (e.g. poky). Because available files reside in this directory, you can create and add your own custom files to the directory. Subsequent use of the wic list images command would then include your kickstart files. In this example, the existing directdisk-gpt file already does most of what is needed. However, for the hardware in this example, the image will need to boot from sdb instead of sda, which is what the directdisk-gpt kickstart file uses. The example begins by making a copy of the directdisk-gpt.wks file in the scripts/lib/image/canned-wks directory and then by changing the lines that specify the target disk from which to boot. $ cp /home/scottrif/poky/scripts/lib/wic/canned-wks/directdisk-gpt.wks \ /home/scottrif/poky/scripts/lib/wic/canned-wks/directdisksdb-gpt.wks Next, the example modifies the directdisksdb-gpt.wks file and changes all instances of "--ondisk sda" to "--ondisk sdb". The example changes the following two lines and leaves the remaining lines untouched: part /boot --source bootimg-pcbios --ondisk sdb --label boot --active --align 1024 part / --source rootfs --ondisk sdb --fstype=ext4 --label platform --align 1024 --use-uuid Once the lines are changed, the example generates the directdisksdb-gpt image. The command points the process at the core-image-minimal artifacts for the Next Unit of Computing (nuc) MACHINE the local.conf. $ wic create directdisksdb-gpt -e core-image-minimal INFO: Building wic-tools... . . . Initialising tasks: 100% |#######################################| Time: 0:00:01 NOTE: Executing SetScene Tasks NOTE: Executing RunQueue Tasks NOTE: Tasks Summary: Attempted 1161 tasks of which 1157 didn't need to be rerun and all succeeded. INFO: Creating image(s)... INFO: The new image(s) can be found here: ./directdisksdb-gpt-201710090938-sdb.direct The following build artifacts were used to create the image(s): ROOTFS_DIR: /home/scottrif/poky/build/tmp.wic.hk3wl6zn/rootfs_copy BOOTIMG_DIR: /home/scottrif/poky/build/tmp/work/qemux86-poky-linux/core-image-minimal/1.0-r0/recipe-sysroot/usr/share KERNEL_DIR: /home/scottrif/poky/build/tmp/deploy/images/qemux86 NATIVE_SYSROOT: /home/scottrif/poky/build/tmp/work/i586-poky-linux/wic-tools/1.0-r0/recipe-sysroot-native INFO: The image(s) were created using OE kickstart file: /home/scottrif/poky/scripts/lib/wic/canned-wks/directdisksdb-gpt.wks Continuing with the example, you can now directly dd the image to a USB stick, or whatever media for which you built your image, and boot the resulting media: $ sudo dd if=directdisksdb-gpt-201710090938-sdb.direct of=/dev/sdb 140966+0 records in 140966+0 records out 72174592 bytes (72 MB, 69 MiB) copied, 78.0282 s, 925 kB/s $ sudo eject /dev/sdb

This next example manually specifies each build artifact (runs in Raw Mode) and uses a modified kickstart file. The example also uses the -o option to cause Wic to create the output somewhere other than the default output directory, which is the current directory: $ wic create /home/scottrif/my_yocto/test.wks -o /home/scottrif/testwic \ --rootfs-dir /home/scottrif/poky/build/tmp/work/qemux86-poky-linux/core-image-minimal/1.0-r0/rootfs \ --bootimg-dir /home/scottrif/poky/build/tmp/work/qemux86-poky-linux/core-image-minimal/1.0-r0/recipe-sysroot/usr/share \ --kernel-dir /home/scottrif/poky/build/tmp/deploy/images/qemux86 \ --native-sysroot /home/scottrif/poky/build/tmp/work/i586-poky-linux/wic-tools/1.0-r0/recipe-sysroot-native INFO: Creating image(s)... INFO: The new image(s) can be found here: /home/scottrif/testwic/test-201710091445-sdb.direct The following build artifacts were used to create the image(s): ROOTFS_DIR: /home/scottrif/testwic/tmp.wic.x4wipbmb/rootfs_copy BOOTIMG_DIR: /home/scottrif/poky/build/tmp/work/qemux86-poky-linux/core-image-minimal/1.0-r0/recipe-sysroot/usr/share KERNEL_DIR: /home/scottrif/poky/build/tmp/deploy/images/qemux86 NATIVE_SYSROOT: /home/scottrif/poky/build/tmp/work/i586-poky-linux/wic-tools/1.0-r0/recipe-sysroot-native INFO: The image(s) were created using OE kickstart file: /home/scottrif/my_yocto/test.wks For this example, MACHINE did not have to be specified in the local.conf file since the artifact is manually specified.

Wic image manipulation allows you to shorten turnaround time during image development. For example, you can use Wic to delete the kernel partition of a Wic image and then insert a newly built kernel. This saves you time from having to rebuild the entire image each time you modify the kernel. In order to use Wic to manipulate a Wic image as in this example, your development machine must have the mtools package installed. The following example examines the contents of the Wic image, deletes the existing kernel, and then inserts a new kernel: List the Partitions: Use the wic ls command to list all the partitions in the Wic image: $ wic ls tmp/deploy/images/qemux86/core-image-minimal-qemux86.wic Num Start End Size Fstype 1 1048576 25041919 23993344 fat16 2 25165824 72157183 46991360 ext4 The previous output shows two partitions in the core-image-minimal-qemux86.wic image. Examine a Particular Partition: Use the wic ls command again but in a different form to examine a particular partition. You can get command usage on any Wic command using the following form: $ wic help command For example, the following command shows you the various ways to use the wic ls command: $ wic help ls The following command shows what is in Partition one: $ wic ls tmp/deploy/images/qemux86/core-image-minimal-qemux86.wic:1 Volume in drive : is boot Volume Serial Number is E894-1809 Directory for ::/ libcom32 c32 186500 2017-10-09 16:06 libutil c32 24148 2017-10-09 16:06 syslinux cfg 220 2017-10-09 16:06 vesamenu c32 27104 2017-10-09 16:06 vmlinuz 6904608 2017-10-09 16:06 5 files 7 142 580 bytes 16 582 656 bytes free The previous output shows five files, with the vmlinuz being the kernel. If you see the following error, you need to update or create a ~/.mtoolsrc file and be sure to have the line “mtools_skip_check=1“ in the file. Then, run the Wic command again: ERROR: _exec_cmd: /usr/bin/mdir -i /tmp/wic-parttfokuwra ::/ returned '1' instead of 0 output: Total number of sectors (47824) not a multiple of sectors per track (32)! Add mtools_skip_check=1 to your .mtoolsrc file to skip this test Remove the Old Kernel: Use the wic rm command to remove the vmlinuz file (kernel): $ wic rm tmp/deploy/images/qemux86/core-image-minimal-qemux86.wic:1/vmlinuz Add In the New Kernel: Use the wic cp command to add the updated kernel to the Wic image. Depending on how you built your kernel, it could be in different places. If you used devtool and an SDK to build your kernel, it resides in the tmp/work directory of the extensible SDK. If you used make to build the kernel, the kernel will be in the workspace/sources area. The following example assumes devtool was used to build the kernel: cp ~/poky_sdk/tmp/work/qemux86-poky-linux/linux-yocto/4.12.12+git999-r0/linux-yocto-4.12.12+git999/arch/x86/boot/bzImage \ ~/poky/build/tmp/deploy/images/qemux86/core-image-minimal-qemux86.wic:1/vmlinuz Once the new kernel is added back into the image, you can use the dd command or bmaptool to flash your wic image onto an SD card or USB stick and test your target. Using bmaptool is generally 10 to 20 times faster than using dd.

An initial RAM filesystem (initramfs) image provides a temporary root filesystem used for early system initialization (e.g. loading of modules needed to locate and mount the "real" root filesystem). The initramfs image is the successor of initial RAM disk (initrd). It is a "copy in and out" (cpio) archive of the initial filesystem that gets loaded into memory during the Linux startup process. Because Linux uses the contents of the archive during initialization, the initramfs image needs to contain all of the device drivers and tools needed to mount the final root filesystem. Follow these steps to create an initramfs image: Create the initramfs Image Recipe: You can reference the core-image-minimal-initramfs.bb recipe found in the meta/recipes-core directory of the Source Directory as an example from which to work. Decide if You Need to Bundle the initramfs Image Into the Kernel Image: If you want the initramfs image that is built to be bundled in with the kernel image, set the INITRAMFS_IMAGE_BUNDLE variable to "1" in your local.conf configuration file and set the INITRAMFS_IMAGE variable in the recipe that builds the kernel image. It is recommended that you do bundle the initramfs image with the kernel image to avoid circular dependencies between the kernel recipe and the initramfs recipe should the initramfs image include kernel modules. Setting the INITRAMFS_IMAGE_BUNDLE flag causes the initramfs image to be unpacked into the ${B}/usr/ directory. The unpacked initramfs image is then passed to the kernel's Makefile using the CONFIG_INITRAMFS_SOURCE variable, allowing the initramfs image to be built into the kernel normally. If you choose to not bundle the initramfs image with the kernel image, you are essentially using an Initial RAM Disk (initrd). Creating an initrd is handled primarily through the INITRD_IMAGE, INITRD_LIVE, and INITRD_IMAGE_LIVE variables. For more information, see the image-live.bbclass file. Optionally Add Items to the initramfs Image Through the initramfs Image Recipe: If you add items to the initramfs image by way of its recipe, you should use PACKAGE_INSTALL rather than IMAGE_INSTALL. PACKAGE_INSTALL gives more direct control of what is added to the image as compared to the defaults you might not necessarily want that are set by the image or core-image classes. Build the Kernel Image and the initramfs Image: Build your kernel image using BitBake. Because the initramfs image recipe is a dependency of the kernel image, the initramfs image is built as well and bundled with the kernel image if you used the INITRAMFS_IMAGE_BUNDLE variable described earlier.

An easy way to flash an image to a bootable device is to use bmaptool, which is integrated into the OpenEmbedded build system. Following, is an example that shows how to flash a Wic image. You can use bmaptool to flash any type of image. Use these steps to flash an image using bmaptool: Unless you are able to install the bmap-tools package as mentioned in the note in the second bullet of step 3 further down, you will need to build bmaptool before using it. Build the tool using the following command: $ bitbake bmap-tools-native Update the local.conf File: Add the following to your local.conf file: IMAGE_FSTYPES += "wic wic.bmap" Get Your Image: Either have your image ready (pre-built) or take the step build the image: $ bitbake image Flash the Device: Flash the device with the image by using bmaptool depending on your particular setup: If you have write access to the media, use this command form: $ oe-run-native bmap-tools-native bmaptool copy ./tmp/deploy/images/qemux86-64-core-image-minimal-machine.wic /dev/sdX If you do not have write access to the media, use the following commands: $ sudo chmod 666 /dev/sdX $ oe-run-native bmap-tools-native bmaptool copy ./tmp/deploy/images/qemux86-64-core-image-minimal-machine.wic /dev/sdX If you are using Ubuntu or Debian distributions, you can install the bmap-tools package using the following command and then use the tool without specifying PATH even from the root account: $ sudo apt-get install bmap-tools For help on the bmaptool command, use the following command: $ bmaptool --help

Security is of increasing concern for embedded devices. Consider the issues and problems discussed in just this sampling of work found across the Internet: "Security Risks of Embedded Systems" by Bruce Schneier "Internet Census 2012" by Carna Botnet "Security Issues for Embedded Devices" by Jake Edge When securing your image is of concern, there are steps, tools, and variables that you can consider to help you reach the security goals you need for your particular device. Not all situations are identical when it comes to making an image secure. Consequently, this section provides some guidance and suggestions for consideration when you want to make your image more secure. Because the security requirements and risks are different for every type of device, this section cannot provide a complete reference on securing your custom OS. It is strongly recommended that you also consult other sources of information on embedded Linux system hardening and on security.

General considerations exist that help you create more secure images. You should consider the following suggestions to help make your device more secure: Scan additional code you are adding to the system (e.g. application code) by using static analysis tools. Look for buffer overflows and other potential security problems. Pay particular attention to the security for any web-based administration interface. Web interfaces typically need to perform administrative functions and tend to need to run with elevated privileges. Thus, the consequences resulting from the interface's security becoming compromised can be serious. Look for common web vulnerabilities such as cross-site-scripting (XSS), unvalidated inputs, and so forth. As with system passwords, the default credentials for accessing a web-based interface should not be the same across all devices. This is particularly true if the interface is enabled by default as it can be assumed that many end-users will not change the credentials. Ensure you can update the software on the device to mitigate vulnerabilities discovered in the future. This consideration especially applies when your device is network-enabled. Ensure you remove or disable debugging functionality before producing the final image. For information on how to do this, see the "Considerations Specific to the OpenEmbedded Build System" section. Ensure you have no network services listening that are not needed. Remove any software from the image that is not needed. Enable hardware support for secure boot functionality when your device supports this functionality.

The Yocto Project has security flags that you can enable that help make your build output more secure. The security flags are in the meta/conf/distro/include/security_flags.inc file in your Source Directory (e.g. poky). Depending on the recipe, certain security flags are enabled and disabled by default. Use the following line in your local.conf file or in your custom distribution configuration file to enable the security compiler and linker flags for your build: require conf/distro/include/security_flags.inc

You can take some steps that are specific to the OpenEmbedded build system to make your images more secure: Ensure "debug-tweaks" is not one of your selected IMAGE_FEATURES. When creating a new project, the default is to provide you with an initial local.conf file that enables this feature using the EXTRA_IMAGE_FEATURES variable with the line: EXTRA_IMAGE_FEATURES = "debug-tweaks" To disable that feature, simply comment out that line in your local.conf file, or make sure IMAGE_FEATURES does not contain "debug-tweaks" before producing your final image. Among other things, leaving this in place sets the root password as blank, which makes logging in for debugging or inspection easy during development but also means anyone can easily log in during production. It is possible to set a root password for the image and also to set passwords for any extra users you might add (e.g. administrative or service type users). When you set up passwords for multiple images or users, you should not duplicate passwords. To set up passwords, use the extrausers class, which is the preferred method. For an example on how to set up both root and user passwords, see the "extrausers.bbclass" section. When adding extra user accounts or setting a root password, be cautious about setting the same password on every device. If you do this, and the password you have set is exposed, then every device is now potentially compromised. If you need this access but want to ensure security, consider setting a different, random password for each device. Typically, you do this as a separate step after you deploy the image onto the device. Consider enabling a Mandatory Access Control (MAC) framework such as SMACK or SELinux and tuning it appropriately for your device's usage. You can find more information in the meta-selinux layer.

The Yocto Project provides tools for making your image more secure. You can find these tools in the meta-security layer of the Yocto Project Source Repositories.

When you build an image using the Yocto Project and do not alter any distribution Metadata, you are creating a Poky distribution. If you wish to gain more control over package alternative selections, compile-time options, and other low-level configurations, you can create your own distribution. To create your own distribution, the basic steps consist of creating your own distribution layer, creating your own distribution configuration file, and then adding any needed code and Metadata to the layer. The following steps provide some more detail: Create a layer for your new distro: Create your distribution layer so that you can keep your Metadata and code for the distribution separate. It is strongly recommended that you create and use your own layer for configuration and code. Using your own layer as compared to just placing configurations in a local.conf configuration file makes it easier to reproduce the same build configuration when using multiple build machines. See the "Creating a General Layer Using the bitbake-layers Script" section for information on how to quickly set up a layer. Create the distribution configuration file: The distribution configuration file needs to be created in the conf/distro directory of your layer. You need to name it using your distribution name (e.g. mydistro.conf). The DISTRO variable in your local.conf file determines the name of your distribution. You can split out parts of your configuration file into include files and then "require" them from within your distribution configuration file. Be sure to place the include files in the conf/distro/include directory of your layer. A common example usage of include files would be to separate out the selection of desired version and revisions for individual recipes. Your configuration file needs to set the following required variables: DISTRO_NAME DISTRO_VERSION These following variables are optional and you typically set them from the distribution configuration file: DISTRO_FEATURES DISTRO_EXTRA_RDEPENDS DISTRO_EXTRA_RRECOMMENDS TCLIBC If you want to base your distribution configuration file on the very basic configuration from OE-Core, you can use conf/distro/defaultsetup.conf as a reference and just include variables that differ as compared to defaultsetup.conf. Alternatively, you can create a distribution configuration file from scratch using the defaultsetup.conf file or configuration files from other distributions such as Poky or Angstrom as references. Provide miscellaneous variables: Be sure to define any other variables for which you want to create a default or enforce as part of the distribution configuration. You can include nearly any variable from the local.conf file. The variables you use are not limited to the list in the previous bulleted item. Point to Your distribution configuration file: In your local.conf file in the Build Directory, set your DISTRO variable to point to your distribution's configuration file. For example, if your distribution's configuration file is named mydistro.conf, then you point to it as follows: DISTRO = "mydistro" Add more to the layer if necessary: Use your layer to hold other information needed for the distribution: Add recipes for installing distro-specific configuration files that are not already installed by another recipe. If you have distro-specific configuration files that are included by an existing recipe, you should add an append file (.bbappend) for those. For general information and recommendations on how to add recipes to your layer, see the "Creating Your Own Layer" and "Following Best Practices When Creating Layers" sections. Add any image recipes that are specific to your distribution. Add a psplash append file for a branded splash screen. For information on append files, see the "Using .bbappend Files in Your Layer" section. Add any other append files to make custom changes that are specific to individual recipes.

If you are producing your own customized version of the build system for use by other users, you might want to customize the message shown by the setup script or you might want to change the template configuration files (i.e. local.conf and bblayers.conf) that are created in a new build directory. The OpenEmbedded build system uses the environment variable TEMPLATECONF to locate the directory from which it gathers configuration information that ultimately ends up in the Build Directory conf directory. By default, TEMPLATECONF is set as follows in the poky repository: TEMPLATECONF=${TEMPLATECONF:-meta-poky/conf} This is the directory used by the build system to find templates from which to build some key configuration files. If you look at this directory, you will see the bblayers.conf.sample, local.conf.sample, and conf-notes.txt files. The build system uses these files to form the respective bblayers.conf file, local.conf file, and display the list of BitBake targets when running the setup script. To override these default configuration files with configurations you want used within every new Build Directory, simply set the TEMPLATECONF variable to your directory. The TEMPLATECONF variable is set in the .templateconf file, which is in the top-level Source Directory folder (e.g. poky). Edit the .templateconf so that it can locate your directory. Best practices dictate that you should keep your template configuration directory in your custom distribution layer. For example, suppose you have a layer named meta-mylayer located in your home directory and you want your template configuration directory named myconf. Changing the .templateconf as follows causes the OpenEmbedded build system to look in your directory and base its configuration files on the *.sample configuration files it finds. The final configuration files (i.e. local.conf and bblayers.conf ultimately still end up in your Build Directory, but they are based on your *.sample files. TEMPLATECONF=${TEMPLATECONF:-meta-mylayer/myconf} Aside from the *.sample configuration files, the conf-notes.txt also resides in the default meta-poky/conf directory. The script that sets up the build environment (i.e. &OE_INIT_FILE;) uses this file to display BitBake targets as part of the script output. Customizing this conf-notes.txt file is a good way to make sure your list of custom targets appears as part of the script's output. Here is the default list of targets displayed as a result of running either of the setup scripts: You can now run 'bitbake <target>' Common targets are: core-image-minimal core-image-sato meta-toolchain meta-ide-support Changing the listed common targets is as easy as editing your version of conf-notes.txt in your custom template configuration directory and making sure you have TEMPLATECONF set to your directory.

Very small distributions have some significant advantages such as requiring less on-die or in-package memory (cheaper), better performance through efficient cache usage, lower power requirements due to less memory, faster boot times, and reduced development overhead. Some real-world examples where a very small distribution gives you distinct advantages are digital cameras, medical devices, and small headless systems. This section presents information that shows you how you can trim your distribution to even smaller sizes than the poky-tiny distribution, which is around 5 Mbytes, that can be built out-of-the-box using the Yocto Project.

The following list presents the overall steps you need to consider and perform to create distributions with smaller root filesystems, achieve faster boot times, maintain your critical functionality, and avoid initial RAM disks: Determine your goals and guiding principles. Understand what contributes to your image size. Reduce the size of the root filesystem. Reduce the size of the kernel. Eliminate packaging requirements. Look for other ways to minimize size. Iterate on the process.

Before you can reach your destination, you need to know where you are going. Here is an example list that you can use as a guide when creating very small distributions: Determine how much space you need (e.g. a kernel that is 1 Mbyte or less and a root filesystem that is 3 Mbytes or less). Find the areas that are currently taking 90% of the space and concentrate on reducing those areas. Do not create any difficult "hacks" to achieve your goals. Leverage the device-specific options. Work in a separate layer so that you keep changes isolated. For information on how to create layers, see the "Understanding and Creating Layers" section.

It is easiest to have something to start with when creating your own distribution. You can use the Yocto Project out-of-the-box to create the poky-tiny distribution. Ultimately, you will want to make changes in your own distribution that are likely modeled after poky-tiny. To use poky-tiny in your build, set the DISTRO variable in your local.conf file to "poky-tiny" as described in the "Creating Your Own Distribution" section. Understanding some memory concepts will help you reduce the system size. Memory consists of static, dynamic, and temporary memory. Static memory is the TEXT (code), DATA (initialized data in the code), and BSS (uninitialized data) sections. Dynamic memory represents memory that is allocated at runtime: stacks, hash tables, and so forth. Temporary memory is recovered after the boot process. This memory consists of memory used for decompressing the kernel and for the __init__ functions. To help you see where you currently are with kernel and root filesystem sizes, you can use two tools found in the Source Directory in the scripts/tiny/ directory: ksize.py: Reports component sizes for the kernel build objects. dirsize.py: Reports component sizes for the root filesystem. This next tool and command help you organize configuration fragments and view file dependencies in a human-readable form: merge_config.sh: Helps you manage configuration files and fragments within the kernel. With this tool, you can merge individual configuration fragments together. The tool allows you to make overrides and warns you of any missing configuration options. The tool is ideal for allowing you to iterate on configurations, create minimal configurations, and create configuration files for different machines without having to duplicate your process. The merge_config.sh script is part of the Linux Yocto kernel Git repositories (i.e. linux-yocto-3.14, linux-yocto-3.10, linux-yocto-3.8, and so forth) in the scripts/kconfig directory. For more information on configuration fragments, see the "Creating Configuration Fragments" section in the Yocto Project Linux Kernel Development Manual. bitbake -u taskexp -g bitbake_target: Using the BitBake command with these options brings up a Dependency Explorer from which you can view file dependencies. Understanding these dependencies allows you to make informed decisions when cutting out various pieces of the kernel and root filesystem.

The root filesystem is made up of packages for booting, libraries, and applications. To change things, you can configure how the packaging happens, which changes the way you build them. You can also modify the filesystem itself or select a different filesystem. First, find out what is hogging your root filesystem by running the dirsize.py script from your root directory: $ cd root-directory-of-image $ dirsize.py 100000 > dirsize-100k.log $ cat dirsize-100k.log You can apply a filter to the script to ignore files under a certain size. The previous example filters out any files below 100 Kbytes. The sizes reported by the tool are uncompressed, and thus will be smaller by a relatively constant factor in a compressed root filesystem. When you examine your log file, you can focus on areas of the root filesystem that take up large amounts of memory. You need to be sure that what you eliminate does not cripple the functionality you need. One way to see how packages relate to each other is by using the Dependency Explorer UI with the BitBake command: $ cd image-directory $ bitbake -u taskexp -g image Use the interface to select potential packages you wish to eliminate and see their dependency relationships. When deciding how to reduce the size, get rid of packages that result in minimal impact on the feature set. For example, you might not need a VGA display. Or, you might be able to get by with devtmpfs and mdev instead of udev. Use your local.conf file to make changes. For example, to eliminate udev and glib, set the following in the local configuration file: VIRTUAL-RUNTIME_dev_manager = "" Finally, you should consider exactly the type of root filesystem you need to meet your needs while also reducing its size. For example, consider cramfs, squashfs, ubifs, ext2, or an initramfs using initramfs. Be aware that ext3 requires a 1 Mbyte journal. If you are okay with running read-only, you do not need this journal. After each round of elimination, you need to rebuild your system and then use the tools to see the effects of your reductions.

The kernel is built by including policies for hardware-independent aspects. What subsystems do you enable? For what architecture are you building? Which drivers do you build by default? You can modify the kernel source if you want to help with boot time. Run the ksize.py script from the top-level Linux build directory to get an idea of what is making up the kernel: $ cd top-level-linux-build-directory $ ksize.py > ksize.log $ cat ksize.log When you examine the log, you will see how much space is taken up with the built-in .o files for drivers, networking, core kernel files, filesystem, sound, and so forth. The sizes reported by the tool are uncompressed, and thus will be smaller by a relatively constant factor in a compressed kernel image. Look to reduce the areas that are large and taking up around the "90% rule." To examine, or drill down, into any particular area, use the -d option with the script: $ ksize.py -d > ksize.log Using this option breaks out the individual file information for each area of the kernel (e.g. drivers, networking, and so forth). Use your log file to see what you can eliminate from the kernel based on features you can let go. For example, if you are not going to need sound, you do not need any drivers that support sound. After figuring out what to eliminate, you need to reconfigure the kernel to reflect those changes during the next build. You could run menuconfig and make all your changes at once. However, that makes it difficult to see the effects of your individual eliminations and also makes it difficult to replicate the changes for perhaps another target device. A better method is to start with no configurations using allnoconfig, create configuration fragments for individual changes, and then manage the fragments into a single configuration file using merge_config.sh. The tool makes it easy for you to iterate using the configuration change and build cycle. Each time you make configuration changes, you need to rebuild the kernel and check to see what impact your changes had on the overall size.

Packaging requirements add size to the image. One way to reduce the size of the image is to remove all the packaging requirements from the image. This reduction includes both removing the package manager and its unique dependencies as well as removing the package management data itself. To eliminate all the packaging requirements for an image, be sure that "package-management" is not part of your IMAGE_FEATURES statement for the image. When you remove this feature, you are removing the package manager as well as its dependencies from the root filesystem.

Depending on your particular circumstances, other areas that you can trim likely exist. The key to finding these areas is through tools and methods described here combined with experimentation and iteration. Here are a couple of areas to experiment with: glibc: In general, follow this process: Remove glibc features from DISTRO_FEATURES that you think you do not need. Build your distribution. If the build fails due to missing symbols in a package, determine if you can reconfigure the package to not need those features. For example, change the configuration to not support wide character support as is done for ncurses. Or, if support for those characters is needed, determine what glibc features provide the support and restore the configuration. Rebuild and repeat the process. busybox: For BusyBox, use a process similar as described for glibc. A difference is you will need to boot the resulting system to see if you are able to do everything you expect from the running system. You need to be sure to integrate configuration fragments into Busybox because BusyBox handles its own core features and then allows you to add configuration fragments on top.

If you have not reached your goals on system size, you need to iterate on the process. The process is the same. Use the tools and see just what is taking up 90% of the root filesystem and the kernel. Decide what you can eliminate without limiting your device beyond what you need. Depending on your system, a good place to look might be Busybox, which provides a stripped down version of Unix tools in a single, executable file. You might be able to drop virtual terminal services or perhaps ipv6.

A common scenario developers face is creating images for several different machines that use the same software environment. In this situation, it is tempting to set the tunings and optimization flags for each build specifically for the targeted hardware (i.e. "maxing out" the tunings). Doing so can considerably add to build times and package feed maintenance collectively for the machines. For example, selecting tunes that are extremely specific to a CPU core used in a system might enable some micro optimizations in GCC for that particular system but would otherwise not gain you much of a performance difference across the other systems as compared to using a more general tuning across all the builds (e.g. setting DEFAULTTUNE specifically for each machine's build). Rather than "max out" each build's tunings, you can take steps that cause the OpenEmbedded build system to reuse software across the various machines where it makes sense. If build speed and package feed maintenance are considerations, you should consider the points in this section that can help you optimize your tunings to best consider build times and package feed maintenance. Share the Build Directory: If at all possible, share the TMPDIR across builds. The Yocto Project supports switching between different MACHINE values in the same TMPDIR. This practice is well supported and regularly used by developers when building for multiple machines. When you use the same TMPDIR for multiple machine builds, the OpenEmbedded build system can reuse the existing native and often cross-recipes for multiple machines. Thus, build time decreases. If DISTRO settings change or fundamental configuration settings such as the filesystem layout, you need to work with a clean TMPDIR. Sharing TMPDIR under these circumstances might work but since it is not guaranteed, you should use a clean TMPDIR. Enable the Appropriate Package Architecture: By default, the OpenEmbedded build system enables three levels of package architectures: "all", "tune" or "package", and "machine". Any given recipe usually selects one of these package architectures (types) for its output. Depending for what a given recipe creates packages, making sure you enable the appropriate package architecture can directly impact the build time. A recipe that just generates scripts can enable "all" architecture because there are no binaries to build. To specifically enable "all" architecture, be sure your recipe inherits the allarch class. This class is useful for "all" architectures because it configures many variables so packages can be used across multiple architectures. If your recipe needs to generate packages that are machine-specific or when one of the build or runtime dependencies is already machine-architecture dependent, which makes your recipe also machine-architecture dependent, make sure your recipe enables the "machine" package architecture through the MACHINE_ARCH variable: PACKAGE_ARCH = "${MACHINE_ARCH}" When you do not specifically enable a package architecture through the PACKAGE_ARCH, The OpenEmbedded build system defaults to the TUNE_PKGARCH setting: PACKAGE_ARCH = "${TUNE_PKGARCH}" Choose a Generic Tuning File if Possible: Some tunes are more generic and can run on multiple targets (e.g. an armv5 set of packages could run on armv6 and armv7 processors in most cases). Similarly, i486 binaries could work on i586 and higher processors. You should realize, however, that advances on newer processor versions would not be used. If you select the same tune for several different machines, the OpenEmbedded build system reuses software previously built, thus speeding up the overall build time. Realize that even though a new sysroot for each machine is generated, the software is not recompiled and only one package feed exists. Manage Granular Level Packaging: Sometimes cases exist where injecting another level of package architecture beyond the three higher levels noted earlier can be useful. For example, consider the emgd graphics stack in the meta-intel layer. In this layer, a subset of software exists that is compiled against something different from the rest of the generic packages. You can examine the key code in the Source Repositories "daisy" branch in classes/emgd-gl.bbclass. For a specific set of packages, the code redefines PACKAGE_ARCH. PACKAGE_EXTRA_ARCHS is then appended with this extra tune name in meta-intel-emgd.inc. The result is that when searching for packages, the build system uses a four-level search and the packages in this new level are preferred as compared to the standard tune. The overall result is that the build system reuses most software from the common tune except for specific cases as needed. Use Tools to Debug Issues: Sometimes you can run into situations where software is being rebuilt when you think it should not be. For example, the OpenEmbedded build system might not be using shared state between machines when you think it should be. These types of situations are usually due to references to machine-specific variables such as MACHINE, SERIAL_CONSOLE, XSERVER, MACHINE_FEATURES, and so forth in code that is supposed to only be tune-specific or when the recipe depends (DEPENDS, RDEPENDS, RRECOMMENDS, RSUGGESTS, and so forth) on some other recipe that already has PACKAGE_ARCH defined as "${MACHINE_ARCH}". Patches to fix any issues identified are most welcome as these issues occasionally do occur. For such cases, you can use some tools to help you sort out the situation: sstate-diff-machines.sh: You can find this tool in the scripts directory of the Source Repositories. See the comments in the script for information on how to use the tool. BitBake's "-S printdiff" Option: Using this option causes BitBake to try to establish the closest signature match it can (e.g. in the shared state cache) and then run bitbake-diffsigs over the matches to determine the stamps and delta where these two stamp trees diverge.

This section describes a few tasks that involve packages: Excluding packages from an image Incrementing a binary package version Handling optional module packaging Using Runtime Package Management Setting up and running package test (ptest)

You might find it necessary to prevent specific packages from being installed into an image. If so, you can use several variables to direct the build system to essentially ignore installing recommended packages or to not install a package at all. The following list introduces variables you can use to prevent packages from being installed into your image. Each of these variables only works with IPK and RPM package types. Support for Debian packages does not exist. Also, you can use these variables from your local.conf file or attach them to a specific image recipe by using a recipe name override. For more detail on the variables, see the descriptions in the Yocto Project Reference Manual's glossary chapter. BAD_RECOMMENDATIONS: Use this variable to specify "recommended-only" packages that you do not want installed. NO_RECOMMENDATIONS: Use this variable to prevent all "recommended-only" packages from being installed. PACKAGE_EXCLUDE: Use this variable to prevent specific packages from being installed regardless of whether they are "recommended-only" or not. You need to realize that the build process could fail with an error when you prevent the installation of a package whose presence is required by an installed package.

This section provides some background on how binary package versioning is accomplished and presents some of the services, variables, and terminology involved. In order to understand binary package versioning, you need to consider the following: Binary Package: The binary package that is eventually built and installed into an image. Binary Package Version: The binary package version is composed of two components - a version and a revision. Technically, a third component, the "epoch" (i.e. PE) is involved but this discussion for the most part ignores PE. The version and revision are taken from the PV and PR variables, respectively. PV: The recipe version. PV represents the version of the software being packaged. Do not confuse PV with the binary package version. PR: The recipe revision. SRCPV: The OpenEmbedded build system uses this string to help define the value of PV when the source code revision needs to be included in it. PR Service: A network-based service that helps automate keeping package feeds compatible with existing package manager applications such as RPM, APT, and OPKG. Whenever the binary package content changes, the binary package version must change. Changing the binary package version is accomplished by changing or "bumping" the PR and/or PV values. Increasing these values occurs one of two ways: Automatically using a Package Revision Service (PR Service). Manually incrementing the PR and/or PV variables. Given a primary challenge of any build system and its users is how to maintain a package feed that is compatible with existing package manager applications such as RPM, APT, and OPKG, using an automated system is much preferred over a manual system. In either system, the main requirement is that binary package version numbering increases in a linear fashion and that a number of version components exist that support that linear progression. For information on how to ensure package revisioning remains linear, see the "Automatically Incrementing a Binary Package Revision Number" section. The following three sections provide related information on the PR Service, the manual method for "bumping" PR and/or PV, and on how to ensure binary package revisioning remains linear.

As mentioned, attempting to maintain revision numbers in the Metadata is error prone, inaccurate, and causes problems for people submitting recipes. Conversely, the PR Service automatically generates increasing numbers, particularly the revision field, which removes the human element. For additional information on using a PR Service, you can see the PR Service wiki page. The Yocto Project uses variables in order of decreasing priority to facilitate revision numbering (i.e. PE, PV, and PR for epoch, version, and revision, respectively). The values are highly dependent on the policies and procedures of a given distribution and package feed. Because the OpenEmbedded build system uses "signatures", which are unique to a given build, the build system knows when to rebuild packages. All the inputs into a given task are represented by a signature, which can trigger a rebuild when different. Thus, the build system itself does not rely on the PR, PV, and PE numbers to trigger a rebuild. The signatures, however, can be used to generate these values. The PR Service works with both OEBasic and OEBasicHash generators. The value of PR bumps when the checksum changes and the different generator mechanisms change signatures under different circumstances. As implemented, the build system includes values from the PR Service into the PR field as an addition using the form ".x" so r0 becomes r0.1, r0.2 and so forth. This scheme allows existing PR values to be used for whatever reasons, which include manual PR bumps, should it be necessary. By default, the PR Service is not enabled or running. Thus, the packages generated are just "self consistent". The build system adds and removes packages and there are no guarantees about upgrade paths but images will be consistent and correct with the latest changes. The simplest form for a PR Service is for it to exist for a single host development system that builds the package feed (building system). For this scenario, you can enable a local PR Service by setting PRSERV_HOST in your local.conf file in the Build Directory: PRSERV_HOST = "localhost:0" Once the service is started, packages will automatically get increasing PR values and BitBake takes care of starting and stopping the server. If you have a more complex setup where multiple host development systems work against a common, shared package feed, you have a single PR Service running and it is connected to each building system. For this scenario, you need to start the PR Service using the bitbake-prserv command: bitbake-prserv --host ip --port port --start In addition to hand-starting the service, you need to update the local.conf file of each building system as described earlier so each system points to the server and port. It is also recommended you use build history, which adds some sanity checks to binary package versions, in conjunction with the server that is running the PR Service. To enable build history, add the following to each building system's local.conf file: # It is recommended to activate "buildhistory" for testing the PR service INHERIT += "buildhistory" BUILDHISTORY_COMMIT = "1" For information on build history, see the "Maintaining Build Output Quality" section in the Yocto Project Reference Manual. The OpenEmbedded build system does not maintain PR information as part of the shared state (sstate) packages. If you maintain an sstate feed, its expected that either all your building systems that contribute to the sstate feed use a shared PR Service, or you do not run a PR Service on any of your building systems. Having some systems use a PR Service while others do not leads to obvious problems. For more information on shared state, see the "Shared State Cache" section in the Yocto Project Reference Manual.

The alternative to setting up a PR Service is to manually "bump" the PR variable. If a committed change results in changing the package output, then the value of the PR variable needs to be increased (or "bumped") as part of that commit. For new recipes you should add the PR variable and set its initial value equal to "r0", which is the default. Even though the default value is "r0", the practice of adding it to a new recipe makes it harder to forget to bump the variable when you make changes to the recipe in future. If you are sharing a common .inc file with multiple recipes, you can also use the INC_PR variable to ensure that the recipes sharing the .inc file are rebuilt when the .inc file itself is changed. The .inc file must set INC_PR (initially to "r0"), and all recipes referring to it should set PR to "${INC_PR}.0" initially, incrementing the last number when the recipe is changed. If the .inc file is changed then its INC_PR should be incremented. When upgrading the version of a binary package, assuming the PV changes, the PR variable should be reset to "r0" (or "${INC_PR}.0" if you are using INC_PR). Usually, version increases occur only to binary packages. However, if for some reason PV changes but does not increase, you can increase the PE variable (Package Epoch). The PE variable defaults to "0". Binary package version numbering strives to follow the Debian Version Field Policy Guidelines. These guidelines define how versions are compared and what "increasing" a version means.

When fetching a repository, BitBake uses the SRCREV variable to determine the specific source code revision from which to build. You set the SRCREV variable to AUTOREV to cause the OpenEmbedded build system to automatically use the latest revision of the software: SRCREV = "${AUTOREV}" Furthermore, you need to reference SRCPV in PV in order to automatically update the version whenever the revision of the source code changes. Here is an example: PV = "1.0+git${SRCPV}" The OpenEmbedded build system substitutes SRCPV with the following: AUTOINC+source_code_revision The build system replaces the AUTOINC with a number. The number used depends on the state of the PR Service: If PR Service is enabled, the build system increments the number, which is similar to the behavior of PR. This behavior results in linearly increasing package versions, which is desirable. Here is an example: hello-world-git_0.0+git0+b6558dd387-r0.0_armv7a-neon.ipk hello-world-git_0.0+git1+dd2f5c3565-r0.0_armv7a-neon.ipk If PR Service is not enabled, the build system replaces the AUTOINC placeholder with zero (i.e. "0"). This results in changing the package version since the source revision is included. However, package versions are not increased linearly. Here is an example: hello-world-git_0.0+git0+b6558dd387-r0.0_armv7a-neon.ipk hello-world-git_0.0+git0+dd2f5c3565-r0.0_armv7a-neon.ipk In summary, the OpenEmbedded build system does not track the history of binary package versions for this purpose. AUTOINC, in this case, is comparable to PR. If PR server is not enabled, AUTOINC in the package version is simply replaced by "0". If PR server is enabled, the build system keeps track of the package versions and bumps the number when the package revision changes.

Many pieces of software split functionality into optional modules (or plug-ins) and the plug-ins that are built might depend on configuration options. To avoid having to duplicate the logic that determines what modules are available in your recipe or to avoid having to package each module by hand, the OpenEmbedded build system provides functionality to handle module packaging dynamically. To handle optional module packaging, you need to do two things: Ensure the module packaging is actually done. Ensure that any dependencies on optional modules from other recipes are satisfied by your recipe.

To ensure the module packaging actually gets done, you use the do_split_packages function within the populate_packages Python function in your recipe. The do_split_packages function searches for a pattern of files or directories under a specified path and creates a package for each one it finds by appending to the PACKAGES variable and setting the appropriate values for FILES_packagename, RDEPENDS_packagename, DESCRIPTION_packagename, and so forth. Here is an example from the lighttpd recipe: python populate_packages_prepend () { lighttpd_libdir = d.expand('${libdir}') do_split_packages(d, lighttpd_libdir, '^mod_(.*)\.so$', 'lighttpd-module-%s', 'Lighttpd module for %s', extra_depends='') } The previous example specifies a number of things in the call to do_split_packages. A directory within the files installed by your recipe through do_install in which to search. A regular expression used to match module files in that directory. In the example, note the parentheses () that mark the part of the expression from which the module name should be derived. A pattern to use for the package names. A description for each package. An empty string for extra_depends, which disables the default dependency on the main lighttpd package. Thus, if a file in ${libdir} called mod_alias.so is found, a package called lighttpd-module-alias is created for it and the DESCRIPTION is set to "Lighttpd module for alias". Often, packaging modules is as simple as the previous example. However, more advanced options exist that you can use within do_split_packages to modify its behavior. And, if you need to, you can add more logic by specifying a hook function that is called for each package. It is also perfectly acceptable to call do_split_packages multiple times if you have more than one set of modules to package. For more examples that show how to use do_split_packages, see the connman.inc file in the meta/recipes-connectivity/connman/ directory of the poky source repository. You can also find examples in meta/classes/kernel.bbclass. Following is a reference that shows do_split_packages mandatory and optional arguments: Mandatory arguments root The path in which to search file_regex Regular expression to match searched files. Use parentheses () to mark the part of this expression that should be used to derive the module name (to be substituted where %s is used in other function arguments as noted below) output_pattern Pattern to use for the package names. Must include %s. description Description to set for each package. Must include %s. Optional arguments postinst Postinstall script to use for all packages (as a string) recursive True to perform a recursive search - default False hook A hook function to be called for every match. The function will be called with the following arguments (in the order listed): f Full path to the file/directory match pkg The package name file_regex As above output_pattern As above modulename The module name derived using file_regex extra_depends Extra runtime dependencies (RDEPENDS) to be set for all packages. The default value of None causes a dependency on the main package (${PN}) - if you do not want this, pass empty string '' for this parameter. aux_files_pattern Extra item(s) to be added to FILES for each package. Can be a single string item or a list of strings for multiple items. Must include %s. postrm postrm script to use for all packages (as a string) allow_dirs True to allow directories to be matched - default False prepend If True, prepend created packages to PACKAGES instead of the default False which appends them match_path match file_regex on the whole relative path to the root rather than just the file name aux_files_pattern_verbatim Extra item(s) to be added to FILES for each package, using the actual derived module name rather than converting it to something legal for a package name. Can be a single string item or a list of strings for multiple items. Must include %s. allow_links True to allow symlinks to be matched - default False summary Summary to set for each package. Must include %s; defaults to description if not set.

The second part for handling optional module packaging is to ensure that any dependencies on optional modules from other recipes are satisfied by your recipe. You can be sure these dependencies are satisfied by using the PACKAGES_DYNAMIC variable. Here is an example that continues with the lighttpd recipe shown earlier: PACKAGES_DYNAMIC = "lighttpd-module-.*" The name specified in the regular expression can of course be anything. In this example, it is lighttpd-module- and is specified as the prefix to ensure that any RDEPENDS and RRECOMMENDS on a package name starting with the prefix are satisfied during build time. If you are using do_split_packages as described in the previous section, the value you put in PACKAGES_DYNAMIC should correspond to the name pattern specified in the call to do_split_packages.

During a build, BitBake always transforms a recipe into one or more packages. For example, BitBake takes the bash recipe and currently produces the bash-dbg, bash-staticdev, bash-dev, bash-doc, bash-locale, and bash packages. Not all generated packages are included in an image. In several situations, you might need to update, add, remove, or query the packages on a target device at runtime (i.e. without having to generate a new image). Examples of such situations include: You want to provide in-the-field updates to deployed devices (e.g. security updates). You want to have a fast turn-around development cycle for one or more applications that run on your device. You want to temporarily install the "debug" packages of various applications on your device so that debugging can be greatly improved by allowing access to symbols and source debugging. You want to deploy a more minimal package selection of your device but allow in-the-field updates to add a larger selection for customization. In all these situations, you have something similar to a more traditional Linux distribution in that in-field devices are able to receive pre-compiled packages from a server for installation or update. Being able to install these packages on a running, in-field device is what is termed "runtime package management". In order to use runtime package management, you need a host/server machine that serves up the pre-compiled packages plus the required metadata. You also need package manipulation tools on the target. The build machine is a likely candidate to act as the server. However, that machine does not necessarily have to be the package server. The build machine could push its artifacts to another machine that acts as the server (e.g. Internet-facing). A simple build that targets just one device produces more than one package database. In other words, the packages produced by a build are separated out into a couple of different package groupings based on criteria such as the target's CPU architecture, the target board, or the C library used on the target. For example, a build targeting the qemuarm device produces the following three package databases: all, armv5te, and qemuarm. If you wanted your qemuarm device to be aware of all the packages that were available to it, you would need to point it to each of these databases individually. In a similar way, a traditional Linux distribution usually is configured to be aware of a number of software repositories from which it retrieves packages. Using runtime package management is completely optional and not required for a successful build or deployment in any way. But if you want to make use of runtime package management, you need to do a couple things above and beyond the basics. The remainder of this section describes what you need to do.

This section describes build considerations of which you need to be aware in order to provide support for runtime package management. When BitBake generates packages, it needs to know what format or formats to use. In your configuration, you use the PACKAGE_CLASSES variable to specify the format: Open the local.conf file inside your Build Directory (e.g. ~/poky/build/conf/local.conf). Select the desired package format as follows: PACKAGE_CLASSES ?= “package_packageformat” where packageformat can be "ipk", "rpm", and "deb", which are the supported package formats. Because the Yocto Project supports three different package formats, you can set the variable with more than one argument. However, the OpenEmbedded build system only uses the first argument when creating an image or Software Development Kit (SDK). If you would like your image to start off with a basic package database containing the packages in your current build as well as to have the relevant tools available on the target for runtime package management, you can include "package-management" in the IMAGE_FEATURES variable. Including "package-management" in this configuration variable ensures that when the image is assembled for your target, the image includes the currently-known package databases as well as the target-specific tools required for runtime package management to be performed on the target. However, this is not strictly necessary. You could start your image off without any databases but only include the required on-target package tool(s). As an example, you could include "opkg" in your IMAGE_INSTALL variable if you are using the IPK package format. You can then initialize your target's package database(s) later once your image is up and running. Whenever you perform any sort of build step that can potentially generate a package or modify an existing package, it is always a good idea to re-generate the package index with: $ bitbake package-index Realize that it is not sufficient to simply do the following: $ bitbake some-package package-index The reason for this restriction is because BitBake does not properly schedule the package-index target fully after any other target has completed. Thus, be sure to run the package update step separately. You can use the PACKAGE_FEED_ARCHS, PACKAGE_FEED_BASE_PATHS, and PACKAGE_FEED_URIS variables to pre-configure target images to use a package feed. If you do not define these variables, then manual steps as described in the subsequent sections are necessary to configure the target. You should set these variables before building the image in order to produce a correctly configured image. When your build is complete, your packages reside in the ${TMPDIR}/deploy/packageformat directory. For example, if ${TMPDIR} is tmp and your selected package type is IPK, then your IPK packages are available in tmp/deploy/ipk.

Although other protocols are possible, a server using HTTP typically serves packages. If you want to use HTTP, then set up and configure a web server such as Apache 2, lighttpd, or SimpleHTTPServer on the machine serving the packages. To keep things simple, this section describes how to set up a SimpleHTTPServer web server to share package feeds from the developer's machine. Although this server might not be the best for a production environment, the setup is simple and straight forward. Should you want to use a different server more suited for production (e.g. Apache 2, Lighttpd, or Nginx), take the appropriate steps to do so. From within the build directory where you have built an image based on your packaging choice (i.e. the PACKAGE_CLASSES setting), simply start the server. The following example assumes a build directory of ~/poky/build/tmp/deploy/rpm and a PACKAGE_CLASSES setting of "package_rpm": $ cd ~/poky/build/tmp/deploy/rpm $ python -m SimpleHTTPServer

Setting up the target differs depending on the package management system. This section provides information for RPM, IPK, and DEB.

The dnf application performs runtime package management of RPM packages. You must perform an initial setup for dnf on the target machine if the PACKAGE_FEED_ARCHS, PACKAGE_FEED_BASE_PATHS, and PACKAGE_FEED_URIS variables have not been set or the target image was built before the variables were set. As an example, assume the target is able to use the following package databases: all, i586, and qemux86 from a server named my.server. You must inform dnf of the availability of these databases by creating a /etc/yum.repos.d/oe-packages.repo file with the following content: [oe-packages] baseurl=http://my.server/rpm/i586 http://my.server/rpm/qemux86 http://my.server/rpm/all From the target machine, fetch the repository: # dnf makecache After everything is set up, dnf is able to find, install, and upgrade packages from the specified repository. See the DNF documentation for additional information.

The opkg application performs runtime package management of IPK packages. You must perform an initial setup for opkg on the target machine if the PACKAGE_FEED_ARCHS, PACKAGE_FEED_BASE_PATHS, and PACKAGE_FEED_URIS variables have not been set or the target image was built before the variables were set. The opkg application uses configuration files to find available package databases. Thus, you need to create a configuration file inside the /etc/opkg/ direction, which informs opkg of any repository you want to use. As an example, suppose you are serving packages from a ipk/ directory containing the i586, all, and qemux86 databases through an HTTP server named my.server. On the target, create a configuration file (e.g. my_repo.conf) inside the /etc/opkg/ directory containing the following: src/gz all http://my.server/ipk/all src/gz i586 http://my.server/ipk/i586 src/gz qemux86 http://my.server/ipk/qemux86 Next, instruct opkg to fetch the repository information: # opkg update The opkg application is now able to find, install, and upgrade packages from the specified repository.

The apt application performs runtime package management of DEB packages. This application uses a source list file to find available package databases. You must perform an initial setup for apt on the target machine if the PACKAGE_FEED_ARCHS, PACKAGE_FEED_BASE_PATHS, and PACKAGE_FEED_URIS variables have not been set or the target image was built before the variables were set. To inform apt of the repository you want to use, you might create a list file (e.g. my_repo.list) inside the /etc/apt/sources.list.d/ directory. As an example, suppose you are serving packages from a deb/ directory containing the i586, all, and qemux86 databases through an HTTP server named my.server. The list file should contain: deb http://my.server/deb/all ./ deb http://my.server/deb/i586 ./ deb http://my.server/deb/qemux86 ./ Next, instruct the apt application to fetch the repository information: # apt-get update After this step, apt is able to find, install, and upgrade packages from the specified repository.

In order to add security to RPM packages used during a build, you can take steps to securely sign them. Once a signature is verified, the OpenEmbedded build system can use the package in the build. If security fails for a signed package, the build system aborts the build. This section describes how to sign RPM packages during a build and how to use signed package feeds (repositories) when doing a build.

To enable signing RPM packages, you must set up the following configurations in either your local.config or distro.config file: # Inherit sign_rpm.bbclass to enable signing functionality INHERIT += " sign_rpm" # Define the GPG key that will be used for signing. RPM_GPG_NAME = "key_name" # Provide passphrase for the key RPM_GPG_PASSPHRASE = "passphrase" Be sure to supply appropriate values for both key_name and passphrase Aside from the RPM_GPG_NAME and RPM_GPG_PASSPHRASE variables in the previous example, two optional variables related to signing exist: GPG_BIN: Specifies a gpg binary/wrapper that is executed when the package is signed. GPG_PATH: Specifies the gpg home directory used when the package is signed.

In addition to being able to sign RPM packages, you can also enable signed package feeds for IPK and RPM packages. The steps you need to take to enable signed package feed use are similar to the steps used to sign RPM packages. You must define the following in your local.config or distro.config file: INHERIT += "sign_package_feed" PACKAGE_FEED_GPG_NAME = "key_name" PACKAGE_FEED_GPG_PASSPHRASE_FILE = "path_to_file_containing_passphrase" For signed package feeds, the passphrase must exist in a separate file, which is pointed to by the PACKAGE_FEED_GPG_PASSPHRASE_FILE variable. Regarding security, keeping a plain text passphrase out of the configuration is more secure. Aside from the PACKAGE_FEED_GPG_NAME and PACKAGE_FEED_GPG_PASSPHRASE_FILE variables, three optional variables related to signed package feeds exist: GPG_BIN: Specifies a gpg binary/wrapper that is executed when the package is signed. GPG_PATH: Specifies the gpg home directory used when the package is signed. PACKAGE_FEED_GPG_SIGNATURE_TYPE: Specifies the type of gpg signature. This variable applies only to RPM and IPK package feeds. Allowable values for the PACKAGE_FEED_GPG_SIGNATURE_TYPE are "ASC", which is the default and specifies ascii armored, and "BIN", which specifies binary.

A Package Test (ptest) runs tests against packages built by the OpenEmbedded build system on the target machine. A ptest contains at least two items: the actual test, and a shell script (run-ptest) that starts the test. The shell script that starts the test must not contain the actual test - the script only starts the test. On the other hand, the test can be anything from a simple shell script that runs a binary and checks the output to an elaborate system of test binaries and data files. The test generates output in the format used by Automake: result: testname where the result can be PASS, FAIL, or SKIP, and the testname can be any identifying string. For a list of Yocto Project recipes that are already enabled with ptest, see the Ptest wiki page. A recipe is "ptest-enabled" if it inherits the ptest class.

To add package testing to your build, add the DISTRO_FEATURES and EXTRA_IMAGE_FEATURES variables to your local.conf file, which is found in the Build Directory: DISTRO_FEATURES_append = " ptest" EXTRA_IMAGE_FEATURES += "ptest-pkgs" Once your build is complete, the ptest files are installed into the /usr/lib/package/ptest directory within the image, where package is the name of the package.

The ptest-runner package installs a shell script that loops through all installed ptest test suites and runs them in sequence. Consequently, you might want to add this package to your image.

In order to enable a recipe to run installed ptests on target hardware, you need to prepare the recipes that build the packages you want to test. Here is what you have to do for each recipe: Be sure the recipe inherits the ptest class: Include the following line in each recipe: inherit ptest Create run-ptest: This script starts your test. Locate the script where you will refer to it using SRC_URI. Here is an example that starts a test for dbus: #!/bin/sh cd test make -k runtest-TESTS Ensure dependencies are met: If the test adds build or runtime dependencies that normally do not exist for the package (such as requiring "make" to run the test suite), use the DEPENDS and RDEPENDS variables in your recipe in order for the package to meet the dependencies. Here is an example where the package has a runtime dependency on "make": RDEPENDS_${PN}-ptest += "make" Add a function to build the test suite: Not many packages support cross-compilation of their test suites. Consequently, you usually need to add a cross-compilation function to the package. Many packages based on Automake compile and run the test suite by using a single command such as make check. However, the host make check builds and runs on the same computer, while cross-compiling requires that the package is built on the host but executed for the target architecture (though often, as in the case for ptest, the execution occurs on the host). The built version of Automake that ships with the Yocto Project includes a patch that separates building and execution. Consequently, packages that use the unaltered, patched version of make check automatically cross-compiles. Regardless, you still must add a do_compile_ptest function to build the test suite. Add a function similar to the following to your recipe: do_compile_ptest() { oe_runmake buildtest-TESTS } Ensure special configurations are set: If the package requires special configurations prior to compiling the test code, you must insert a do_configure_ptest function into the recipe. Install the test suite: The ptest class automatically copies the file run-ptest to the target and then runs make install-ptest to run the tests. If this is not enough, you need to create a do_install_ptest function and make sure it gets called after the "make install-ptest" completes.

The OpenEmbedded build system works with source files located through the SRC_URI variable. When you build something using BitBake, a big part of the operation is locating and downloading all the source tarballs. For images, downloading all the source for various packages can take a significant amount of time. This section presents information for working with source files that can lead to more efficient use of resources and time.

As mentioned, a good deal that goes into a Yocto Project build is simply downloading all of the source tarballs. Maybe you have been working with another build system (OpenEmbedded or Angstrom) for which you have built up a sizable directory of source tarballs. Or, perhaps someone else has such a directory for which you have read access. If so, you can save time by adding statements to your configuration file so that the build process checks local directories first for existing tarballs before checking the Internet. Here is an efficient way to set it up in your local.conf file: SOURCE_MIRROR_URL ?= "file:///home/you/your-download-dir/" INHERIT += "own-mirrors" BB_GENERATE_MIRROR_TARBALLS = "1" # BB_NO_NETWORK = "1" In the previous example, the BB_GENERATE_MIRROR_TARBALLS variable causes the OpenEmbedded build system to generate tarballs of the Git repositories and store them in the DL_DIR directory. Due to performance reasons, generating and storing these tarballs is not the build system's default behavior. You can also use the PREMIRRORS variable. For an example, see the variable's glossary entry in the Yocto Project Reference Manual.

Another technique you can use to ready yourself for a successive string of build operations, is to pre-fetch all the source files without actually starting a build. This technique lets you work through any download issues and ultimately gathers all the source files into your download directory build/downloads, which is located with DL_DIR. Use the following BitBake command form to fetch all the necessary sources without starting the build: $ bitbake -c fetchall target This variation of the BitBake command guarantees that you have all the sources for that BitBake target should you disconnect from the Internet and want to do the build later offline.

By default, the OpenEmbedded build system uses the Build Directory when building source code. The build process involves fetching the source files, unpacking them, and then patching them if necessary before the build takes place. Situations exist where you might want to build software from source files that are external to and thus outside of the OpenEmbedded build system. For example, suppose you have a project that includes a new BSP with a heavily customized kernel. And, you want to minimize exposing the build system to the development team so that they can focus on their project and maintain everyone's workflow as much as possible. In this case, you want a kernel source directory on the development machine where the development occurs. You want the recipe's SRC_URI variable to point to the external directory and use it as is, not copy it. To build from software that comes from an external source, all you need to do is inherit the externalsrc class and then set the EXTERNALSRC variable to point to your external source code. Here are the statements to put in your local.conf file: INHERIT += "externalsrc" EXTERNALSRC_pn-myrecipe = "path-to-your-source-tree" This next example shows how to accomplish the same thing by setting EXTERNALSRC in the recipe itself or in the recipe's append file: EXTERNALSRC = "path" EXTERNALSRC_BUILD = "path" In order for these settings to take effect, you must globally or locally inherit the externalsrc class. By default, externalsrc.bbclass builds the source code in a directory separate from the external source directory as specified by EXTERNALSRC. If you need to have the source built in the same directory in which it resides, or some other nominated directory, you can set EXTERNALSRC_BUILD to point to that directory: EXTERNALSRC_BUILD_pn-myrecipe = "path-to-your-source-tree"

By default, the Yocto Project uses SysVinit as the initialization manager. However, support also exists for systemd, which is a full replacement for init with parallel starting of services, reduced shell overhead and other features that are used by many distributions. If you want to use SysVinit, you do not have to do anything. But, if you want to use systemd, you must take some steps as described in the following sections.

Set the these variables in your distribution configuration file as follows: DISTRO_FEATURES_append = " systemd" VIRTUAL-RUNTIME_init_manager = "systemd" You can also prevent the SysVinit distribution feature from being automatically enabled as follows: DISTRO_FEATURES_BACKFILL_CONSIDERED = "sysvinit" Doing so removes any redundant SysVinit scripts. To remove initscripts from your image altogether, set this variable also: VIRTUAL-RUNTIME_initscripts = "" For information on the backfill variable, see DISTRO_FEATURES_BACKFILL_CONSIDERED.

Set these variables in your distribution configuration file as follows: DISTRO_FEATURES_append = " systemd" VIRTUAL-RUNTIME_init_manager = "systemd" Doing so causes your main image to use the packagegroup-core-boot.bb recipe and systemd. The rescue/minimal image cannot use this package group. However, it can install SysVinit and the appropriate packages will have support for both systemd and SysVinit.

The Yocto Project provides multiple ways to manage the device manager (/dev): Persistent and Pre-Populated/dev: For this case, the /dev directory is persistent and the required device nodes are created during the build. Use devtmpfs with a Device Manager: For this case, the /dev directory is provided by the kernel as an in-memory file system and is automatically populated by the kernel at runtime. Additional configuration of device nodes is done in user space by a device manager like udev or busybox-mdev.

To use the static method for device population, you need to set the USE_DEVFS variable to "0" as follows: USE_DEVFS = "0" The content of the resulting /dev directory is defined in a Device Table file. The IMAGE_DEVICE_TABLES variable defines the Device Table to use and should be set in the machine or distro configuration file. Alternatively, you can set this variable in your local.conf configuration file. If you do not define the IMAGE_DEVICE_TABLES variable, the default device_table-minimal.txt is used: IMAGE_DEVICE_TABLES = "device_table-mymachine.txt" The population is handled by the makedevs utility during image creation:

To use the dynamic method for device population, you need to use (or be sure to set) the USE_DEVFS variable to "1", which is the default: USE_DEVFS = "1" With this setting, the resulting /dev directory is populated by the kernel using devtmpfs. Make sure the corresponding kernel configuration variable CONFIG_DEVTMPFS is set when building you build a Linux kernel. All devices created by devtmpfs will be owned by root and have permissions 0600. To have more control over the device nodes, you can use a device manager like udev or busybox-mdev. You choose the device manager by defining the VIRTUAL-RUNTIME_dev_manager variable in your machine or distro configuration file. Alternatively, you can set this variable in your local.conf configuration file: VIRTUAL-RUNTIME_dev_manager = "udev" # Some alternative values # VIRTUAL-RUNTIME_dev_manager = "busybox-mdev" # VIRTUAL-RUNTIME_dev_manager = "systemd"

If you're working on a recipe that pulls from an external Source Code Manager (SCM), it is possible to have the OpenEmbedded build system notice new recipe changes added to the SCM and then build the resulting packages that depend on the new recipes by using the latest versions. This only works for SCMs from which it is possible to get a sensible revision number for changes. Currently, you can do this with Apache Subversion (SVN), Git, and Bazaar (BZR) repositories. To enable this behavior, the PV of the recipe needs to reference SRCPV. Here is an example: PV = "1.2.3+git${SRCPV}" Then, you can add the following to your local.conf: SRCREV_pn-PN = "${AUTOREV}" PN is the name of the recipe for which you want to enable automatic source revision updating. If you do not want to update your local configuration file, you can add the following directly to the recipe to finish enabling the feature: SRCREV = "${AUTOREV}" The Yocto Project provides a distribution named poky-bleeding, whose configuration file contains the line: require conf/distro/include/poky-floating-revisions.inc This line pulls in the listed include file that contains numerous lines of exactly that form: #SRCREV_pn-opkg-native ?= "${AUTOREV}" #SRCREV_pn-opkg-sdk ?= "${AUTOREV}" #SRCREV_pn-opkg ?= "${AUTOREV}" #SRCREV_pn-opkg-utils-native ?= "${AUTOREV}" #SRCREV_pn-opkg-utils ?= "${AUTOREV}" SRCREV_pn-gconf-dbus ?= "${AUTOREV}" SRCREV_pn-matchbox-common ?= "${AUTOREV}" SRCREV_pn-matchbox-config-gtk ?= "${AUTOREV}" SRCREV_pn-matchbox-desktop ?= "${AUTOREV}" SRCREV_pn-matchbox-keyboard ?= "${AUTOREV}" SRCREV_pn-matchbox-panel-2 ?= "${AUTOREV}" SRCREV_pn-matchbox-themes-extra ?= "${AUTOREV}" SRCREV_pn-matchbox-terminal ?= "${AUTOREV}" SRCREV_pn-matchbox-wm ?= "${AUTOREV}" SRCREV_pn-settings-daemon ?= "${AUTOREV}" SRCREV_pn-screenshot ?= "${AUTOREV}" . . . These lines allow you to experiment with building a distribution that tracks the latest development source for numerous packages. The poky-bleeding distribution is not tested on a regular basis. Keep this in mind if you use it.

Suppose, for security reasons, you need to disable your target device's root filesystem's write permissions (i.e. you need a read-only root filesystem). Or, perhaps you are running the device's operating system from a read-only storage device. For either case, you can customize your image for that behavior. Supporting a read-only root filesystem requires that the system and applications do not try to write to the root filesystem. You must configure all parts of the target system to write elsewhere, or to gracefully fail in the event of attempting to write to the root filesystem.

To create the read-only root filesystem, simply add the "read-only-rootfs" feature to your image. Using either of the following statements in your image recipe or from within the local.conf file found in the Build Directory causes the build system to create a read-only root filesystem: IMAGE_FEATURES = "read-only-rootfs" or EXTRA_IMAGE_FEATURES += "read-only-rootfs" For more information on how to use these variables, see the "Customizing Images Using Custom IMAGE_FEATURES and EXTRA_IMAGE_FEATURES" section. For information on the variables, see IMAGE_FEATURES and EXTRA_IMAGE_FEATURES.

It is very important that you make sure all post-Installation (pkg_postinst) scripts for packages that are installed into the image can be run at the time when the root filesystem is created during the build on the host system. These scripts cannot attempt to run during first-boot on the target device. With the "read-only-rootfs" feature enabled, the build system checks during root filesystem creation to make sure all post-installation scripts succeed. If any of these scripts still need to be run after the root filesystem is created, the build immediately fails. These build-time checks ensure that the build fails rather than the target device fails later during its initial boot operation. Most of the common post-installation scripts generated by the build system for the out-of-the-box Yocto Project are engineered so that they can run during root filesystem creation (e.g. post-installation scripts for caching fonts). However, if you create and add custom scripts, you need to be sure they can be run during this file system creation. Here are some common problems that prevent post-installation scripts from running during root filesystem creation: Not using $D in front of absolute paths: The build system defines $D when the root filesystem is created. Furthermore, $D is blank when the script is run on the target device. This implies two purposes for $D: ensuring paths are valid in both the host and target environments, and checking to determine which environment is being used as a method for taking appropriate actions. Attempting to run processes that are specific to or dependent on the target architecture: You can work around these attempts by using native tools, which run on the host system, to accomplish the same tasks, or by alternatively running the processes under QEMU, which has the qemu_run_binary function. For more information, see the qemu class.

With the "read-only-rootfs" feature enabled, any attempt by the target to write to the root filesystem at runtime fails. Consequently, you must make sure that you configure processes and applications that attempt these types of writes do so to directories with write access (e.g. /tmp or /var/run).

The OpenEmbedded build system makes available a series of automated tests for images to verify runtime functionality. You can run these tests on either QEMU or actual target hardware. Tests are written in Python making use of the unittest module, and the majority of them run commands on the target system over SSH. This section describes how you set up the environment to use these tests, run available tests, and write and add your own tests. For information on the test and QA infrastructure available within the Yocto Project, see the "Testing and Quality Assurance" section in the Yocto Project Reference Manual.

Depending on whether you are planning to run tests using QEMU or on the hardware, you have to take different steps to enable the tests. See the following subsections for information on how to enable both types of tests.

In order to run tests, you need to do the following: Set up to avoid interaction with sudo for networking: To accomplish this, you must do one of the following: Add NOPASSWD for your user in /etc/sudoers either for all commands or just for runqemu-ifup. You must provide the full path as that can change if you are using multiple clones of the source repository. On some distributions, you also need to comment out "Defaults requiretty" in /etc/sudoers. Manually configure a tap interface for your system. Run as root the script in scripts/runqemu-gen-tapdevs, which should generate a list of tap devices. This is the option typically chosen for Autobuilder-type environments. Set the DISPLAY variable: You need to set this variable so that you have an X server available (e.g. start vncserver for a headless machine). Be sure your host's firewall accepts incoming connections from 192.168.7.0/24: Some of the tests (in particular DNF tests) start an HTTP server on a random high number port, which is used to serve files to the target. The DNF module serves ${WORKDIR}/oe-rootfs-repo so it can run DNF channel commands. That means your host's firewall must accept incoming connections from 192.168.7.0/24, which is the default IP range used for tap devices by runqemu. Be sure your host has the correct packages installed: Depending your host's distribution, you need to have the following packages installed: Ubuntu and Debian: sysstat and iproute2 OpenSUSE: sysstat and iproute2 Fedora: sysstat and iproute CentOS: sysstat and iproute Once you start running the tests, the following happens: A copy of the root filesystem is written to ${WORKDIR}/testimage. The image is booted under QEMU using the standard runqemu script. A default timeout of 500 seconds occurs to allow for the boot process to reach the login prompt. You can change the timeout period by setting TEST_QEMUBOOT_TIMEOUT in the local.conf file. Once the boot process is reached and the login prompt appears, the tests run. The full boot log is written to ${WORKDIR}/testimage/qemu_boot_log. Each test module loads in the order found in TEST_SUITES. You can find the full output of the commands run over SSH in ${WORKDIR}/testimgage/ssh_target_log. If no failures occur, the task running the tests ends successfully. You can find the output from the unittest in the task log at ${WORKDIR}/temp/log.do_testimage.

The OpenEmbedded build system can run tests on real hardware, and for certain devices it can also deploy the image to be tested onto the device beforehand. For automated deployment, a "master image" is installed onto the hardware once as part of setup. Then, each time tests are to be run, the following occurs: The master image is booted into and used to write the image to be tested to a second partition. The device is then rebooted using an external script that you need to provide. The device boots into the image to be tested. When running tests (independent of whether the image has been deployed automatically or not), the device is expected to be connected to a network on a pre-determined IP address. You can either use static IP addresses written into the image, or set the image to use DHCP and have your DHCP server on the test network assign a known IP address based on the MAC address of the device. In order to run tests on hardware, you need to set TEST_TARGET to an appropriate value. For QEMU, you do not have to change anything, the default value is "QemuTarget". For running tests on hardware, the following options exist: "SimpleRemoteTarget": Choose "SimpleRemoteTarget" if you are going to run tests on a target system that is already running the image to be tested and is available on the network. You can use "SimpleRemoteTarget" in conjunction with either real hardware or an image running within a separately started QEMU or any other virtual machine manager. "Systemd-bootTarget": Choose "Systemd-bootTarget" if your hardware is an EFI-based machine with systemd-boot as bootloader and core-image-testmaster (or something similar) is installed. Also, your hardware under test must be in a DHCP-enabled network that gives it the same IP address for each reboot. If you choose "Systemd-bootTarget", there are additional requirements and considerations. See the "Selecting Systemd-bootTarget" section, which follows, for more information. "BeagleBoneTarget": Choose "BeagleBoneTarget" if you are deploying images and running tests on the BeagleBone "Black" or original "White" hardware. For information on how to use these tests, see the comments at the top of the BeagleBoneTarget meta-yocto-bsp/lib/oeqa/controllers/beaglebonetarget.py file. "EdgeRouterTarget": Choose "EdgeRouterTarget" is you are deploying images and running tests on the Ubiquiti Networks EdgeRouter Lite. For information on how to use these tests, see the comments at the top of the EdgeRouterTarget meta-yocto-bsp/lib/oeqa/controllers/edgeroutertarget.py file. "GrubTarget": Choose the "supports deploying images and running tests on any generic PC that boots using GRUB. For information on how to use these tests, see the comments at the top of the GrubTarget meta-yocto-bsp/lib/oeqa/controllers/grubtarget.py file. "your-target": Create your own custom target if you want to run tests when you are deploying images and running tests on a custom machine within your BSP layer. To do this, you need to add a Python unit that defines the target class under lib/oeqa/controllers/ within your layer. You must also provide an empty __init__.py. For examples, see files in meta-yocto-bsp/lib/oeqa/controllers/.

If you did not set TEST_TARGET to "Systemd-bootTarget", then you do not need any information in this section. You can skip down to the "Running Tests" section. If you did set TEST_TARGET to "Systemd-bootTarget", you also need to perform a one-time setup of your master image by doing the following: Set EFI_PROVIDER: Be sure that EFI_PROVIDER is as follows: EFI_PROVIDER = "systemd-boot" Build the master image: Build the core-image-testmaster image. The core-image-testmaster recipe is provided as an example for a "master" image and you can customize the image recipe as you would any other recipe. Here are the image recipe requirements: Inherits core-image so that kernel modules are installed. Installs normal linux utilities not busybox ones (e.g. bash, coreutils, tar, gzip, and kmod). Uses a custom Initial RAM Disk (initramfs) image with a custom installer. A normal image that you can install usually creates a single rootfs partition. This image uses another installer that creates a specific partition layout. Not all Board Support Packages (BSPs) can use an installer. For such cases, you need to manually create the following partition layout on the target: First partition mounted under /boot, labeled "boot". The main rootfs partition where this image gets installed, which is mounted under /. Another partition labeled "testrootfs" where test images get deployed. Install image: Install the image that you just built on the target system. The final thing you need to do when setting TEST_TARGET to "Systemd-bootTarget" is to set up the test image: Set up your local.conf file: Make sure you have the following statements in your local.conf file: IMAGE_FSTYPES += "tar.gz" INHERIT += "testimage" TEST_TARGET = "Systemd-bootTarget" TEST_TARGET_IP = "192.168.2.3" Build your test image: Use BitBake to build the image: $ bitbake core-image-sato

For most hardware targets other than SimpleRemoteTarget, you can control power: You can use TEST_POWERCONTROL_CMD together with TEST_POWERCONTROL_EXTRA_ARGS as a command that runs on the host and does power cycling. The test code passes one argument to that command: off, on or cycle (off then on). Here is an example that could appear in your local.conf file: TEST_POWERCONTROL_CMD = "powercontrol.exp test 10.11.12.1 nuc1" In this example, the expect script does the following: ssh test@10.11.12.1 "pyctl nuc1 arg" It then runs a Python script that controls power for a label called nuc1. You need to customize TEST_POWERCONTROL_CMD and TEST_POWERCONTROL_EXTRA_ARGS for your own setup. The one requirement is that it accepts "on", "off", and "cycle" as the last argument. When no command is defined, it connects to the device over SSH and uses the classic reboot command to reboot the device. Classic reboot is fine as long as the machine actually reboots (i.e. the SSH test has not failed). It is useful for scenarios where you have a simple setup, typically with a single board, and where some manual interaction is okay from time to time. If you have no hardware to automatically perform power control but still wish to experiment with automated hardware testing, you can use the dialog-power-control script that shows a dialog prompting you to perform the required power action. This script requires either KDialog or Zenity to be installed. To use this script, set the TEST_POWERCONTROL_CMD variable as follows: TEST_POWERCONTROL_CMD = "${COREBASE}/scripts/contrib/dialog-power-control"

For test target classes requiring a serial console to interact with the bootloader (e.g. BeagleBoneTarget, EdgeRouterTarget, and GrubTarget), you need to specify a command to use to connect to the serial console of the target machine by using the TEST_SERIALCONTROL_CMD variable and optionally the TEST_SERIALCONTROL_EXTRA_ARGS variable. These cases could be a serial terminal program if the machine is connected to a local serial port, or a telnet or ssh command connecting to a remote console server. Regardless of the case, the command simply needs to connect to the serial console and forward that connection to standard input and output as any normal terminal program does. For example, to use the picocom terminal program on serial device /dev/ttyUSB0 at 115200bps, you would set the variable as follows: TEST_SERIALCONTROL_CMD = "picocom /dev/ttyUSB0 -b 115200" For local devices where the serial port device disappears when the device reboots, an additional "serdevtry" wrapper script is provided. To use this wrapper, simply prefix the terminal command with ${COREBASE}/scripts/contrib/serdevtry: TEST_SERIALCONTROL_CMD = "${COREBASE}/scripts/contrib/serdevtry picocom -b 115200 /dev/ttyUSB0"

You can start the tests automatically or manually: Automatically running tests: To run the tests automatically after the OpenEmbedded build system successfully creates an image, first set the TEST_IMAGE variable to "1" in your local.conf file in the Build Directory: TEST_IMAGE = "1" Next, build your image. If the image successfully builds, the tests will be run: bitbake core-image-sato Manually running tests: To manually run the tests, first globally inherit the testimage class by editing your local.conf file: INHERIT += "testimage" Next, use BitBake to run the tests: bitbake -c testimage image All test files reside in meta/lib/oeqa/runtime in the Source Directory. A test name maps directly to a Python module. Each test module may contain a number of individual tests. Tests are usually grouped together by the area tested (e.g tests for systemd reside in meta/lib/oeqa/runtime/systemd.py). You can add tests to any layer provided you place them in the proper area and you extend BBPATH in the local.conf file as normal. Be sure that tests reside in layer/lib/oeqa/runtime. Be sure that module names do not collide with module names used in the default set of test modules in meta/lib/oeqa/runtime. You can change the set of tests run by appending or overriding TEST_SUITES variable in local.conf. Each name in TEST_SUITES represents a required test for the image. Test modules named within TEST_SUITES cannot be skipped even if a test is not suitable for an image (e.g. running the RPM tests on an image without rpm). Appending "auto" to TEST_SUITES causes the build system to try to run all tests that are suitable for the image (i.e. each test module may elect to skip itself). The order you list tests in TEST_SUITES is important and influences test dependencies. Consequently, tests that depend on other tests should be added after the test on which they depend. For example, since the ssh test depends on the ping test, "ssh" needs to come after "ping" in the list. The test class provides no re-ordering or dependency handling. Each module can have multiple classes with multiple test methods. And, Python unittest rules apply. Here are some things to keep in mind when running tests: The default tests for the image are defined as: DEFAULT_TEST_SUITES_pn-image = "ping ssh df connman syslog xorg scp vnc date rpm dnf dmesg" Add your own test to the list of the by using the following: TEST_SUITES_append = " mytest" Run a specific list of tests as follows: TEST_SUITES = "test1 test2 test3" Remember, order is important. Be sure to place a test that is dependent on another test later in the order.

You can export tests so that they can run independently of the build system. Exporting tests is required if you want to be able to hand the test execution off to a scheduler. You can only export tests that are defined in TEST_SUITES. If your image is already built, make sure the following are set in your local.conf file: INHERIT +="testexport" TEST_TARGET_IP = "IP-address-for-the-test-target" TEST_SERVER_IP = "IP-address-for-the-test-server" You can then export the tests with the following BitBake command form: $ bitbake image -c testexport Exporting the tests places them in the Build Directory in tmp/testexport/image, which is controlled by the TEST_EXPORT_DIR variable. You can now run the tests outside of the build environment: $ cd tmp/testexport/image $ ./runexported.py testdata.json Here is a complete example that shows IP addresses and uses the core-image-sato image: INHERIT +="testexport" TEST_TARGET_IP = "192.168.7.2" TEST_SERVER_IP = "192.168.7.1" Use BitBake to export the tests: $ bitbake core-image-sato -c testexport Run the tests outside of the build environment using the following: $ cd tmp/testexport/core-image-sato $ ./runexported.py testdata.json

As mentioned previously, all new test files need to be in the proper place for the build system to find them. New tests for additional functionality outside of the core should be added to the layer that adds the functionality, in layer/lib/oeqa/runtime (as long as BBPATH is extended in the layer's layer.conf file as normal). Just remember the following: Filenames need to map directly to test (module) names. Do not use module names that collide with existing core tests. Minimally, an empty __init__.py file must exist in the runtime directory. To create a new test, start by copying an existing module (e.g. syslog.py or gcc.py are good ones to use). Test modules can use code from meta/lib/oeqa/utils, which are helper classes. Structure shell commands such that you rely on them and they return a single code for success. Be aware that sometimes you will need to parse the output. See the df.py and date.py modules for examples. You will notice that all test classes inherit oeRuntimeTest, which is found in meta/lib/oetest.py. This base class offers some helper attributes, which are described in the following sections:

Class methods are as follows: hasPackage(pkg): Returns "True" if pkg is in the installed package list of the image, which is based on the manifest file that is generated during the do_rootfs task. hasFeature(feature): Returns "True" if the feature is in IMAGE_FEATURES or DISTRO_FEATURES.

Class attributes are as follows: pscmd: Equals "ps -ef" if procps is installed in the image. Otherwise, pscmd equals "ps" (busybox). tc: The called test context, which gives access to the following attributes: d: The BitBake datastore, which allows you to use stuff such as oeRuntimeTest.tc.d.getVar("VIRTUAL-RUNTIME_init_manager"). testslist and testsrequired: Used internally. The tests do not need these. filesdir: The absolute path to meta/lib/oeqa/runtime/files, which contains helper files for tests meant for copying on the target such as small files written in C for compilation. target: The target controller object used to deploy and start an image on a particular target (e.g. QemuTarget, SimpleRemote, and Systemd-bootTarget). Tests usually use the following: ip: The target's IP address. server_ip: The host's IP address, which is usually used by the DNF test suite. run(cmd, timeout=None): The single, most used method. This command is a wrapper for: ssh root@host "cmd". The command returns a tuple: (status, output), which are what their names imply - the return code of "cmd" and whatever output it produces. The optional timeout argument represents the number of seconds the test should wait for "cmd" to return. If the argument is "None", the test uses the default instance's timeout period, which is 300 seconds. If the argument is "0", the test runs until the command returns. copy_to(localpath, remotepath): scp localpath root@ip:remotepath. copy_from(remotepath, localpath): scp root@host:remotepath localpath.

A single instance attribute exists, which is target. The target instance attribute is identical to the class attribute of the same name, which is described in the previous section. This attribute exists as both an instance and class attribute so tests can use self.target.run(cmd) in instance methods instead of oeRuntimeTest.tc.target.run(cmd).

When a test requires a package built by BitBake, it is possible to install that package. Installing the package does not require a package manager be installed in the device under test (DUT). It does, however, require an SSH connection and the target must be using the sshcontrol class. This method uses scp to copy files from the host to the target, which causes permissions and special attributes to be lost. A JSON file is used to define the packages needed by a test. This file must be in the same path as the file used to define the tests. Furthermore, the filename must map directly to the test module name with a .json extension. The JSON file must include an object with the test name as keys of an object or an array. This object (or array of objects) uses the following data: "pkg" - A mandatory string that is the name of the package to be installed. "rm" - An optional boolean, which defaults to "false", that specifies to remove the package after the test. "extract" - An optional boolean, which defaults to "false", that specifies if the package must be extracted from the package format. When set to "true", the package is not automatically installed into the DUT. Following is an example JSON file that handles test "foo" installing package "bar" and test "foobar" installing packages "foo" and "bar". Once the test is complete, the packages are removed from the DUT. { "foo": { "pkg": "bar" }, "foobar": [ { "pkg": "foo", "rm": true }, { "pkg": "bar", "rm": true } ] }

GDB allows you to examine running programs, which in turn helps you to understand and fix problems. It also allows you to perform post-mortem style analysis of program crashes. GDB is available as a package within the Yocto Project and is installed in SDK images by default. See the "Images" chapter in the Yocto Project Reference Manual for a description of these images. You can find information on GDB at . For best results, install debug (-dbg) packages for the applications you are going to debug. Doing so makes extra debug symbols available that give you more meaningful output. Sometimes, due to memory or disk space constraints, it is not possible to use GDB directly on the remote target to debug applications. These constraints arise because GDB needs to load the debugging information and the binaries of the process being debugged. Additionally, GDB needs to perform many computations to locate information such as function names, variable names and values, stack traces and so forth - even before starting the debugging process. These extra computations place more load on the target system and can alter the characteristics of the program being debugged. To help get past the previously mentioned constraints, you can use gdbserver, which runs on the remote target and does not load any debugging information from the debugged process. Instead, a GDB instance processes the debugging information that is run on a remote computer - the host GDB. The host GDB then sends control commands to gdbserver to make it stop or start the debugged program, as well as read or write memory regions of that debugged program. All the debugging information loaded and processed as well as all the heavy debugging is done by the host GDB. Offloading these processes gives the gdbserver running on the target a chance to remain small and fast. Because the host GDB is responsible for loading the debugging information and for doing the necessary processing to make actual debugging happen, you have to make sure the host can access the unstripped binaries complete with their debugging information and also be sure the target is compiled with no optimizations. The host GDB must also have local access to all the libraries used by the debugged program. Because gdbserver does not need any local debugging information, the binaries on the remote target can remain stripped. However, the binaries must also be compiled without optimization so they match the host's binaries. To remain consistent with GDB documentation and terminology, the binary being debugged on the remote target machine is referred to as the "inferior" binary. For documentation on GDB see the GDB site. The following steps show you how to debug using the GNU project debugger. Configure your build system to construct the companion debug filesystem: In your local.conf file, set the following: IMAGE_GEN_DEBUGFS = "1" IMAGE_FSTYPES_DEBUGFS = "tar.bz2" These options cause the OpenEmbedded build system to generate a special companion filesystem fragment, which contains the matching source and debug symbols to your deployable filesystem. The build system does this by looking at what is in the deployed filesystem, and pulling the corresponding -dbg packages. The companion debug filesystem is not a complete filesystem, but only contains the debug fragments. This filesystem must be combined with the full filesystem for debugging. Subsequent steps in this procedure show how to combine the partial filesystem with the full filesystem. Configure the system to include gdbserver in the target filesystem: Make the following addition in either your local.conf file or in an image recipe: IMAGE_INSTALL_append = “ gdbserver" The change makes sure the gdbserver package is included. Build the environment: Use the following command to construct the image and the companion Debug Filesystem: $ bitbake image Build the cross GDB component and make it available for debugging. Build the SDK that matches the image. Building the SDK is best for a production build that can be used later for debugging, especially during long term maintenance: $ bitbake -c populate_sdk image Alternatively, you can build the minimal toolchain components that match the target. Doing so creates a smaller than typical SDK and only contains a minimal set of components with which to build simple test applications, as well as run the debugger: $ bitbake meta-toolchain A final method is to build Gdb itself within the build system: $ bitbake gdb-cross-architecture Doing so produces a temporary copy of cross-gdb you can use for debugging during development. While this is the quickest approach, the two previous methods in this step are better when considering long-term maintenance strategies. If you run bitbake gdb-cross, the OpenEmbedded build system suggests the actual image (e.g. gdb-cross-i586). The suggestion is usually the actual name you want to use. Set up the debugfs Run the following commands to set up the debugfs: $ mkdir debugfs $ cd debugfs $ tar xvfj build-dir/tmp-glibc/deploy/images/machine/image.rootfs.tar.bz2 $ tar xvfj build-dir/tmp-glibc/deploy/images/machine/image-dbg.rootfs.tar.bz2 Set up GDB Install the SDK (if you built one) and then source the correct environment file. Sourcing the environment file puts the SDK in your PATH environment variable. If you are using the build system, Gdb is located in build-dir/tmp/sysroots/host/usr/bin/architecture/architecture-gdb Boot the target: For information on how to run QEMU, see the QEMU Documentation. Be sure to verify that your host can access the target via TCP. Debug a program: Debugging a program involves running gdbserver on the target and then running Gdb on the host. The example in this step debugs gzip: root@qemux86:~# gdbserver localhost:1234 /bin/gzip —help For additional gdbserver options, see the GDB Server Documentation. After running gdbserver on the target, you need to run Gdb on the host and configure it and connect to the target. Use these commands: $ cd directory-holding-the-debugfs-directory $ arch-gdb (gdb) set sysroot debugfs (gdb) set substitute-path /usr/src/debug debugfs/usr/src/debug (gdb) target remote IP-of-target:1234 At this point, everything should automatically load (i.e. matching binaries, symbols and headers). The Gdb set commands in the previous example can be placed into the users ~/.gdbinit file. Upon starting, Gdb automatically runs whatever commands are in that file. Deploying without a full image rebuild: In many cases, during development you want a quick method to deploy a new binary to the target and debug it, without waiting for a full image build. One approach to solving this situation is to just build the component you want to debug. Once you have built the component, copy the executable directly to both the target and the host debugfs. If the binary is processed through the debug splitting in OpenEmbedded, you should also copy the debug items (i.e. .debug contents and corresponding /usr/src/debug files) from the work directory. Here is an example: $ bitbake bash $ bitbake -c devshell bash $ cd .. $ scp packages-split/bash/bin/bash target:/bin/bash $ cp -a packages-split/bash-dbg/* path/debugfs

The previous section addressed using GDB remotely for debugging purposes, which is the most usual case due to the inherent hardware limitations on many embedded devices. However, debugging in the target hardware itself is also possible with more powerful devices. This section describes what you need to do in order to support using GDB to debug on the target hardware. To support this kind of debugging, you need do the following: Ensure that GDB is on the target. You can do this by adding "gdb" to IMAGE_INSTALL: IMAGE_INSTALL_append = " gdb" Alternatively, you can add "tools-debug" to IMAGE_FEATURES: IMAGE_FEATURES_append = " tools-debug" Ensure that debug symbols are present. You can make sure these symbols are present by installing -dbg: IMAGE_INSTALL_append = " packagename-dbg" Alternatively, you can do the following to include all the debug symbols: IMAGE_FEATURES_append = " dbg-pkgs" To improve the debug information accuracy, you can reduce the level of optimization used by the compiler. For example, when adding the following line to your local.conf file, you will reduce optimization from FULL_OPTIMIZATION of "-O2" to DEBUG_OPTIMIZATION of "-O -fno-omit-frame-pointer": DEBUG_BUILD = "1" Consider that this will reduce the application's performance and is recommended only for debugging purposes.

A parallel make race occurs when the build consists of several parts that are run simultaneously and a situation occurs when the output or result of one part is not ready for use with a different part of the build that depends on that output. Parallel make races are annoying and can sometimes be difficult to reproduce and fix. However, some simple tips and tricks exist that can help you debug and fix them. This section presents a real-world example of an error encountered on the Yocto Project autobuilder and the process used to fix it. If you cannot properly fix a make race condition, you can work around it by clearing either the PARALLEL_MAKE or PARALLEL_MAKEINST variables.

For this example, assume that you are building an image that depends on the "neard" package. And, during the build, BitBake runs into problems and creates the following output. This example log file has longer lines artificially broken to make the listing easier to read. If you examine the output or the log file, you see the failure during make: | DEBUG: SITE files ['endian-little', 'bit-32', 'ix86-common', 'common-linux', 'common-glibc', 'i586-linux', 'common'] | DEBUG: Executing shell function do_compile | NOTE: make -j 16 | make --no-print-directory all-am | /bin/mkdir -p include/near | /bin/mkdir -p include/near | /bin/mkdir -p include/near | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/types.h include/near/types.h | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/log.h include/near/log.h | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/plugin.h include/near/plugin.h | /bin/mkdir -p include/near | /bin/mkdir -p include/near | /bin/mkdir -p include/near | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/tag.h include/near/tag.h | /bin/mkdir -p include/near | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/adapter.h include/near/adapter.h | /bin/mkdir -p include/near | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/ndef.h include/near/ndef.h | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/tlv.h include/near/tlv.h | /bin/mkdir -p include/near | /bin/mkdir -p include/near | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/setting.h include/near/setting.h | /bin/mkdir -p include/near | /bin/mkdir -p include/near | /bin/mkdir -p include/near | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/device.h include/near/device.h | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/nfc_copy.h include/near/nfc_copy.h | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/snep.h include/near/snep.h | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/version.h include/near/version.h | ln -s /home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/work/i586-poky-linux/neard/ 0.14-r0/neard-0.14/include/dbus.h include/near/dbus.h | ./src/genbuiltin nfctype1 nfctype2 nfctype3 nfctype4 p2p > src/builtin.h | i586-poky-linux-gcc -m32 -march=i586 --sysroot=/home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/ build/build/tmp/sysroots/qemux86 -DHAVE_CONFIG_H -I. -I./include -I./src -I./gdbus -I/home/pokybuild/ yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/sysroots/qemux86/usr/include/glib-2.0 -I/home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/tmp/sysroots/qemux86/usr/ lib/glib-2.0/include -I/home/pokybuild/yocto-autobuilder/yocto-slave/nightly-x86/build/build/ tmp/sysroots/qemux86/usr/include/dbus-1.0 -I/home/pokybuild/yocto-autobuilder/yocto-slave/ nightly-x86/build/build/tmp/sysroots/qemux86/usr/lib/dbus-1.0/include -I/home/pokybuild/yocto-autobuilder/ yocto-slave/nightly-x86/build/build/tmp/sysroots/qemux86/usr/include/libnl3 -DNEAR_PLUGIN_BUILTIN -DPLUGINDIR=\""/usr/lib/near/plugins"\" -DCONFIGDIR=\""/etc/neard\"" -O2 -pipe -g -feliminate-unused-debug-types -c -o tools/snep-send.o tools/snep-send.c | In file included from tools/snep-send.c:16:0: | tools/../src/near.h:41:23: fatal error: near/dbus.h: No such file or directory | #include <near/dbus.h> | ^ | compilation terminated. | make[1]: *** [tools/snep-send.o] Error 1 | make[1]: *** Waiting for unfinished jobs.... | make: *** [all] Error 2 | ERROR: oe_runmake failed

Because race conditions are intermittent, they do not manifest themselves every time you do the build. In fact, most times the build will complete without problems even though the potential race condition exists. Thus, once the error surfaces, you need a way to reproduce it. In this example, compiling the "neard" package is causing the problem. So the first thing to do is build "neard" locally. Before you start the build, set the PARALLEL_MAKE variable in your local.conf file to a high number (e.g. "-j 20"). Using a high value for PARALLEL_MAKE increases the chances of the race condition showing up: $ bitbake neard Once the local build for "neard" completes, start a devshell build: $ bitbake neard -c devshell For information on how to use a devshell, see the "Using a Development Shell" section. In the devshell, do the following: $ make clean $ make tools/snep-send.o The devshell commands cause the failure to clearly be visible. In this case, a missing dependency exists for the "neard" Makefile target. Here is some abbreviated, sample output with the missing dependency clearly visible at the end: i586-poky-linux-gcc -m32 -march=i586 --sysroot=/home/scott-lenovo/...... . . . tools/snep-send.c In file included from tools/snep-send.c:16:0: tools/../src/near.h:41:23: fatal error: near/dbus.h: No such file or directory #include <near/dbus.h> ^ compilation terminated. make: *** [tools/snep-send.o] Error 1 $

Because there is a missing dependency for the Makefile target, you need to patch the Makefile.am file, which is generated from Makefile.in. You can use Quilt to create the patch: $ quilt new parallelmake.patch Patch patches/parallelmake.patch is now on top $ quilt add Makefile.am File Makefile.am added to patch patches/parallelmake.patch For more information on using Quilt, see the "Using Quilt in Your Workflow" section. At this point you need to make the edits to Makefile.am to add the missing dependency. For our example, you have to add the following line to the file: tools/snep-send.$(OBJEXT): include/near/dbus.h Once you have edited the file, use the refresh command to create the patch: $ quilt refresh Refreshed patch patches/parallelmake.patch Once the patch file exists, you need to add it back to the originating recipe folder. Here is an example assuming a top-level Source Directory named poky: $ cp patches/parallelmake.patch poky/meta/recipes-connectivity/neard/neard The final thing you need to do to implement the fix in the build is to update the "neard" recipe (i.e. neard-0.14.bb) so that the SRC_URI statement includes the patch file. The recipe file is in the folder above the patch. Here is what the edited SRC_URI statement would look like: SRC_URI = "${KERNELORG_MIRROR}/linux/network/nfc/${BPN}-${PV}.tar.xz \ file://neard.in \ file://neard.service.in \ file://parallelmake.patch \ " With the patch complete and moved to the correct folder and the SRC_URI statement updated, you can exit the devshell: $ exit

With everything in place, you can get back to trying the build again locally: $ bitbake neard This build should succeed. Now you can open up a devshell again and repeat the clean and make operations as follows: $ bitbake neard -c devshell $ make clean $ make tools/snep-send.o The build should work without issue. As with all solved problems, if they originated upstream, you need to submit the fix for the recipe in OE-Core and upstream so that the problem is taken care of at its source. See the "Submitting a Change to the Yocto Project" section for more information.

One of the concerns for a development organization using open source software is how to maintain compliance with various open source licensing during the lifecycle of the product. While this section does not provide legal advice or comprehensively cover all scenarios, it does present methods that you can use to assist you in meeting the compliance requirements during a software release. With hundreds of different open source licenses that the Yocto Project tracks, it is difficult to know the requirements of each and every license. However, the requirements of the major FLOSS licenses can begin to be covered by assuming that three main areas of concern exist: Source code must be provided. License text for the software must be provided. Compilation scripts and modifications to the source code must be provided. There are other requirements beyond the scope of these three and the methods described in this section (e.g. the mechanism through which source code is distributed). As different organizations have different methods of complying with open source licensing, this section is not meant to imply that there is only one single way to meet your compliance obligations, but rather to describe one method of achieving compliance. The remainder of this section describes methods supported to meet the previously mentioned three requirements. Once you take steps to meet these requirements, and prior to releasing images, sources, and the build system, you should audit all artifacts to ensure completeness. The Yocto Project generates a license manifest during image creation that is located in ${DEPLOY_DIR}/licenses/image_name-datestamp to assist with any audits.

Compliance activities should begin before you generate the final image. The first thing you should look at is the requirement that tops the list for most compliance groups - providing the source. The Yocto Project has a few ways of meeting this requirement. One of the easiest ways to meet this requirement is to provide the entire DL_DIR used by the build. This method, however, has a few issues. The most obvious is the size of the directory since it includes all sources used in the build and not just the source used in the released image. It will include toolchain source, and other artifacts, which you would not generally release. However, the more serious issue for most companies is accidental release of proprietary software. The Yocto Project provides an archiver class to help avoid some of these concerns. Before you employ DL_DIR or the archiver class, you need to decide how you choose to provide source. The source archiver class can generate tarballs and SRPMs and can create them with various levels of compliance in mind. One way of doing this (but certainly not the only way) is to release just the source as a tarball. You can do this by adding the following to the local.conf file found in the Build Directory: INHERIT += "archiver" ARCHIVER_MODE[src] = "original" During the creation of your image, the source from all recipes that deploy packages to the image is placed within subdirectories of DEPLOY_DIR/sources based on the LICENSE for each recipe. Releasing the entire directory enables you to comply with requirements concerning providing the unmodified source. It is important to note that the size of the directory can get large. A way to help mitigate the size issue is to only release tarballs for licenses that require the release of source. Let us assume you are only concerned with GPL code as identified by running the following script: # Script to archive a subset of packages matching specific license(s) # Source and license files are copied into sub folders of package folder # Must be run from build folder #!/bin/bash src_release_dir="source-release" mkdir -p $src_release_dir for a in tmp/deploy/sources/*; do for d in $a/*; do # Get package name from path p=`basename $d` p=${p%-*} p=${p%-*} # Only archive GPL packages (update *GPL* regex for your license check) numfiles=`ls tmp/deploy/licenses/$p/*GPL* 2> /dev/null | wc -l` if [ $numfiles -gt 1 ]; then echo Archiving $p mkdir -p $src_release_dir/$p/source cp $d/* $src_release_dir/$p/source 2> /dev/null mkdir -p $src_release_dir/$p/license cp tmp/deploy/licenses/$p/* $src_release_dir/$p/license 2> /dev/null fi done done At this point, you could create a tarball from the gpl_source_release directory and provide that to the end user. This method would be a step toward achieving compliance with section 3a of GPLv2 and with section 6 of GPLv3.

One requirement that is often overlooked is inclusion of license text. This requirement also needs to be dealt with prior to generating the final image. Some licenses require the license text to accompany the binary. You can achieve this by adding the following to your local.conf file: COPY_LIC_MANIFEST = "1" COPY_LIC_DIRS = "1" LICENSE_CREATE_PACKAGE = "1" Adding these statements to the configuration file ensures that the licenses collected during package generation are included on your image. Setting all three variables to "1" results in the image having two copies of the same license file. One copy resides in /usr/share/common-licenses and the other resides in /usr/share/license. The reason for this behavior is because COPY_LIC_DIRS and COPY_LIC_MANIFEST add a copy of the license when the image is built but do not offer a path for adding licenses for newly installed packages to an image. LICENSE_CREATE_PACKAGE adds a separate package and an upgrade path for adding licenses to an image. As the source archiver class has already archived the original unmodified source that contains the license files, you would have already met the requirements for inclusion of the license information with source as defined by the GPL and other open source licenses.

At this point, we have addressed all we need to prior to generating the image. The next two requirements are addressed during the final packaging of the release. By releasing the version of the OpenEmbedded build system and the layers used during the build, you will be providing both compilation scripts and the source code modifications in one step. If the deployment team has a BSP layer and a distro layer, and those those layers are used to patch, compile, package, or modify (in any way) any open source software included in your released images, you might be required to release those layers under section 3 of GPLv2 or section 1 of GPLv3. One way of doing that is with a clean checkout of the version of the Yocto Project and layers used during your build. Here is an example: # We built using the &DISTRO_NAME_NO_CAP; branch of the poky repo $ git clone -b &DISTRO_NAME_NO_CAP; git://git.yoctoproject.org/poky $ cd poky # We built using the release_branch for our layers $ git clone -b release_branch git://git.mycompany.com/meta-my-bsp-layer $ git clone -b release_branch git://git.mycompany.com/meta-my-software-layer # clean up the .git repos $ find . -name ".git" -type d -exec rm -rf {} \; One thing a development organization might want to consider for end-user convenience is to modify meta-poky/conf/bblayers.conf.sample to ensure that when the end user utilizes the released build system to build an image, the development organization's layers are included in the bblayers.conf file automatically: # LAYER_CONF_VERSION is increased each time build/conf/bblayers.conf # changes incompatibly LCONF_VERSION = "6" BBPATH = "${TOPDIR}" BBFILES ?= "" BBLAYERS ?= " \ ##OEROOT##/meta \ ##OEROOT##/meta-poky \ ##OEROOT##/meta-yocto-bsp \ ##OEROOT##/meta-mylayer \ " Creating and providing an archive of the Metadata layers (recipes, configuration files, and so forth) enables you to meet your requirements to include the scripts to control compilation as well as any modifications to the original source.

The error reporting tool allows you to submit errors encountered during builds to a central database. Outside of the build environment, you can use a web interface to browse errors, view statistics, and query for errors. The tool works using a client-server system where the client portion is integrated with the installed Yocto Project Source Directory (e.g. poky). The server receives the information collected and saves it in a database. A live instance of the error reporting server exists at . This server exists so that when you want to get help with build failures, you can submit all of the information on the failure easily and then point to the URL in your bug report or send an email to the mailing list. If you send error reports to this server, the reports become publicly visible.

By default, the error reporting tool is disabled. You can enable it by inheriting the report-error class by adding the following statement to the end of your local.conf file in your Build Directory. INHERIT += "report-error" By default, the error reporting feature stores information in ${LOG_DIR}/error-report. However, you can specify a directory to use by adding the following to your local.conf file: ERR_REPORT_DIR = "path" Enabling error reporting causes the build process to collect the errors and store them in a file as previously described. When the build system encounters an error, it includes a command as part of the console output. You can run the command to send the error file to the server. For example, the following command sends the errors to an upstream server: $ send-error-report /home/brandusa/project/poky/build/tmp/log/error-report/error_report_201403141617.txt In the previous example, the errors are sent to a public database available at , which is used by the entire community. If you specify a particular server, you can send the errors to a different database. Use the following command for more information on available options: $ send-error-report --help When sending the error file, you are prompted to review the data being sent as well as to provide a name and optional email address. Once you satisfy these prompts, the command returns a link from the server that corresponds to your entry in the database. For example, here is a typical link: http://errors.yoctoproject.org/Errors/Details/9522/ Following the link takes you to a web interface where you can browse, query the errors, and view statistics.

To disable the error reporting feature, simply remove or comment out the following statement from the end of your local.conf file in your Build Directory. INHERIT += "report-error"

If you want to set up your own error reporting server, you can obtain the code from the Git repository at . Instructions on how to set it up are in the README document.