meta-intel/README

meta-intel
==========

This is the location for Intel maintained BSPs.

Please see the README files contained in the individual BSP layers for
BSP-specific information. For details on the intel-common BSPs, see the
conf/machine/README file.

If you have problems with or questions about a particular BSP, please
contact the maintainer listed in the MAINTAINERS file directly (cc:ing
the Yocto mailing list puts it in the archive and helps other people
who might have the same questions in the future), but please try to do
the following first:

  - look in the Yocto Project Bugzilla
    (http://bugzilla.yoctoproject.org/) to see if a problem has
    already been reported

  - look through recent entries of the meta-intel
    (https://lists.yoctoproject.org/pipermail/meta-intel/) and Yocto
    (https://lists.yoctoproject.org/pipermail/yocto/) mailing list
    archives to see if other people have run into similar problems or
    had similar questions answered.

If you believe you have encountered a bug, you can open a new bug and
enter the details in the Yocto Project Bugzilla
(http://bugzilla.yoctoproject.org/).  If you're relatively certain
that it's a bug against the BSP itself, please use the 'Yocto Project
Components: BSPs | meta-intel' category for the bug; otherwise, please
submit the bug against the most likely category for the problem - if
you're wrong, it's not a big deal and the bug will be recategorized
upon triage.

Guidelines for submitting patches
=================================

Please submit any patches against meta-intel BSPs to the meta-intel
mailing list (meta-intel@yoctoproject.org).  Also, if your patches are
available via a public git repository, please also include a URL to
the repo and branch containing your patches as that makes it easier
for maintainers to grab and test your patches.

There are patch submission scripts available that will, among other
things, automatically include the repo URL and branch as mentioned.
Please see the Yocto Project Development Manual sections entitled
'Using Scripts to Push a Change Upstream and Request a Pull' and
'Using Email to Submit a Patch' for details.

Regardless of how you submit a patch or patchset, the patches should
at minimum follow the suggestions outlined in the 'How to Submit a
Change' secion in the Yocto Project Development Manual.  Specifically,
they should:

  - Include a 'Signed-off-by:' line.  A commit can't legally be pulled
    in without this.

  - Provide a single-line, short summary of the change.  This short
    description should be prefixed by the BSP or recipe name, as
    appropriate, followed by a colon.  Capitalize the first character
    of the summary (following the colon).

  - For the body of the commit message, provide detailed information
    that describes what you changed, why you made the change, and the
    approach you used.

  - If the change addresses a specific bug or issue that is associated
    with a bug-tracking ID, include a reference to that ID in your
    detailed description in the following format: [YOCTO #<bug-id>].

  - Pay attention to line length - please don't allow any particular
    line in the commit message to stretch past 72 characters.

  - For any non-trivial patch, provide information about how you
    tested the patch, and for any non-trivial or non-obvious testing
    setup, provide details of that setup.

Doing a quick 'git log' in meta-intel will provide you with many
examples of good example commits if you have questions about any
aspect of the preferred format.

The meta-intel maintainers will do their best to review and/or pull in
a patch or patchset within 24 hours of the time it was posted.  For
larger and/or more involved patches and patchsets, the review process
may take longer.


Intel-specific machine features
===============================

The meta-intel layer makes some additional machine features available
to BSPs.  These machine features can be used in a BSP layer in the
same way that machine features are used in other layers based on
oe-core, via the MACHINE_FEATURES variable.

Requirements
------------

The meta-intel-specific machine features are only available to a BSP
when the meta-intel layer is included in the build configuration, and
the meta-intel.inc file is included in the machine configuration of
that BSP.

To make these features available for your machine, you will need to:

  1. include a configuration line such as the below in bblayers.conf
	BBLAYERS += "<local path>/meta-intel"
  2. include the following line in the machine configuration file
	require conf/machine/include/meta-intel.inc

Once the above requirements are met, the machine features provided by
the meta-intel layer will be available for the BSP to use.

Building for Intel Quark X1000 microprocessor
---------------------------------------------

To target the Intel Quark X1000. 

 1. In conf/local.conf set the MACHINE type to be intel-quark.

    MACHINE ??= "intel-quark"

 2. Build a target image of your choice. 
    
    $ bitbake core-image-minimal

 3. Use the provided wic script to create an SD card image. 

    $ wic list images
  mkgalileodisk                         Create an Galileo Gen 1/2 disk image
  mkgummidisk                           Create an EFI disk image
    ...

    $ wic create mkgalileodisk -e core-image-minimal

  4. Write the output image to an SD Card

    $ sudo dd if=~/mkgalileodisk-*-mmcblk0.direct of=/dev/mmcblk0
 
  5. Insert the SD Card into the reference platform and power on. 

Available machine features
--------------------------

Currently, the meta-intel layer makes the following set of
Intel-specific machine features available:

  * intel-ucode

These machine features can be included by listing them in the
MACHINE_FEATURES variable in the machine configuration file.  For
example:

	MACHINE_FEATURES += "intel-ucode"

Machine feature details
-----------------------

 * intel-ucode

    This feature provides support for microcode updates to Intel
    processors.  The intel-ucode feature runs at early boot and uses
    the microcode data file added by the feature into the BSP's
    initrd.  It also puts the userland microcode-updating tool,
    iucode_tool, into the target images along with the microcode data
    file.

    Q. Why might a user want to enable the intel-ucode feature?

    A. Intel releases microcode updates to correct processor behavior
       as documented in the respective processor specification
       updates.  While the normal approach to getting such microcode
       updates is via a BIOS upgrade, this can be an administrative
       hassle and not always possible in the field.  The intel-ucode
       feature enables the microcode update capability present in the
       Linux kernel.  It provides an easy path for upgrading processor
       microcode without the need to change the BIOS.  If the feature
       is enabled, it is also possible to update the existing target
       images with a newer microcode update in the future.

    Q. How would a user bundle only target-specific microcode in the
       target image?

    A. The Intel microcode data file released by Intel contains
       microcode updates for multiple processors.  If the BSP image is
       meant to run on only a certain subset of processor types, a
       processor-specific subset of microcode can be bundled into the
       target image via the UCODE_FILTER_PARAMETERS variable.  This
       works by listing a sequence of iucode-tool parameters in the
       UCODE_FILTER_PARAMETERS variable, which in this case will
       select only the specific microcode relevant to the BSP.  For
       more information on the underlying parameters refer to the
       iucode-tool manual page at http://manned.org/iucode-tool

       To define a set of parameters for microcode-filtering via the
       UCODE_FILTER_PARAMETERS variable, one needs to identify the
       cpuid signatures of all the processors the BSP is meant to run
       on.  One way to determine the cpuid signature for a specific
       processor is to build and run an intel-ucode-feature-enabled
       image on the target hardware, without first assigning any value
       to the UCODE_FILTER_PARAMETERS variable, and then once the
       image is booted, run the "ucode_tool -S" command to have the
       ucode tool scan the system for processor signatures.  These
       signatures can then be used in the UCODE_FILTER_PARAMETERS
       variable in conjunction with -s parameter.  For example, for
       the fri2 BSP, the cpuid can be determined as such:

         [root@fri2 ~]# iucode_tool -S
	 iucode_tool: system has processor(s) with signature 0x00020661

       Given that output, a suitable UCODE_FILTER_PARAMETERS variable
       definition could be specified in the machine configuration as
       such:

         UCODE_FILTER_PARAMETERS = "-s 0x00020661"

    Q. Are there any reasons a user might want to disable the
       intel-ucode feature?

    A. The microcode data file and associated tools occupy a small
       amount of space (a few KB) on the target image.  BSPs which are
       highly sensitive to target image size and which are not
       experiencing microcode-related issues might consider not
       enabling this feature.