1
0
forked from brl/citadel
citadel/poky/documentation/ref-manual/faq.xml

835 lines
37 KiB
XML
Raw Normal View History

<!DOCTYPE chapter PUBLIC "-//OASIS//DTD DocBook XML V4.2//EN"
"http://www.oasis-open.org/docbook/xml/4.2/docbookx.dtd"
[<!ENTITY % poky SYSTEM "../poky.ent"> %poky; ] >
<chapter id='faq'>
<title>FAQ</title>
<qandaset>
<qandaentry>
<question>
<para>
How does Poky differ from <ulink url='&OE_HOME_URL;'>OpenEmbedded</ulink>?
</para>
</question>
<answer>
<para>
The term "<link link='poky'>Poky</link>"
refers to the specific reference build system that
the Yocto Project provides.
Poky is based on <link linkend='oe-core'>OE-Core</link>
and <link linkend='bitbake-term'>BitBake</link>.
Thus, the generic term used here for the build system is
the "OpenEmbedded build system."
Development in the Yocto Project using Poky is closely tied to OpenEmbedded, with
changes always being merged to OE-Core or BitBake first before being pulled back
into Poky.
This practice benefits both projects immediately.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para id='faq-not-meeting-requirements'>
My development system does not meet the
required Git, tar, and Python versions.
In particular, I do not have Python 3.4.0 or greater.
Can I still use the Yocto Project?
</para>
</question>
<answer>
<para>
You can get the required tools on your host development
system a couple different ways (i.e. building a tarball or
downloading a tarball).
See the
"<link linkend='required-git-tar-and-python-versions'>Required Git, tar, and Python Versions</link>"
section for steps on how to update your build tools.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
How can you claim Poky / OpenEmbedded-Core is stable?
</para>
</question>
<answer>
<para>
There are three areas that help with stability;
<itemizedlist>
<listitem><para>The Yocto Project team keeps
<link linkend='oe-core'>OE-Core</link> small
and focused, containing around 830 recipes as opposed to the thousands
available in other OpenEmbedded community layers.
Keeping it small makes it easy to test and maintain.</para></listitem>
<listitem><para>The Yocto Project team runs manual and automated tests
using a small, fixed set of reference hardware as well as emulated
targets.</para></listitem>
<listitem><para>The Yocto Project uses an autobuilder,
which provides continuous build and integration tests.</para></listitem>
</itemizedlist>
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
How do I get support for my board added to the Yocto Project?
</para>
</question>
<answer>
<para>
Support for an additional board is added by creating a
Board Support Package (BSP) layer for it.
For more information on how to create a BSP layer, see the
"<ulink url='&YOCTO_DOCS_DEV_URL;#understanding-and-creating-layers'>Understanding and Creating Layers</ulink>"
section in the Yocto Project Development Tasks Manual and the
<ulink url='&YOCTO_DOCS_BSP_URL;'>Yocto Project Board Support Package (BSP) Developer's Guide</ulink>.
</para>
<para>
Usually, if the board is not completely exotic, adding support in
the Yocto Project is fairly straightforward.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
Are there any products built using the OpenEmbedded build system?
</para>
</question>
<answer>
<para>
The software running on the <ulink url='http://vernier.com/labquest/'>Vernier LabQuest</ulink>
is built using the OpenEmbedded build system.
See the <ulink url='http://www.vernier.com/products/interfaces/labq/'>Vernier LabQuest</ulink>
website for more information.
There are a number of pre-production devices using the OpenEmbedded build system
and the Yocto Project team
announces them as soon as they are released.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
What does the OpenEmbedded build system produce as output?
</para>
</question>
<answer>
<para>
Because you can use the same set of recipes to create output of
various formats, the output of an OpenEmbedded build depends on
how you start it.
Usually, the output is a flashable image ready for the target
device.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
How do I add my package to the Yocto Project?
</para>
</question>
<answer>
<para>
To add a package, you need to create a BitBake recipe.
For information on how to create a BitBake recipe, see the
"<ulink url='&YOCTO_DOCS_DEV_URL;#new-recipe-writing-a-new-recipe'>Writing a New Recipe</ulink>"
in the Yocto Project Development Tasks Manual.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
Do I have to reflash my entire board with a new Yocto Project image when recompiling
a package?
</para>
</question>
<answer>
<para>
The OpenEmbedded build system can build packages in various
formats such as IPK for OPKG, Debian package
(<filename>.deb</filename>), or RPM.
You can then upgrade the packages using the package tools on
the device, much like on a desktop distribution such as
Ubuntu or Fedora.
However, package management on the target is entirely optional.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
I see the error '<filename>chmod: XXXXX new permissions are r-xrwxrwx, not r-xr-xr-x</filename>'.
What is wrong?
</para>
</question>
<answer>
<para>
You are probably running the build on an NTFS filesystem.
Use <filename>ext2</filename>, <filename>ext3</filename>, or <filename>ext4</filename> instead.
</para>
</answer>
</qandaentry>
<!-- <qandaentry>
<question>
<para>
How do I make the Yocto Project work in RHEL/CentOS?
</para>
</question>
<answer>
<para>
To get the Yocto Project working under RHEL/CentOS 5.1 you need to first
install some required packages.
The standard CentOS packages needed are:
<itemizedlist>
<listitem><para>"Development tools" (selected during installation)</para></listitem>
<listitem><para><filename>texi2html</filename></para></listitem>
<listitem><para><filename>compat-gcc-34</filename></para></listitem>
</itemizedlist>
On top of these, you need the following external packages:
<itemizedlist>
<listitem><para><filename>python-sqlite2</filename> from
<ulink url='http://dag.wieers.com/rpm/packages/python-sqlite2/'>DAG repository</ulink>
</para></listitem>
<listitem><para><filename>help2man</filename> from
<ulink url='http://centos.karan.org/el4/extras/stable/x86_64/RPMS/repodata/repoview/help2man-0-1.33.1-2.html'>Karan repository</ulink></para></listitem>
</itemizedlist>
</para>
<para>
Once these packages are installed, the OpenEmbedded build system will be able
to build standard images.
However, there might be a problem with the QEMU emulator segfaulting.
You can either disable the generation of binary locales by setting
<filename><link linkend='var-ENABLE_BINARY_LOCALE_GENERATION'>ENABLE_BINARY_LOCALE_GENERATION</link>
</filename> to "0" or by removing the <filename>linux-2.6-execshield.patch</filename>
from the kernel and rebuilding it since that is the patch that causes the problems with QEMU.
</para>
<note>
<para>For information on distributions that the Yocto Project
uses during validation, see the
<ulink url='&YOCTO_WIKI_URL;/wiki/Distribution_Support'>Distribution Support</ulink>
Wiki page.</para>
<para>For notes about using the Yocto Project on a RHEL 4-based
host, see the
<ulink url='&YOCTO_WIKI_URL;/wiki/BuildingOnRHEL4'>Building on RHEL4</ulink>
Wiki page.</para>
</note>
</answer>
</qandaentry> -->
<qandaentry>
<question>
<para>
I see lots of 404 responses for files when the OpenEmbedded
build system is trying to download sources.
Is something wrong?
</para>
</question>
<answer>
<para>
Nothing is wrong.
The OpenEmbedded build system checks any configured source mirrors before downloading
from the upstream sources.
The build system does this searching for both source archives and
pre-checked out versions of SCM-managed software.
These checks help in large installations because it can reduce load on the SCM servers
themselves.
The address above is one of the default mirrors configured into the
build system.
Consequently, if an upstream source disappears, the team
can place sources there so builds continue to work.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
I have machine-specific data in a package for one machine only but the package is
being marked as machine-specific in all cases, how do I prevent this?
</para>
</question>
<answer>
<para>
Set <filename><link linkend='var-SRC_URI_OVERRIDES_PACKAGE_ARCH'>SRC_URI_OVERRIDES_PACKAGE_ARCH</link>
</filename> = "0" in the <filename>.bb</filename> file but make sure the package is
manually marked as
machine-specific for the case that needs it.
The code that handles
<filename>SRC_URI_OVERRIDES_PACKAGE_ARCH</filename> is in
the <filename>meta/classes/base.bbclass</filename> file.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para id='i-am-behind-a-firewall-and-need-to-use-a-proxy-server'>
I'm behind a firewall and need to use a proxy server. How do I do that?
</para>
</question>
<answer>
<para>
Most source fetching by the OpenEmbedded build system is done
by <filename>wget</filename> and you therefore need to specify
the proxy settings in a <filename>.wgetrc</filename> file,
which can be in your home directory if you are a single user
or can be in <filename>/usr/local/etc/wgetrc</filename> as
a global user file.
</para>
<para>
Following is the applicable code for setting various proxy
types in the <filename>.wgetrc</filename> file.
By default, these settings are disabled with comments.
To use them, remove the comments:
<literallayout class='monospaced'>
# You can set the default proxies for Wget to use for http, https, and ftp.
# They will override the value in the environment.
#https_proxy = http://proxy.yoyodyne.com:18023/
#http_proxy = http://proxy.yoyodyne.com:18023/
#ftp_proxy = http://proxy.yoyodyne.com:18023/
# If you do not want to use proxy at all, set this to off.
#use_proxy = on
</literallayout>
The Yocto Project also includes a
<filename>meta-poky/conf/site.conf.sample</filename> file that
shows how to configure CVS and Git proxy servers if needed.
For more information on setting up various proxy types and
configuring proxy servers, see the
"<ulink url='&YOCTO_WIKI_URL;/wiki/Working_Behind_a_Network_Proxy'>Working Behind a Network Proxy</ulink>"
Wiki page.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
Whats the difference between <replaceable>target</replaceable> and <replaceable>target</replaceable><filename>-native</filename>?
</para>
</question>
<answer>
<para>
The <filename>*-native</filename> targets are designed to run on the system
being used for the build.
These are usually tools that are needed to assist the build in some way such as
<filename>quilt-native</filename>, which is used to apply patches.
The non-native version is the one that runs on the target device.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
I'm seeing random build failures. Help?!
</para>
</question>
<answer>
<para>
If the same build is failing in totally different and random
ways, the most likely explanation is:
<itemizedlist>
<listitem><para>The hardware you are running the build on
has some problem.</para></listitem>
<listitem><para>You are running the build under
virtualization, in which case the virtualization
probably has bugs.</para></listitem>
</itemizedlist>
The OpenEmbedded build system processes a massive amount of
data that causes lots of network, disk and CPU activity and
is sensitive to even single-bit failures in any of these areas.
True random failures have always been traced back to hardware
or virtualization issues.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
When I try to build a native recipe, the build fails with <filename>iconv.h</filename> problems.
</para>
</question>
<answer>
<para>
If you get an error message that indicates GNU
<filename>libiconv</filename> is not in use but
<filename>iconv.h</filename> has been included from
<filename>libiconv</filename>, you need to check to see if
you have a previously installed version of the header file
in <filename>/usr/local/include</filename>.
<literallayout class='monospaced'>
#error GNU libiconv not in use but included iconv.h is from libiconv
</literallayout>
If you find a previously installed file, you should either
uninstall it or temporarily rename it and try the build again.
</para>
<para>
This issue is just a single manifestation of "system
leakage" issues caused when the OpenEmbedded build system
finds and uses previously installed files during a native
build.
This type of issue might not be limited to
<filename>iconv.h</filename>.
Be sure that leakage cannot occur from
<filename>/usr/local/include</filename> and
<filename>/opt</filename> locations.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
What do we need to ship for license compliance?
</para>
</question>
<answer>
<para>
This is a difficult question and you need to consult your lawyer
for the answer for your specific case.
It is worth bearing in mind that for GPL compliance, there needs
to be enough information shipped to allow someone else to
rebuild and produce the same end result you are shipping.
This means sharing the source code, any patches applied to it,
and also any configuration information about how that package
was configured and built.
</para>
<para>
You can find more information on licensing in the
"<link linkend='licensing'>Licensing</link>" section and in the
"<ulink url='&YOCTO_DOCS_DEV_URL;#maintaining-open-source-license-compliance-during-your-products-lifecycle'>Maintaining Open Source License Compliance During Your Product's Lifecycle</ulink>"
section, which is in the Yocto Project Development Tasks
Manual.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
How do I disable the cursor on my touchscreen device?
</para>
</question>
<answer>
<para>
You need to create a form factor file as described in the
"<ulink url='&YOCTO_DOCS_BSP_URL;#bsp-filelayout-misc-recipes'>Miscellaneous BSP-Specific Recipe Files</ulink>"
section in the Yocto Project Board Support Packages (BSP)
Developer's Guide.
Set the <filename>HAVE_TOUCHSCREEN</filename> variable equal to
one as follows:
<literallayout class='monospaced'>
HAVE_TOUCHSCREEN=1
</literallayout>
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
How do I make sure connected network interfaces are brought up by default?
</para>
</question>
<answer>
<para>
The default interfaces file provided by the netbase recipe does not
automatically bring up network interfaces.
Therefore, you will need to add a BSP-specific netbase that includes an interfaces
file.
See the "<ulink url='&YOCTO_DOCS_BSP_URL;#bsp-filelayout-misc-recipes'>Miscellaneous BSP-Specific Recipe Files</ulink>"
section in the Yocto Project Board Support Packages (BSP)
Developer's Guide for information on creating these types of
miscellaneous recipe files.
</para>
<para>
For example, add the following files to your layer:
<literallayout class='monospaced'>
meta-MACHINE/recipes-bsp/netbase/netbase/MACHINE/interfaces
meta-MACHINE/recipes-bsp/netbase/netbase_5.0.bbappend
</literallayout>
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
How do I create images with more free space?
</para>
</question>
<answer>
<para>
By default, the OpenEmbedded build system creates images
that are 1.3 times the size of the populated root filesystem.
To affect the image size, you need to set various
configurations:
<itemizedlist>
<listitem><para><emphasis>Image Size:</emphasis>
The OpenEmbedded build system uses the
<link linkend='var-IMAGE_ROOTFS_SIZE'><filename>IMAGE_ROOTFS_SIZE</filename></link>
variable to define the size of the image in Kbytes.
The build system determines the size by taking into
account the initial root filesystem size before any
modifications such as requested size for the image and
any requested additional free disk space to be
added to the image.</para></listitem>
<listitem><para><emphasis>Overhead:</emphasis>
Use the
<link linkend='var-IMAGE_OVERHEAD_FACTOR'><filename>IMAGE_OVERHEAD_FACTOR</filename></link>
variable to define the multiplier that the build system
applies to the initial image size, which is 1.3 by
default.</para></listitem>
<listitem><para><emphasis>Additional Free Space:</emphasis>
Use the
<link linkend='var-IMAGE_ROOTFS_EXTRA_SPACE'><filename>IMAGE_ROOTFS_EXTRA_SPACE</filename></link>
variable to add additional free space to the image.
The build system adds this space to the image after
it determines its
<filename>IMAGE_ROOTFS_SIZE</filename>.
</para></listitem>
</itemizedlist>
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
Why don't you support directories with spaces in the pathnames?
</para>
</question>
<answer>
<para>
The Yocto Project team has tried to do this before but too
many of the tools the OpenEmbedded build system depends on,
such as <filename>autoconf</filename>, break when they find
spaces in pathnames.
Until that situation changes, the team will not support spaces
in pathnames.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
How do I use an external toolchain?
</para>
</question>
<answer>
<para>
The toolchain configuration is very flexible and customizable.
It is primarily controlled with the
<filename><link linkend='var-TCMODE'>TCMODE</link></filename>
variable.
This variable controls which <filename>tcmode-*.inc</filename>
file to include from the
<filename>meta/conf/distro/include</filename> directory within
the
<link linkend='source-directory'>Source Directory</link>.
</para>
<para>
The default value of <filename>TCMODE</filename> is "default",
which tells the OpenEmbedded build system to use its internally
built toolchain (i.e. <filename>tcmode-default.inc</filename>).
However, other patterns are accepted.
In particular, "external-*" refers to external toolchains.
One example is the Sourcery G++ Toolchain.
The support for this toolchain resides in the separate
<filename>meta-sourcery</filename> layer at
<ulink url='http://github.com/MentorEmbedded/meta-sourcery/'></ulink>.
</para>
<para>
In addition to the toolchain configuration, you also need a
corresponding toolchain recipe file.
This recipe file needs to package up any pre-built objects in
the toolchain such as <filename>libgcc</filename>,
<filename>libstdcc++</filename>, any locales, and
<filename>libc</filename>.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para id='how-does-the-yocto-project-obtain-source-code-and-will-it-work-behind-my-firewall-or-proxy-server'>
How does the OpenEmbedded build system obtain source code and
will it work behind my firewall or proxy server?
</para>
</question>
<answer>
<para>
The way the build system obtains source code is highly
configurable.
You can setup the build system to get source code in most
environments if HTTP transport is available.
</para>
<para>
When the build system searches for source code, it first
tries the local download directory.
If that location fails, Poky tries
<link linkend='var-PREMIRRORS'><filename>PREMIRRORS</filename></link>,
the upstream source, and then
<link linkend='var-MIRRORS'><filename>MIRRORS</filename></link>
in that order.
</para>
<para>
Assuming your distribution is "poky", the OpenEmbedded build
system uses the Yocto Project source
<filename>PREMIRRORS</filename> by default for SCM-based
sources, upstreams for normal tarballs, and then falls back
to a number of other mirrors including the Yocto Project
source mirror if those fail.
</para>
<para>
As an example, you could add a specific server for the
build system to attempt before any others by adding something
like the following to the <filename>local.conf</filename>
configuration file:
<literallayout class='monospaced'>
PREMIRRORS_prepend = "\
git://.*/.* http://www.yoctoproject.org/sources/ \n \
ftp://.*/.* http://www.yoctoproject.org/sources/ \n \
http://.*/.* http://www.yoctoproject.org/sources/ \n \
https://.*/.* http://www.yoctoproject.org/sources/ \n"
</literallayout>
</para>
<para>
These changes cause the build system to intercept Git, FTP,
HTTP, and HTTPS requests and direct them to the
<filename>http://</filename> sources mirror.
You can use <filename>file://</filename> URLs to point to
local directories or network shares as well.
</para>
<para>
Aside from the previous technique, these options also exist:
<literallayout class='monospaced'>
BB_NO_NETWORK = "1"
</literallayout>
This statement tells BitBake to issue an error instead of
trying to access the Internet.
This technique is useful if you want to ensure code builds
only from local sources.
</para>
<para>
Here is another technique:
<literallayout class='monospaced'>
BB_FETCH_PREMIRRORONLY = "1"
</literallayout>
This statement limits the build system to pulling source
from the <filename>PREMIRRORS</filename> only.
Again, this technique is useful for reproducing builds.
</para>
<para>
Here is another technique:
<literallayout class='monospaced'>
BB_GENERATE_MIRROR_TARBALLS = "1"
</literallayout>
This statement tells the build system to generate mirror
tarballs.
This technique is useful if you want to create a mirror server.
If not, however, the technique can simply waste time during
the build.
</para>
<para>
Finally, consider an example where you are behind an
HTTP-only firewall.
You could make the following changes to the
<filename>local.conf</filename> configuration file as long as
the <filename>PREMIRRORS</filename> server is current:
<literallayout class='monospaced'>
PREMIRRORS_prepend = "\
ftp://.*/.* http://www.yoctoproject.org/sources/ \n \
http://.*/.* http://www.yoctoproject.org/sources/ \n \
https://.*/.* http://www.yoctoproject.org/sources/ \n"
BB_FETCH_PREMIRRORONLY = "1"
</literallayout>
These changes would cause the build system to successfully
fetch source over HTTP and any network accesses to anything
other than the <filename>PREMIRRORS</filename> would fail.
</para>
<para>
The build system also honors the standard shell environment
variables <filename>http_proxy</filename>,
<filename>ftp_proxy</filename>,
<filename>https_proxy</filename>, and
<filename>all_proxy</filename> to redirect requests through
proxy servers.
</para>
<note>
You can find more information on the
"<ulink url='&YOCTO_WIKI_URL;/wiki/Working_Behind_a_Network_Proxy'>Working Behind a Network Proxy</ulink>"
Wiki page.
</note>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
Can I get rid of build output so I can start over?
</para>
</question>
<answer>
<para>
Yes - you can easily do this.
When you use BitBake to build an image, all the build output
goes into the directory created when you run the
build environment setup script (i.e.
<link linkend='structure-core-script'><filename>&OE_INIT_FILE;</filename></link>).
By default, this
<link linkend='build-directory'>Build Directory</link>
is named <filename>build</filename> but can be named
anything you want.
</para>
<para>
Within the Build Directory, is the <filename>tmp</filename>
directory.
To remove all the build output yet preserve any source code or
downloaded files from previous builds, simply remove the
<filename>tmp</filename> directory.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
Why do <filename>${bindir}</filename> and <filename>${libdir}</filename> have strange values for <filename>-native</filename> recipes?
</para>
</question>
<answer>
<para>
Executables and libraries might need to be used from a
directory other than the directory into which they were
initially installed.
Complicating this situation is the fact that sometimes these
executables and libraries are compiled with the expectation
of being run from that initial installation target directory.
If this is the case, moving them causes problems.
</para>
<para>
This scenario is a fundamental problem for package maintainers
of mainstream Linux distributions as well as for the
OpenEmbedded build system.
As such, a well-established solution exists.
Makefiles, Autotools configuration scripts, and other build
systems are expected to respect environment variables such as
<filename>bindir</filename>, <filename>libdir</filename>,
and <filename>sysconfdir</filename> that indicate where
executables, libraries, and data reside when a program is
actually run.
They are also expected to respect a
<filename>DESTDIR</filename> environment variable, which is
prepended to all the other variables when the build system
actually installs the files.
It is understood that the program does not actually run from
within <filename>DESTDIR</filename>.
</para>
<para>
When the OpenEmbedded build system uses a recipe to build a
target-architecture program (i.e. one that is intended for
inclusion on the image being built), that program eventually
runs from the root file system of that image.
Thus, the build system provides a value of "/usr/bin" for
<filename>bindir</filename>, a value of "/usr/lib" for
<filename>libdir</filename>, and so forth.
</para>
<para>
Meanwhile, <filename>DESTDIR</filename> is a path within the
<link linkend='build-directory'>Build Directory</link>.
However, when the recipe builds a native program (i.e. one
that is intended to run on the build machine), that program
is never installed directly to the build machine's root
file system.
Consequently, the build system uses paths within the Build
Directory for <filename>DESTDIR</filename>,
<filename>bindir</filename> and related variables.
To better understand this, consider the following two paths
where the first is relatively normal and the second is not:
<note>
Due to these lengthy examples, the paths are artificially
broken across lines for readability.
</note>
<literallayout class='monospaced'>
/home/maxtothemax/poky-bootchart2/build/tmp/work/i586-poky-linux/zlib/
1.2.8-r0/sysroot-destdir/usr/bin
/home/maxtothemax/poky-bootchart2/build/tmp/work/x86_64-linux/
zlib-native/1.2.8-r0/sysroot-destdir/home/maxtothemax/poky-bootchart2/
build/tmp/sysroots/x86_64-linux/usr/bin
</literallayout>
Even if the paths look unusual, they both are correct -
the first for a target and the second for a native recipe.
These paths are a consequence of the
<filename>DESTDIR</filename> mechanism and while they
appear strange, they are correct and in practice very effective.
</para>
</answer>
</qandaentry>
<qandaentry>
<question>
<para>
The files provided by my <filename>*-native</filename> recipe do
not appear to be available to other recipes.
Files are missing from the native sysroot, my recipe is
installing to the wrong place, or I am getting permissions
errors during the do_install task in my recipe! What is wrong?
</para>
</question>
<answer>
<para>
This situation results when a build system does
not recognize the environment variables supplied to it by
<link linkend='bitbake-term'>BitBake</link>.
The incident that prompted this FAQ entry involved a Makefile
that used an environment variable named
<filename>BINDIR</filename> instead of the more standard
variable <filename>bindir</filename>.
The makefile's hardcoded default value of "/usr/bin" worked
most of the time, but not for the recipe's
<filename>-native</filename> variant.
For another example, permissions errors might be caused
by a Makefile that ignores <filename>DESTDIR</filename> or uses
a different name for that environment variable.
Check the the build system to see if these kinds of
issues exist.
</para>
</answer>
</qandaentry>
</qandaset>
</chapter>
<!--
vim: expandtab tw=80 ts=4
-->