generic-poky/documentation/dev-manual/dev-manual-kernel-appendix.xml

1133 lines
57 KiB
XML

<!DOCTYPE appendix PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN"
"http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd">
<appendix id='dev-manual-kernel-appendix'>
<title>Kernel Modification Example</title>
<para>
Kernel modification involves changing or adding configurations to an existing kernel,
changing or adding recipes to the kernel that are needed to support specific hardware features,
or even altering the source code itself.
This appendix presents simple examples that modify the kernel source code,
change the kernel configuration, and add a kernel source recipe.
<!-- [WRITER'S NOTE: I might want to work in information about applying a local
change to a kernel layer and also pushing a change upstream into the tree]
<orderedlist>
<listitem><para>Iteratively determine and set kernel configurations and make
kernel recipe changes.</para></listitem>
<listitem><para>Apply your configuration changes to your local kernel layer.
</para></listitem>
<listitem><para>Push your configuration and recipe changes upstream into the
Yocto Project source repositories to make them available to the community.
</para></listitem>
</orderedlist> -->
</para>
<section id='modifying-the-kernel-source-code'>
<title>Modifying the Kernel Source Code</title>
<para>
This example adds some simple QEMU emulator console output at boot time by
adding <filename>printk</filename> statements to the kernel's
<filename>calibrate.c</filename> source code file.
Booting the modified image causes the added messages to appear on the emulator's
console.
</para>
<section id='understanding-the-files-you-need'>
<title>Understanding the Files You Need</title>
<para>
Before you modify the kernel, you need to know what Git repositories and file
structures you need.
Briefly, you need the following:
<itemizedlist>
<listitem><para>A local Yocto Project files Git repository</para></listitem>
<listitem><para>The <filename>poky-extras</filename> Git repository placed
within the local Yocto Project files Git repository</para></listitem>
<listitem><para>A bare clone of the Linux Yocto kernel upstream Git
repository to which you want to push your modifications.
</para></listitem>
<listitem><para>A copy of that bare clone in which you make your source
modifcations</para></listitem>
</itemizedlist>
</para>
<para>
The following figure summarizes these four areas.
Within each rectangular that represents a data structure, a
host development directory pathname appears at the
lower left-hand corner of the box.
These pathnames are the locations used in this example.
The figure also provides key statements and commands used during the kernel
modification process:
</para>
<para>
<imagedata fileref="figures/kernel-example-repos-edison.png" width="7in" depth="5in"
align="center" scale="100" />
</para>
<para>
Here is a brief description of the four areas:
<itemizedlist>
<listitem><para><emphasis>Local Yocto Project Files Git Repository:</emphasis>
This area contains all the metadata that supports building images in the
Yocto Project build environment - the local Yocto Project files.
In this example, the local Yocto Project files Git repository also
contains the build directory, which contains the configuration directory
that lets you control the build.
In this example, the repository also contains the
<filename>poky-extras</filename> Git repository.</para>
<para>See the bulleted item
"<link linkend='local-yp-release'>Yocto Project Release</link>"
for information on how to get these files.</para></listitem>
<listitem><para><emphasis><filename>poky-extras</filename> Git Repository:</emphasis>
This area contains the <filename>meta-kernel-dev</filename> layer,
which is where you make changes that append the kernel build recipes.
You edit <filename>.bbappend</filename> files to locate your
local kernel source files and to identify the kernel being built.
This Git repository is a gathering place for extensions to the Linux Yocto
(or really any) kernel recipes that faciliate the creation and development
of kernel features, BSPs or configurations.</para>
<para>See the bulleted item
"<link linkend='poky-extras-repo'>The
<filename>poky-extras</filename> Git Repository</link>"
for information on how to get these files.</para></listitem>
<listitem><para><emphasis>Bare Clone of the Linux Yocto kernel:</emphasis>
This bare Git repository tracks the upstream Git repository of the Linux
Yocto kernel source code you are changing.
When you modify the kernel you must work through a bare clone.
All source code changes you make to the kernel must be committed and
pushed to the bare clone using Git commands.
As mentioned, the <filename>.bbappend</filename> file in the
<filename>poky-extras</filename> repository points to the bare clone
so that the build process can locate the locally changed source files.</para>
<para>See the bulleted item
"<link linkend='local-kernel-files'>Linux Yocto Kernel</link>"
for information on how to set up the bare clone.
</para></listitem>
<listitem><para><emphasis>Copy of the Linux Yocto Kernel Bare Clone:</emphasis>
This Git repository contains the actual source files that you modify.
Any changes you make to files in this location need to ultimately be pushed
to the bare clone using the <filename>git push</filename> command.</para>
<para>See the bulleted item
"<link linkend='local-kernel-files'>Linux Yocto Kernel</link>"
for information on how to set up the bare clone.
<note>Typically, Git workflows follow a scheme where changes made to a local area
are pulled into a Git repository.
However, because the <filename>git pull</filename> command does not work
with bare clones, this workflow pushes changes to the
repository even though you could use other more complicated methods to
get changes into the bare clone.</note>
</para></listitem>
</itemizedlist>
</para>
</section>
<section id='setting-up-the-local-yocto-project-files-git-repository'>
<title>Setting Up the Local Yocto Project Files Git Repository</title>
<para>
You can get the local Yocto Project files through tarball extraction or by
cloning the <filename>poky</filename> Git repository.
This example uses <filename>poky</filename> as the root directory of the
local Yocto Project files Git repository.
See the bulleted item
"<link linkend='local-yp-release'>Yocto Project Release</link>"
for information on how to get these files.
</para>
<para>
Once you have the repository set up,
you have many development branches from which you can work.
From inside the repository you can see the branch names and the tag names used
in the Git repository using either of the following two commands:
<literallayout class='monospaced'>
$ cd poky
$ git branch -a
$ git tag -l
</literallayout>
This example uses the Yocto Project 1.1.1 Release code named "edison",
which maps to the <filename>edison</filename> branch in the repository.
The following commands create and checkout the local <filename>edison</filename>
branch:
<literallayout class='monospaced'>
$ git checkout -b edison origin/edison
Branch edison set up to track remote branch edison from origin.
Switched to a new branch 'edison'
</literallayout>
</para>
</section>
<section id='setting-up-the-poky-extras-git-repository'>
<title>Setting Up the poky-extras Git Repository</title>
<para>
This example places the <filename>poky-extras</filename> Git repository inside
of <filename>poky</filename>.
See the bulleted item
"<link linkend='poky-extras-repo'>The
<filename>poky-extras</filename> Git Repository</link>"
for information on how to get the <filename>poky-extras</filename> repository.
</para>
<para>
Once you have the repository set up,
you have many development branches from which you can work.
From inside the repository you can see the branch names and the tag names used
in the Git repository using either of the following two commands:
<literallayout class='monospaced'>
$ cd poky
$ git branch -a
$ git tag -l
</literallayout>
This example uses the Yocto Project 1.1.1 Release code named "edison",
which maps to the <filename>edison</filename> branch in the repository.
The following commands create and checkout the local <filename>edison</filename>
branch:
<literallayout class='monospaced'>
$ git checkout -b edison origin/edison
Branch edison set up to track remote branch edison from origin.
Switched to a new branch 'edison'
</literallayout>
</para>
</section>
<section id='setting-up-the-bare-clone-and-its-copy'>
<title>Setting Up the Bare Clone and its Copy</title>
<para>
This example modifies the <filename>linux-yocto-3.0-1.1.x</filename> kernel.
Thus, you need to create a bare clone of that kernel and then make a copy of the
bare clone.
See the bulleted item
"<link linkend='local-kernel-files'>Linux Yocto Kernel</link>"
for information on how to do that.
</para>
<para>
The bare clone exists for the kernel build tools and simply as the receiving end
of <filename>git push</filename>
commands after you make edits and commits inside the copy of the clone.
The copy (<filename>my-linux-yocto-3.0-1.1.x-work</filename> in this example) has to have
a local branch created and checked out for your work.
This example uses <filename>common-pc-base</filename> as the local branch.
The following commands create and checkout the branch:
<literallayout class='monospaced'>
$ cd ~/my-linux-yocto-3.0-1.1.x-work
$ git checkout -b common-pc-base origin/yocto/standard/common-pc/base
Checking out files: 100% (7289/7289), done.
Branch common-pc-base set up to track remote branch
yocto/standard/common-pc/base from origin.
Switched to a new branch 'common-pc-base'
</literallayout>
</para>
</section>
<section id='building-and-booting-the-default-qemu-kernel-image'>
<title>Building and Booting the Default QEMU Kernel Image</title>
<para>
Before we make changes to the kernel source files, this example first builds the
default image and then boots it inside the QEMU emulator.
<note>
Because a full build can take hours, you should check two variables in the
<filename>build</filename> directory that is created after you source the
<filename>oe-init-build-env</filename> script.
You can find these variables
<filename>BB_NUMBER_THREADS</filename> and <filename>PARALLEL_MAKE</filename>
in the <filename>build/conf</filename> directory in the
<filename>local.conf</filename> configuration file.
By default, these variables are commented out.
If your host development system supports multi-core and multi-thread capabilities,
you can uncomment these statements and set the variables to significantly shorten
the full build time.
As a guideline, set <filename>BB_NUMBER_THREADS</filename> to twice the number
of cores your machine supports and set <filename>PARALLEL_MAKE</filename> to one and
a half times the number of cores your machine supports.
</note>
The following two commands <filename>source</filename> the build environment setup script
and build the default <filename>qemux86</filename> image.
If necessary, the script creates the build directory:
<literallayout class='monospaced'>
$ cd ~/poky
$ source oe-init-build-env
### Shell environment set up for builds. ###
You can now run 'bitbake &lt;target&gt;'
Common targets are:
core-image-minimal
core-image-sato
meta-toolchain
meta-toolchain-sdk
adt-installer
meta-ide-support
You can also run generated qemu images with a command like 'runqemu qemux86'
</literallayout>
</para>
<para>
The following <filename>bitbake</filename> command starts the build:
<literallayout class='monospaced'>
$ bitbake -k core-image-minimal
</literallayout>
<note>Be sure to check the settings in the <filename>local.conf</filename>
before starting the build.</note>
</para>
<para>
After the build completes, you can start the QEMU emulator using the resulting image
<filename>qemux86</filename> as follows:
<literallayout class='monospaced'>
$ runqemu qemux86
</literallayout>
</para>
<para>
As the image boots in the emulator, console message and status output appears
across the terminal window.
Because the output scrolls by quickly, it is difficult to read.
To examine the output, you log into the system using the
login <filename>root</filename> with no password.
Once you are logged in, issue the following command to scroll through the
console output:
<literallayout class='monospaced'>
# dmesg | less
</literallayout>
</para>
<para>
Take note of the output as you will want to look for your inserted print command output
later in the example.
</para>
</section>
<section id='changing-the-source-code-and-pushing-it-to-the-bare-clone'>
<title>Changing the Source Code and Pushing it to the Bare Clone</title>
<para>
The file you change in this example is named <filename>calibrate.c</filename>
and is located in the <filename>my-linux-yocto-3.0-1.1.x-work</filename> Git repository
(the copy of the bare clone) in <filename>init</filename>.
This example simply inserts several <filename>printk</filename> statements
at the beginning of the <filename>calibrate_delay</filename> function.
</para>
<para>
Here is the unaltered code at the start of this function:
<literallayout class='monospaced'>
void __cpuinit calibrate_delay(void)
{
unsigned long lpj;
static bool printed;
if (preset_lpj) {
.
.
.
</literallayout>
</para>
<para>
Here is the altered code showing five new <filename>printk</filename> statements
just after initializing <filename>lps_precision</filename>:
<literallayout class='monospaced'>
void __cpuinit calibrate_delay(void)
{
unsigned long lpj;
static bool printed;
printk("*************************************\n");
printk("* *\n");
printk("* HELLO YOCTO KERNEL *\n");
printk("* *\n");
printk("*************************************\n");
if (preset_lpj) {
.
.
.
</literallayout>
</para>
<para>
After making and saving your changes, you need to stage them for the push.
The following Git commands are one method of staging and committing your changes:
<literallayout class='monospaced'>
$ git add calibrate.c
$ git commit --signoff
</literallayout>
</para>
<para>
Once the source code has been modified, you need to use Git to push the changes to
the bare clone.
If you do not push the changes, then the Yocto Project build system will not pick
up the changed source files.
</para>
<para>
The following command pushes the changes to the bare clone:
<literallayout class='monospaced'>
$ git push origin common-pc-base:yocto/standard/common-pc/base
</literallayout>
</para>
</section>
<section id='changing-build-parameters-for-your-build'>
<title>Changing Build Parameters for Your Build</title>
<para>
At this point, the source has been changed and pushed.
The example now defines some variables used by the Yocto Project build system
to locate your kernel source.
You essentially need to identify where to find the kernel recipe and the changed source code.
You also need to be sure some basic configurations are in place that identify the
type of machine you are building and to help speed up the build should your host support
multiple-core and thread capabilities.
</para>
<para>
Do the following to make sure the build parameters are set up for the example.
Once you set up these build parameters, they do not have to change unless you
change the target architecture of the machine you are building or you move
the bare clone, copy of the clone, or the <filename>poky-extras</filename> repository:
<itemizedlist>
<listitem><para><emphasis>Build for the Correct Target Architecture:</emphasis> The
<filename>local.conf</filename> file in the build directory defines the build's
target architecture.
By default, <filename>MACHINE</filename> is set to
<filename>qemux86</filename>, which specifies a 32-bit
<trademark class='registered'>Intel</trademark> Architecture
target machine suitable for the QEMU emulator.
In this example, <filename>MACHINE</filename> is correctly configured.
</para></listitem>
<listitem><para><emphasis>Optimize Build Time:</emphasis> Also in the
<filename>local.conf</filename> file are two variables that can speed your
build time if your host supports multi-core and multi-thread capabilities:
<filename>BB_NUMBER_THREADS</filename> and <filename>PARALLEL_MAKE</filename>.
If the host system has multiple cores then you can optimize build time
by setting <filename>BB_NUMBER_THREADS</filename> to twice the number of
cores and setting <filename>PARALLEL_MAKE</filename> to one and a half times the
number of cores.</para></listitem>
<listitem><para><emphasis>Identify Your <filename>meta-kernel-dev</filename>
Layer:</emphasis> The <filename>BBLAYERS</filename> variable in the
<filename>bblayers.conf</filename> file found in the
<filename>poky/build/conf</filename> directory needs to have the path to your local
<filename>meta-kernel-dev</filename> layer.
By default, the <filename>BBLAYERS</filename> variable contains paths to
<filename>meta</filename> and <filename>meta-yocto</filename> in the
<filename>poky</filename> Git repository.
Add the path to your <filename>meta-kernel-dev</filename> location.
Be sure to substitute your user information in the statement.
Here is an example:
<literallayout class='monospaced'>
BBLAYERS = " \
/home/scottrif/poky/meta \
/home/scottrif/poky/meta-yocto \
/home/scottrif/poky/poky-extras/meta-kernel-dev \
"
</literallayout></para></listitem>
<listitem><para><emphasis>Identify Your Source Files:</emphasis> In the
<filename>linux-yocto_3.0.bbappend</filename> file located in the
<filename>poky-extras/meta-kernel-dev/recipes-kernel/linux</filename>
directory, you need to identify the location of the
local source code, which in this example is the bare clone named
<filename>linux-yocto-3.0-1.1.x.git</filename>.
To do this, set the <filename>KSRC_linux_yocto</filename> variable to point to your
local <filename>linux-yocto-3.0-1.1.x.git</filename> Git repository by adding the
following statement.
Be sure to substitute your user information in the statement:
<literallayout class='monospaced'>
KSRC_linux_yocto ?= /home/scottrif/linux-yocto-3.0-1.1.x.git
</literallayout></para></listitem>
<listitem><para><emphasis>Specify the Kernel Machine:</emphasis> Also in the
<filename>linux-yocto_3.0.bbappend</filename> file, you need to specify
the kernel machine with the following statement:
<literallayout class='monospaced'>
KMACHINE_qemux86 = "yocto/standard/common-pc/base"
</literallayout></para></listitem>
</itemizedlist>
</para>
<note>
Before attempting to build the modified kernel, there is one more set of changes you
need to make in the <filename>meta-kernel-dev</filename> layer.
Because all the kernel <filename>.bbappend</filename> files are parsed during the
build process regardless of whether you are using them or not, you should either
comment out the <filename>COMPATIBLE_MACHINE</filename> statements in all
unused <filename>.bbappend</filename> files.
Alternatively, you can simply remove all the files
except the one your are using for the build
(i.e. <filename>linux-yocto_3.0.bbappend</filename> in this example).
</note>
</section>
<section id='building-and-booting-the-modified-qemu-kernel-image'>
<title>Building and Booting the Modified QEMU Kernel Image</title>
<para>
Next, you need to build the modified image.
Do the following:
<orderedlist>
<listitem><para>Your environment should be set up since you previously sourced
the <filename>oe-init-build-env</filename> script.
If it isn't, source the script again from <filename>poky</filename>.
</para></listitem>
<listitem><para>Be sure old images are cleaned out by running the
<filename>cleanall</filename> BitBake task as follows:
<literallayout class='monospaced'>
$ cd ~/poky
$ bitbake -c cleanall linux-yocto
</literallayout></para>
<para><note>Never remove any files by hand from the <filename>tmp/deploy</filename>
directory insided the local Yocto Project files build directory.
Always use the BitBake <filename>cleanall</filename> task to clear
out previous builds.</note></para></listitem>
<listitem><para>Build the kernel image using this command:
<literallayout class='monospaced'>
$ bitbake -k core-image-minimal
</literallayout></para></listitem>
<listitem><para>Finally, boot the modified image in the QEMU emulator
using this command:
<literallayout class='monospaced'>
$ runqemu qemux86
</literallayout></para></listitem>
</orderedlist>
</para>
<para>
Log into the machine using <filename>root</filename> with no password and then
use the following shell command to scroll through the console's boot output.
<literallayout class='monospaced'>
# dmesg | less
</literallayout>
</para>
<para>
You should see the results of your <filename>printk</filename> statements
as part of the output.
</para>
</section>
</section>
<section id='changing-the-kernel-configuration'>
<title>Changing the Kernel Configuration</title>
<para>
This example changes the default behavior, which is "off", of the Symmetric
Multi-processing Support (<filename>CONFIG_SMP</filename>) to "on".
It is a simple example that demonstrates how to reconfigure the kernel.
</para>
<section id='getting-set-up-to-run-this-example'>
<title>Getting Set Up to Run this Example</title>
<para>
If you took the time to work through the example that modifies the kernel source code
in "<link linkend='modifying-the-kernel-source-code'>Modifying the Kernel Source
Code</link>" you should already have the Yocto Project files set up on your
host machine.
If this is the case, go to then next section titled
"<link linkend='examining-the-default-config-smp-behavior'>Examining the Default
<filename>CONFIG_SMP</filename> Behavior</link>" and continue with the
example.
</para>
<para>
If you don't have the Yocto Project files established on your system,
you can get them through tarball extraction or by
cloning the <filename>poky</filename> Git repository.
This example uses <filename>poky</filename> as the root directory of the
local Yocto Project files Git repository.
See the bulleted item
"<link linkend='local-yp-release'>Yocto Project Release</link>"
for information on how to get these files.
</para>
<para>
Once you have the repository set up,
you have many development branches from which you can work.
From inside the repository you can see the branch names and the tag names used
in the Git repository using either of the following two commands:
<literallayout class='monospaced'>
$ cd poky
$ git branch -a
$ git tag -l
</literallayout>
This example uses the Yocto Project 1.1.1 Release code named "edison",
which maps to the <filename>edison</filename> branch in the repository.
The following commands create and checkout the local <filename>edison</filename>
branch:
<literallayout class='monospaced'>
$ git checkout -b edison origin/edison
Branch edison set up to track remote branch edison from origin.
Switched to a new branch 'edison'
</literallayout>
</para>
<para>
Next, you need to build the default <filename>qemux86</filename> image that you
can boot using QEMU.
<note>
Because a full build can take hours, you should check two variables in the
<filename>build</filename> directory that is created after you source the
<filename>oe-init-build-env</filename> script.
You can find these variables
<filename>BB_NUMBER_THREADS</filename> and <filename>PARALLEL_MAKE</filename>
in the <filename>build/conf</filename> directory in the
<filename>local.conf</filename> configuration file.
By default, these variables are commented out.
If your host development system supports multi-core and multi-thread capabilities,
you can uncomment these statements and set the variables to significantly shorten
the full build time.
As a guideline, set <filename>BB_NUMBER_THREADS</filename> to twice the number
of cores your machine supports and set <filename>PARALLEL_MAKE</filename> to one and
a half times the number of cores your machine supports.
</note>
The following two commands <filename>source</filename> the build environment setup script
and build the default <filename>qemux86</filename> image.
If necessary, the script creates the build directory:
<literallayout class='monospaced'>
$ cd ~/poky
$ source oe-init-build-env
### Shell environment set up for builds. ###
You can now run 'bitbake &lt;target&gt;'
Common targets are:
core-image-minimal
core-image-sato
meta-toolchain
meta-toolchain-sdk
adt-installer
meta-ide-support
You can also run generated qemu images with a command like 'runqemu qemux86'
</literallayout>
</para>
<para>
The following <filename>bitbake</filename> command starts the build:
<literallayout class='monospaced'>
$ bitbake -k core-image-minimal
</literallayout>
<note>Be sure to check the settings in the <filename>local.conf</filename>
before starting the build.</note>
</para>
</section>
<section id='examining-the-default-config-smp-behavior'>
<title>Examining the Default <filename>CONFIG_SMP</filename> Behavior</title>
<para>
By default, <filename>CONFIG_SMP</filename> supports single processor machines.
To see this default setting from within the QEMU emulator, boot your image using
the emulator as follows:
<literallayout class='monospaced'>
$ runqemu qemux86 qemuparams="-smp 2"
</literallayout>
</para>
<para>
Login to the machine using <filename>root</filename> with no password.
After logging in, enter the following command to see how many processors are
being supported in the emulator.
The emulator reports support for a single processor:
<literallayout class='monospaced'>
# cat /proc/cpuinfo | grep processor
processor : 0
#
</literallayout>
Logout of the emulator using the <filename>exit</filename> command and
then close it down.
</para>
</section>
<section id='changing-the-config-smp-configuration-using-menuconfig'>
<title>Changing the <filename>CONFIG_SMP</filename> Configuration Using <filename>menuconfig</filename></title>
<para>
The <filename>menuconfig</filename> tool provides an interactive method with which
to set kernel configurations.
You need to run <filename>menuconfig</filename> inside the Yocto BitBake environment.
Thus, the environment must be set up using the <filename>oe-init-build-env</filename>
script found in the Yocto Project files Git repository build directory.
If you have not sourced this script do so with the following commands:
<literallayout class='monospaced'>
$ cd ~/poky
$ source oe-init-build-env
</literallayout>
</para>
<para>
After setting up the environment to run <filename>menuconfig</filename>, you are ready
to use the tool to interactively change the kernel configuration.
In this example, we are basing our changes on the <filename>linux-yocto-3.0-1.1.x</filename>
kernel.
The Yocto Project build environment recognizes this kernel as
<filename>linux-yocto</filename>.
Thus, the following command from the shell in which you previously sourced the
environment initialization script launches <filename>menuconfig</filename>:
<literallayout class='monospaced'>
$ bitbake linux-yocto -c menuconfig
</literallayout>
</para>
<para>
Once <filename>menuconfig</filename> launches, navigate through the user interface
to find the <filename>CONFIG_SMP</filename> configuration setting.
You can find it at <filename>Processor Type and Features</filename>.
The configuration selection is
<filename>Symmetric Multi-processing Support</filename>.
After using the arrow keys to highlight this selection, press "y" to select it.
Then, exit out and save your selections.
</para>
<para>
Once you save the selection, the <filename>.config</filename> configuration file
is updated.
This is the file that the build system uses to configure the Linux Yocto kernel
when it is built.
You can find and examine this file in the Yocto Project files Git repository in
the build directory.
This example uses the following.
Note that this example directory is artificially split and many of the characters
in the actually filename are omitted in order to make it more
readable:
<literallayout class='monospaced'>
~/poky/build/tmp/work/qemux86-poky-linux/linux-yocto-3.0.10+git1+d38...
...3a9ac596f7a-r3/linux-qemux86-standard-build
</literallayout>
</para>
<para>
Within the <filename>.config</filename> file, you can see the following setting:
<literallayout class='monospaced'>
CONFIG_SMP=y
</literallayout>
</para>
<para>
A good method to isolate changed configurations is to use a combination of the
<filename>menuconfig</filename> tool and simple shell commands.
Before changing configurations with <filename>menuconfig</filename>, copy the
existing <filename>.config</filename> and rename it to something else,
use <filename>menuconfig</filename> to make
as many changes an you want and save them, then compare the renamed configuration
file against the newly created file.
You can use the resulting differences as your base to create configuration fragments
to permanently save in your kernel layer.
<note>
Be sure to make a copy of the <filename>.config</filename> and don't just
rename it.
The Yocto Project build system needs an existing <filename>.config</filename>
from which to work.
</note>
</para>
</section>
<section id='recompiling-the-kernel-and-testing-the-new-configuration'>
<title>Recompiling the Kernel and Testing the New Configuration</title>
<para>
At this point, you are ready to recompile your kernel image with
the new setting in effect using the BitBake commands below:
<literallayout class='monospaced'>
$ bitbake linux-yocto -c compile -f
$ bitbake linux-yocto
</literallayout>
</para>
<para>
Now run the QEMU emulator:
<literallayout class='monospaced'>
$ runqemu qemux86 qemuparams="-smp 2"
</literallayout>
</para>
<para>
Login to the machine using <filename>root</filename> with no password
and test for the number of processors the kernel supports:
<literallayout class='monospaced'>
# cat /proc/cpuinfo | grep processor
processor : 0
processor : 1
#
</literallayout>
</para>
<para>
From the output, you can see that you have successfully reconfigured the kernel.
</para>
</section>
</section>
<section id='adding-kernel-recipes'>
<title>Adding Kernel Recipes</title>
<para>
A future release of this manual will present an example that adds kernel recipes, which provide
new functionality to the kernel.
</para>
<para>
<imagedata fileref="figures/wip.png"
width="2in" depth="3in" align="center" scalefit="1" />
</para>
</section>
<!-- <section id='is-vfat-supported'>
<title>Is VFAT Supported?</title>
<para>
<literallayout class='monospaced'>
I entered runqemu qemux86 and it fires upthis fires up the emulator and uses the
image and filesystem in the build area created in the previous section.
Then I copied over a pre-created and formated 5.2MB VFAT file named vfat.img.
I did this with scp vfat.img root@192.168.7.2:
The file is in the root directory.
I had to do this because the mkfs.vfat vfat.img command does not work.
mkfs is not recognized in the qemu terminal session.
when I try mount -o loop -t vfat vfat.img mnt/ I get the error
mount: can't set up loop device: No space left on device.
This error is because the loop module is not currently in the kernel image.
However, this module is available in the
build area in the tarball modules-2.6.37.6-yocto-starndard+-20-qemux86.tgz.
You can add this to the kernel image by adding the
IMAGE_INSTALL += " kernel-module-loop" statement at the top of the local.conf
file in the build area and then rebuilding the kernel using bitbake.
It should just build whatever is necessary and not go through an entire build again.
The <filename>menuconfig</filename> tool provides an interactive method with which
to set kernel configurations.
In order to use <filename>menuconfig</filename> from within the BitBake environment
you need to source an environment setup script.
This script is located in the local Yocto Project file structure and is called
<filename>oe-init-build-env</filename>.
</para>
<para>
The following command sets up the environment:
<literallayout class='monospaced'>
$ cd ~/poky
$ source oe-init-build-env
$ runqemu qemux86
Continuing with the following parameters:
KERNEL: [/home/scottrif/poky/build/tmp/deploy/images/bzImage-qemux86.bin]
ROOTFS: [/home/scottrif/poky/build/tmp/deploy/images/core-image-sato-qemux86.ext3]
FSTYPE: [ext3]
Setting up tap interface under sudo
Acquiring lockfile for tap0...
WARNING: distccd not present, no distcc support loaded.
Running qemu...
/home/scottrif/poky/build/tmp/sysroots/x86_64-linux/usr/bin/qemu
-kernel /home/scottrif/poky/build/tmp/deploy/images/bzImage-qemux86.bin
-net nic,vlan=0 -net tap,vlan=0,ifname=tap0,script=no,downscript=no
-hda /home/scottrif/poky/build/tmp/deploy/images/core-image-sato-qemux86.ext3
-show-cursor -usb -usbdevice wacom-tablet -vga vmware -enable-gl -no-reboot
-m 128 &dash;&dash;append "vga=0 root=/dev/hda rw mem=128M ip=192.168.7.2::192.168.7.1:255.255.255.0 oprofile.timer=1 "
Enabling opengl
vmsvga_value_write: guest runs Linux.
</literallayout>
</para>
</section>
<section id='prepare-to-use-menuconfig'>
<title>Prepare to use <filename>menuconfig</filename></title>
<para>
[WRITER'S NOTE: Stuff from here down are crib notes]
</para>
<para>
Once menuconfig fires up you see all kinds of categories that you can interactively
investigate.
If they have an "M" in it then the feature is "modularized".
I guess that means that means that it needs to be manually linked in when the
kernel is booted??? (Not sure).
If they have an "*" then the feature is automatically part of the kernel.]
</para>
<para>
So the tmp/work/ area was created in poky and there is a .config file in there and
a .config.old file.
The old one must have been created when I exited from menuconfig after poking around
a bit.
Nope - appears to just be created automatically.
</para>
<para>
A good practice is to first determine what configurations you have for the kernel.
You can see the results by looking in the .config file in the build/tmp/work/qemux86-poky-linux area
of the local YP files.
There is a directory named linux-yocto-2.6.37* in the directory.
In that directory is a directory named linux-qemux86-standard-build.
In that directory you will find a file named .config that is the configuration file
for the kernel that will be used when you build the kernel.
You can open that file up and examine it.
If you do a search for "VFAT" you will see that that particular configuration is not
enabled for the kernel.
This means that you cannot print a VFAT text file, or for that matter, even mount one
from the image if you were to build it at this point.
</para>
<para>
You can prove the point by actually trying it at this point.
Here are the commands:
<literallayout class='monospaced'>
$ mkdir ~/vfat-test
$ cd ~/vfat-test
$ dd if=/dev/zero of=vfat.img bs=1024 count=5000 [creates a 5MB disk image]
5+0 records in
5+0 records out
5242880 bytes (5.2 MB) copied, 0.00798912 s, 656 MB/s
$ ls -lah [lists the contents of the new image. l=long, a=all, h=human readable]
total 5.1M
drwxr-xr-x 2 srifenbark scottrif 4.0K 2011-08-01 08:18 .
drwxr-xr-x 66 srifenbark scottrif 4.0K 2011-08-01 08:14 ..
-rw-r&dash;&dash;r&dash;&dash; 1 srifenbark scottrif 5.0M 2011-08-01 08:18 vfat.img
$ mkfs.vfat vfat.img [formats the disk image]
mkfs.vfat 3.0.7 (24 Dec 2009)
$ mkdir mnt [mounts the disk image]
$ sudo su [gives you root privilege]
# mount -o loop vfat.img mnt [mounts it as a loop device]
# ls mnt [shows nothing in mnt]
# mount [lists the mounted filesystems - note/dev/loop0]
/dev/sda1 on / type ext4 (rw,errors=remount-ro)
proc on /proc type proc (rw,noexec,nosuid,nodev)
none on /sys type sysfs (rw,noexec,nosuid,nodev)
none on /sys/fs/fuse/connections type fusectl (rw)
none on /sys/kernel/debug type debugfs (rw)
none on /sys/kernel/security type securityfs (rw)
none on /dev type devtmpfs (rw,mode=0755)
none on /dev/pts type devpts (rw,noexec,nosuid,gid=5,mode=0620)
none on /dev/shm type tmpfs (rw,nosuid,nodev)
none on /var/run type tmpfs (rw,nosuid,mode=0755)
none on /var/lock type tmpfs (rw,noexec,nosuid,nodev)
none on /lib/init/rw type tmpfs (rw,nosuid,mode=0755)
binfmt_misc on /proc/sys/fs/binfmt_misc type binfmt_misc (rw,noexec,nosuid,nodev)
gvfs-fuse-daemon on /home/scottrif/.gvfs type fuse.gvfs-fuse-daemon (rw,nosuid,nodev,user=srifenbark)
/dev/loop0 on /home/scottrif/vfat-test/mnt type vfat (rw)
# echo "hello world" > mnt/hello.txt [creates a text file in the mounted VFAT system]
# ls mnt [verifies the file is there]
hello.txt
# cat mnt/hello.txt [displays the contents of the file created]
hello world
# umount mnt [unmounts the system and destroys the loop]
# exit [gets out of privileged user mode]
exit
$ lsmod [this stuff Darren did to show me ]
Module Size Used by [the status of modules in the regular linux kernel]
nls_iso8859_1 4633 0
nls_cp437 6351 0
vfat 10866 0
fat 55350 1 vfat
snd_hda_codec_atihdmi 3023 1
binfmt_misc 7960 1
snd_hda_codec_realtek 279008 1
ppdev 6375 0
snd_hda_intel 25805 2
fbcon 39270 71
tileblit 2487 1 fbcon
font 8053 1 fbcon
bitblit 5811 1 fbcon
snd_hda_codec 85759 3 snd_hda_codec_atihdmi,snd_hda_codec_realtek,snd_hda_intel
softcursor 1565 1 bitblit
snd_seq_dummy 1782 0
snd_hwdep 6924 1 snd_hda_codec
vga16fb 12757 0
snd_pcm_oss 41394 0
snd_mixer_oss 16299 1 snd_pcm_oss
snd_pcm 87946 3 snd_hda_intel,snd_hda_codec,snd_pcm_oss
vgastate 9857 1 vga16fb
snd_seq_oss 31191 0
snd_seq_midi 5829 0
snd_rawmidi 23420 1 snd_seq_midi
radeon 744506 3
snd_seq_midi_event 7267 2 snd_seq_oss,snd_seq_midi
ttm 61007 1 radeon
snd_seq 57481 6 snd_seq_dummy,snd_seq_oss,snd_seq_midi,snd_seq_midi_event
drm_kms_helper 30742 1 radeon
snd_timer 23649 2 snd_pcm,snd_seq
snd_seq_device 6888 5 snd_seq_dummy,snd_seq_oss,snd_seq_midi,snd_rawmidi,snd_seq
usb_storage 50377 0
snd 71283 16 \
snd_hda_codec_realtek,snd_hda_intel,snd_hda_codec, \
snd_hwdep,snd_pcm_oss,snd_mixer_oss,snd_pcm, \
snd_seq_oss,snd_rawmidi,snd_seq,snd_timer,snd_seq_device
soundcore 8052 1 snd
psmouse 65040 0
drm 198886 5 radeon,ttm,drm_kms_helper
i2c_algo_bit 6024 1 radeon
serio_raw 4918 0
snd_page_alloc 8500 2 snd_hda_intel,snd_pcm
dell_wmi 2177 0
dcdbas 6886 0
lp 9336 0
parport 37160 2 ppdev,lp
usbhid 41116 0
ohci1394 30260 0
hid 83888 1 usbhid
ieee1394 94771 1 ohci1394
tg3 122382 0
</literallayout>
</para>
</section>
</section> -->
</appendix>
<!--
EXTRA STUFF I MIGHT NEED BUT NOW SURE RIGHT NOW.
In the standard layer structure you have several areas that you need to examine or
modify.
For this example the layer contains four areas:
<itemizedlist>
<listitem><para><emphasis><filename>conf</filename></emphasis> - Contains the
<filename>layer.conf</filename> that identifies the location of the recipe files.
</para></listitem>
<listitem><para><emphasis><filename>images</filename></emphasis> - Contains the
image recipe file.
This recipe includes the base image you will be using and specifies other
packages the image might need.</para></listitem>
<listitem><para><emphasis><filename>recipes-bsp</filename></emphasis> - Contains
recipes specific to the hardware for which you are developing the kernel.
</para></listitem>
<listitem><para><emphasis><filename>recipes-kernel</filename></emphasis> - Contains the
"append" files that add information to the main recipe kernel.
</para></listitem>
</itemizedlist>
</para>
<para>
Let's take a look at the <filename>layer.conf</filename> in the
<filename>conf</filename> directory first.
This configuration file enables the Yocto Project build system to locate and
use the information in your new layer.
</para>
<para>
The variable <filename>BBPATH</filename> needs to include the path to your layer
as follows:
<literallayout class='monospaced'>
BBPATH := "${BBPATH}:${LAYERDIR}"
</literallayout>
And, the variable <filename>BBFILES</filename> needs to be modified to include your
recipe and append files:
<literallayout class='monospaced'>
BBFILES := "${BBFILES} ${LAYERDIR}/images/*.bb \
${LAYERDIR}/images/*.bbappend \
${LAYERDIR}/recipes-*/*/*.bb \
${LAYERDIR}/recipes-*/*/*.bbappend"
</literallayout>
Finally, you need to be sure to use your layer name in these variables at the
end of the file:
<literallayout class='monospaced'>
BBFILE_COLLECTIONS += "elc"
BBFILE_PATTERN_elc := "^${LAYERDIR}/"
BBFILE_PRIORITY_elc = "9"
</literallayout>
</para>
<para>
The <filename>images</filename> directory contains an append file that helps
further define the image.
In our example, the base image is <filename>core-image-minimal</filename>.
The image does, however, need some additional modules that we are using
for this example.
These modules support the amixer functionality.
Here is the append file:
<literallayout class='monospaced'>
require recipes-core/images/poky-image-minimal.bb
IMAGE_INSTALL += "dropbear alsa-utils-aplay alsa-utils-alsamixer"
IMAGE_INSTALL_append_qemux86 += " kernel-module-snd-ens1370 \
kernel-module-snd-rawmidi kernel-module-loop kernel-module-nls-cp437 \
kernel-module-nls-iso8859-1 qemux86-audio alsa-utils-amixer"
LICENSE = "MIT"
</literallayout>
</para>
<para>
While the focus of this example is not on the BSP, it is worth mentioning that the
<filename>recipes-bsp</filename> directory has the recipes and append files for
features that the hardware requires.
In this example, there is a script and a recipe to support the
<filename>amixer</filename> functionality in QEMU.
It is beyond the scope of this manual to go too deeply into the script.
Suffice it to say that the script tests for the presence of the mixer, sets up
default mixer values, enables the mixer, unmutes master and then
sets the volume to 100.
</para>
<para>
The recipe <filename>qemu86-audio.bb</filename> installs and runs the
<filename>amixer</filename> when the system boots.
Here is the recipe:
<literallayout class='monospaced'>
SUMMARY = "Provide a basic init script to enable audio"
DESCRIPTION = "Set the volume and unmute the Front mixer setting during boot."
SECTION = "base"
LICENSE = "MIT"
LIC_FILES_CHKSUM = "file://${POKYBASE}/LICENSE;md5=3f40d7994397109285ec7b81fdeb3b58"
PR = "r4"
inherit update-rc.d
RDEPENDS = "alsa-utils-amixer"
SRC_URI = "file://qemux86-audio"
INITSCRIPT_NAME = "qemux86-audio"
INITSCRIPT_PARAMS = "defaults 90"
do_install() {
install -d ${D}${sysconfdir} \
${D}${sysconfdir}/init.d
install -m 0755 ${WORKDIR}/qemux86-audio ${D}${sysconfdir}/init.d
cat ${WORKDIR}/${INITSCRIPT_NAME} | \
sed -e 's,/etc,${sysconfdir},g' \
-e 's,/usr/sbin,${sbindir},g' \
-e 's,/var,${localstatedir},g' \
-e 's,/usr/bin,${bindir},g' \
-e 's,/usr,${prefix},g' > ${D}${sysconfdir}/init.d/${INITSCRIPT_NAME}
chmod 755 ${D}${sysconfdir}/init.d/${INITSCRIPT_NAME}
}
</literallayout>
</para>
<para>
The last area to look at is <filename>recipes-kernel</filename>.
This area holds configuration fragments and kernel append files.
The append file must have the same name as the kernel recipe, which is
<filename>linux-yocto-2.6.37</filename> in this example.
The file can <filename>SRC_URI</filename> statements to point to configuration
fragments you might have in the layer.
The file can also contain <filename>KERNEL_FEATURES</filename> statements that specify
included kernel configurations that ship with the Yocto Project.
</para>
-->
<!--
vim: expandtab tw=80 ts=4
-->