Linux Kernel and Driver Development Training

Linux Kernel and Driver Development Training
Linux Kernel and Driver
Development Training
Free Electrons
©Copyright 2004-2015, Free Electrons.
Creative Commons BY-SA 3.0 license.
Latest update: May 19, 2015.
Document updates and sources:
http://free-electrons.com/doc/training/linux-kernel
Corrections, suggestions, contributions and translations are welcome!
Send them to feedback@free-electrons.com
Embedded Linux
Developers
Free Electrons
Free Electrons - Embedded Linux, kernel, drivers and Android - Development, consulting, training and support. http://free-electrons.com 1/477

Rights to copy
© Copyright 2004-2015, Free Electrons
License: Creative Commons Attribution - Share Alike 3.0
http://creativecommons.org/licenses/by-sa/3.0/legalcode
You are free:
to copy, distribute, display, and perform the work
to make derivative works
to make commercial use of the work
Under the following conditions:
Attribution. You must give the original author credit.
Share Alike. If you alter, transform, or build upon this work, you may distribute
the resulting work only under a license identical to this one.
For any reuse or distribution, you must make clear to others the license terms of
this work.
Any of these conditions can be waived if you get permission from the copyright
holder.
Your fair use and other rights are in no way aﬀected by the above.

Hyperlinks in the document
There are many hyperlinks in the document
Regular hyperlinks:
http://kernel.org/
Kernel documentation links:
Documentation/kmemcheck.txt
Links to kernel source ﬁles and directories:
drivers/input
include/linux/fb.h
Links to the declarations, deﬁnitions and instances of kernel
symbols (functions, types, data, structures):
platform_get_irq()
GFP_KERNEL
struct file_operations

Free Electrons at a glance
Engineering company created in 2004
(not a training company!)
Locations: Orange, Toulouse, Lyon (France)
Serving customers all around the world
See http://free-electrons.com/company/customers/
Head count: 9
Only Free Software enthusiasts!
Focus: Embedded Linux, Linux kernel, Android Free Software
/ Open Source for embedded and real-time systems.
Activities: development, training, consulting, technical
support.
Added value: get the best of the user and development
community and the resources it oﬀers.

Free Electrons on-line resources
All our training materials:
http://free-electrons.com/docs/
Technical blog:
http://free-electrons.com/blog/
Quarterly newsletter:
http://lists.free-
electrons.com/mailman/listinfo/newsletter
News and discussions (Google +):
https://plus.google.com/+FreeElectronsDevelopers
News and discussions (LinkedIn):
http://linkedin.com/groups/Free-Electrons-4501089
Quick news (Twitter):
http://twitter.com/free_electrons
Linux Cross Reference - browse Linux kernel sources on-line:
http://lxr.free-electrons.com

Generic course information
Generic course
information
Free Electrons
Embedded Linux
Developers
Free Electrons

Hardware used in this training session
BeagleBone Black, from CircuitCo
Texas Instruments AM335x (ARM Cortex-A8)
Powerful CPU, with 3D acceleration,
additional processors (PRUs) and lots of
peripherals.
512 MB of RAM
2 GB of on-board eMMC storage
(4 GB in Rev C)
USB host and USB device ports
microSD slot
HDMI port
2 x 46 pins headers, with access to many
expansion buses (I2C, SPI, UART and more)
A huge number of expansion boards, called
capes. See http://beagleboardtoys.com/.

Do not damage your BeagleBone Black!
Do not remove power abruptly:
Boards components have been damaged by removing the
power or USB cable in an abrupt way, not leaving the PMIC
the time to switch off the components in a clean way. See
http://bit.ly/1FWHNZi
Reboot (reboot) or shutdown (halt) the board in software
when Linux is running.
You can also press the RESET button to reset and reboot.
When there is no software way, you can also switch off the
board by pressing the POWER button for 8 seconds.
Do not leave your board powered on a metallic surface (like a
laptop with a metal finish).

Shopping list: hardware for this course
BeagleBone Black - Multiple distributors:
See http://beagleboard.org/Products/
Nintendo Nunchuck with UEXT connector:
Olimex: http://j.mp/1dTYLfs
Breadboard jumper wires - Male ends:
Olimex: http://j.mp/IUaBsr
USB Serial Cable - Male ends:
Olimex: http://j.mp/1eUuY2K
USB Serial Cable - Female ends:
Olimex: http://j.mp/18Hk8yF
Note that both USB serial cables are the same.
Only the gender of their connector changes.

Participate!
During the lectures...
Don’t hesitate to ask questions. Other people in the audience
may have similar questions too.
This helps the trainer to detect any explanation that wasn’t
clear or detailed enough.
Don’t hesitate to share your experience, for example to
compare Linux / Android with other operating systems used
in your company.
Your point of view is most valuable, because it can be similar
to your colleagues’ and diﬀerent from the trainer’s.
Your participation can make our session more interactive and
make the topics easier to learn.

Practical lab guidelines
During practical labs...
We cannot support more than 8 workstations at once (each
with its board and equipment). Having more would make the
whole class progress slower, compromising the coverage of the
whole training agenda (exception for public sessions: up to 10
people).
So, if you are more than 8 participants, please form up to 8
working groups.
Open the electronic copy of your lecture materials, and use it
throughout the practical labs to ﬁnd the slides you need again.
Don’t copy and paste from the PDF slides.
The slides contain UTF-8 characters that look the same as
ASCII ones, but won’t be understood by shells or compilers.

Cooperate!
As in the Free Software and Open Source community, cooperation
during practical labs is valuable in this training session:
If you complete your labs before other people, don’t hesitate
to help other people and investigate the issues they face. The
faster we progress as a group, the more time we have to
explore extra topics.
Explain what you understood to other participants when
needed. It also helps to consolidate your knowledge.
Don’t hesitate to report potential bugs to your instructor.
Don’t hesitate to look for solutions on the Internet as well.

Command memento sheet
This memento sheet gives
command examples for the most
typical needs (looking for ﬁles,
extracting a tar archive...)
It saves us 1 day of UNIX / Linux
command line training.
Our best tip: in the command line
shell, always hit the Tab key to
complete command names and ﬁle
paths. This avoids 95% of typing
mistakes.
Get an electronic copy on
http://free-electrons.com/
doc/training/embedded-
linux/command_memento.pdf

vi basic commands
The vi editor is very useful to
make quick changes to files in an
embedded target.
Though not very user friendly at
first, vi is very powerful and its
main 15 commands are easy to
learn and are sufficient for 99% of
everyone’s needs!
Get an electronic copy on
doc/training/embedded-
linux/vi_memento.pdf
You can also take the quick tutorial
by running vimtutor. This is a
worthy investment!

Practical lab - Training Setup
Prepare your lab environment
Download the lab archive
Enforce correct permissions

Linux Kernel Introduction
Linux Kernel
Introduction
Free Electrons
Embedded Linux
Developers
Free Electrons

Linux Kernel Introduction
Linux features

History
The Linux kernel is one component of a system, which also
requires libraries and applications to provide features to end
users.
The Linux kernel was created as a hobby in 1991 by a Finnish
student, Linus Torvalds.
Linux quickly started to be used as the kernel for free software
operating systems
Linus Torvalds has been able to create a large and dynamic
developer and user community around Linux.
Nowadays, more than one thousand people contribute to each
kernel release, individuals or companies big and small.

Linux kernel key features
Portability and hardware
support. Runs on most
architectures.
Scalability. Can run on
super computers as well as
on tiny devices (4 MB of
RAM is enough).
Compliance to standards
and interoperability.
Exhaustive networking
support.
Security. It can’t hide its
ﬂaws. Its code is reviewed
by many experts.
Stability and reliability.
Modularity. Can include
only what a system needs
even at run time.
Easy to program. You can
learn from existing code.
Many useful resources on
the net.

Linux kernel in the system

Linux kernel main roles
Manage all the hardware resources: CPU, memory, I/O.
Provide a set of portable, architecture and hardware
independent APIs to allow user space applications and
libraries to use the hardware resources.
Handle concurrent accesses and usage of hardware
resources from diﬀerent applications.
Example: a single network interface is used by multiple user
space applications through various network connections. The
kernel is responsible to “multiplex” the hardware resource.

System calls
The main interface between the kernel and user space is the
set of system calls
About 300 system calls that provide the main kernel services
File and device operations, networking operations,
inter-process communication, process management, memory
mapping, timers, threads, synchronization primitives, etc.
This interface is stable over time: only new system calls can
be added by the kernel developers
This system call interface is wrapped by the C library, and
user space applications usually never make a system call
directly but rather use the corresponding C library function

Pseudo filesystems
Linux makes system and kernel information available in user
space through pseudo filesystems, sometimes also called
virtual filesystems
Pseudo filesystems allow applications to see directories and
files that do not exist on any real storage: they are created
and updated on the fly by the kernel
The two most important pseudo filesystems are
proc, usually mounted on /proc:
Operating system related information (processes, memory
management parameters...)
sysfs, usually mounted on /sys:
Representation of the system as a set of devices and buses.
Information about these devices.

Inside the Linux kernel

Supported hardware architectures
See the arch/ directory in the kernel sources
Minimum: 32 bit processors, with or without MMU, and gcc
support
32 bit architectures (arch/ subdirectories)
Examples: arm, avr32, blackfin, c6x, m68k, microblaze,
mips, score, sparc, um
64 bit architectures:
Examples: alpha, arm64, ia64, tile
32/64 bit architectures
Examples: powerpc, x86, sh, sparc
Find details in kernel sources: arch/<arch>/Kconfig,
arch/<arch>/README, or Documentation/<arch>/

Embedded Linux Kernel Usage
Embedded Linux
Kernel Usage
Free Electrons
Embedded Linux
Developers
Free Electrons

Embedded Linux Kernel Usage
Linux kernel sources

Location of kernel sources
The official versions of the Linux kernel, as released by Linus
Torvalds, are available at http://www.kernel.org
These versions follow the development model of the kernel
However, they may not contain the latest development from a
specific area yet. Some features in development might not be
ready for mainline inclusion yet.
Many chip vendors supply their own kernel sources
Focusing on hardware support first
Can have a very important delta with mainline Linux
Useful only when mainline hasn’t caught up yet.
Many kernel sub-communities maintain their own kernel, with
usually newer but less stable features
Architecture communities (ARM, MIPS, PowerPC, etc.),
device drivers communities (I2C, SPI, USB, PCI, network,
etc.), other communities (real-time, etc.)
No official releases, only development trees are available.

Getting Linux sources
The kernel sources are available from
http://kernel.org/pub/linux/kernel as full tarballs
(complete kernel sources) and patches (diﬀerences between
two kernel versions).
However, more and more people use the git version control
system. Absolutely needed for kernel development!
Fetch the entire kernel sources and history
git clone git://git.kernel.org/pub/scm/linux/
kernel/git/torvalds/linux.git
Create a branch that starts at a speciﬁc stable version
git checkout -b <name-of-branch> v3.11
Web interface available at http://git.kernel.org/cgit/
linux/kernel/git/torvalds/linux.git/tree/.
Read more about Git at http://git-scm.com/

Linux kernel size (1)
Linux 3.10 sources:
Raw size: 573 MB (43,000 files, approx 15,800,000 lines)
gzip compressed tar archive: 105 MB
bzip2 compressed tar archive: 83 MB (better)
xz compressed tar archive: 69 MB (best)
Minimum Linux 3.17 compiled kernel size, booting on the
ARM Versatile board (hard drive on PCI, ext2 filesystem, ELF
executable support, framebuffer console and input devices):
876 KB (compressed), 2.3 MB (raw)
Why are these sources so big?
Because they include thousands of device drivers, many
network protocols, support many architectures and
filesystems...
The Linux core (scheduler, memory management...) is pretty
small!

Linux kernel size (2)
As of kernel version 3.10.
drivers/: 49.4%
arch/: 21.9%
fs/: 6.0%
include/: 4.7%
sound/: 4.4%
Documentation/: 4.0%
net/: 3.9%
firmware/: 1.0%
kernel/: 1.0%
tools/: 0.9%
scripts/: 0.5%
mm/: 0.5%
crypto/: 0.4%
security/: 0.4%
lib/: 0.4%
block/: 0.2%
...

Practical lab - Get Linux Kernel Source Code
Clone the mainline Linux source
tree with git

Kernel Source Code
Kernel Source
Code
Free Electrons
Embedded Linux
Developers
Free Electrons

Kernel Source Code
Linux Code and Device Drivers

Programming language
Implemented in C like all Unix systems. (C was created to
implement the ﬁrst Unix systems)
A little Assembly is used too:
CPU and machine initialization, exceptions
Critical library routines.
No C++ used, see http://www.tux.org/lkml/#s15-3
All the code compiled with gcc
Many gcc speciﬁc extensions used in the kernel code, any
ANSI C compiler will not compile the kernel
A few alternate compilers are supported (Intel and Marvell)
See http://gcc.gnu.org/onlinedocs/gcc-4.9.0/gcc/C-
Extensions.html

No C library
The kernel has to be standalone and can’t use user space
code.
User space is implemented on top of kernel services, not the
opposite.
Kernel code has to supply its own library implementations
(string utilities, cryptography, uncompression ...)
So, you can’t use standard C library functions in kernel code.
(printf(), memset(), malloc(),...).
Fortunately, the kernel provides similar C functions for your
convenience, like printk(), memset(), kmalloc(), ...

Portability
The Linux kernel code is designed to be portable
All code outside arch/ should be portable
To this aim, the kernel provides macros and functions to
abstract the architecture speciﬁc details
Endianness
cpu_to_be32()
cpu_to_le32()
be32_to_cpu()
le32_to_cpu()
I/O memory access
Memory barriers to provide ordering guarantees if needed
DMA API to ﬂush and invalidate caches if needed

No floating point computation
Never use floating point numbers in kernel code. Your code
may be run on a processor without a floating point unit (like
on certain ARM CPUs).
Don’t be confused with floating point related configuration
options
They are related to the emulation of floating point operation
performed by the user space applications, triggering an
exception into the kernel.
Using soft-float, i.e. emulation in user space, is however
recommended for performance reasons.

No stable Linux internal API
The internal kernel API to implement kernel code can undergo
changes between two releases.
In-tree drivers are updated by the developer proposing the API
change: works great for mainline code.
An out-of-tree driver compiled for a given version may no
longer compile or work on a more recent one.
See Documentation/stable_api_nonsense.txt in kernel
sources for reasons why.
Of course, the kernel to userspace API does not change
(system calls, /proc, /sys), as it would break existing
programs.

Kernel memory constraints
No memory protection
Accessing illegal memory locations result in (often fatal)
kernel oopses.
Fixed size stack (8 or 4 KB). Unlike in user space, there’s no
way to make it grow.
Kernel memory can’t be swapped out (for the same reasons).

Linux kernel licensing constraints
The Linux kernel is licensed under the GNU General Public
License version 2
This license gives you the right to use, study, modify and share
the software freely
However, when the software is redistributed, either modified
or unmodified, the GPL requires that you redistribute the
software under the same license, with the source code
If modifications are made to the Linux kernel (for example to
adapt it to your hardware), it is a derivative work of the kernel,
and therefore must be released under GPLv2
The validity of the GPL on this point has already been verified
in courts
However, you’re only required to do so
At the time the device starts to be distributed
To your customers, not to the entire world

Proprietary code and the kernel
It is illegal to distribute a binary kernel that includes statically
compiled proprietary drivers
The kernel modules are a gray area: are they derived works of
the kernel or not?
The general opinion of the kernel community is that
proprietary drivers are bad: http://j.mp/fbyuuH
From a legal point of view, each driver is probably a diﬀerent
case
Is it really useful to keep your drivers secret?
There are some examples of proprietary drivers, like the Nvidia
graphics drivers
They use a wrapper between the driver and the kernel
Unclear whether it makes it legal or not

Advantages of GPL drivers
You don’t have to write your driver from scratch. You can
reuse code from similar free software drivers.
You could get free community contributions, support, code
review and testing, though this generally only happens with
code submitted for the mainline kernel.
Your drivers can be freely and easily shipped by others (for
example by Linux distributions or embedded Linux build
systems).
Pre-compiled drivers work with only one kernel version and
one specific configuration, making life difficult for users who
want to change the kernel version.
Legal certainty, you are sure that a GPL driver is fine from a
legal point of view.

Advantages of in-tree kernel drivers
Once your sources are accepted in the mainline tree, they are
maintained by people making changes.
Near cost-free maintenance, security ﬁxes and improvements.
Easy access to your sources by users.
Many more people reviewing your code.

User space device drivers 1/3
In some cases, it is possible to implement device drivers in
user space!
Can be used when
The kernel provides a mechanism that allows userspace
applications to directly access the hardware.
There is no need to leverage an existing kernel subsystem such
as the networking stack or ﬁlesystems.
There is no need for the kernel to act as a “multiplexer” for
the device: only one application accesses the device.

Possibilities for userspace device drivers:
USB with libusb, http://www.libusb.org/
SPI with spidev, Documentation/spi/spidev
I2C with i2cdev, Documentation/i2c/dev-interface
Memory-mapped devices with UIO, including interrupt
handling, Documentation/DocBook/uio-howto/
Certain classes of devices (printers, scanners, 2D/3D graphics
acceleration) are typically handled partly in kernel space,
partly in user space.

Advantages
No need for kernel coding skills. Easier to reuse code between
devices.
Drivers can be written in any language, even Perl!
Drivers can be kept proprietary.
Driver code can be killed and debugged. Cannot crash the
kernel.
Can be swapped out (kernel code cannot be).
Can use ﬂoating-point computation.
Less in-kernel complexity.
Potentially higher performance, especially for memory-mapped
devices, thanks to the avoidance of system calls.
Drawbacks
Less straightforward to handle interrupts.
Increased interrupt latency vs. kernel code.

Kernel Source Code
Linux sources

Linux sources structure 1/5
arch/<ARCH>
Architecture specific code
arch/<ARCH>/mach-<machine>, machine/board specific code
arch/<ARCH>/include/asm, architecture-specific headers
arch/<ARCH>/boot/dts, Device Tree source files, for some
architectures
block/
Block layer core
COPYING
Linux copying conditions (GNU GPL)
CREDITS
Linux main contributors
crypto/
Cryptographic libraries

Documentation/
Kernel documentation. Don’t miss it!
drivers/
All device drivers except sound ones (usb, pci...)
firmware/
Legacy: ﬁrmware images extracted from old drivers
fs/
Filesystems (fs/ext3/, etc.)
include/
Kernel headers
include/linux/
Linux kernel core headers
include/uapi/
User space API headers
init/
Linux initialization (including main.c)
ipc/
Code used for process communication

Kbuild
Part of the kernel build system
Kconfig
Top level description file for configuration parameters
kernel/
Linux kernel core (very small!)
lib/
Misc library routines (zlib, crc32...)
MAINTAINERS
Maintainers of each kernel part. Very useful!
Makefile
Top Linux Makefile (sets arch and version)
mm/
Memory management code (small too!)

net/
Network support code (not drivers)
README
Overview and building instructions
REPORTING-BUGS
Bug report instructions
samples/
Sample code (markers, kprobes, kobjects...)
scripts/
Scripts for internal or external use
security/
Security model implementations (SELinux...)
sound/
Sound support code and drivers
tools/
Code for various user space tools (mostly C)

usr/
Code to generate an initramfs cpio archive
virt/
Virtualization support (KVM)

Kernel Source Code
Kernel source management tools

Cscope
Tool to browse source code (mainly C, but also C++ or Java)
Supports huge projects like the Linux kernel. Typically takes
less than 1 min. to index the whole Linux sources.
In Linux kernel sources, two ways of running it:
cscope -Rk
All files for all architectures at once
make cscope
cscope -d cscope.out
Only files for your current architecture
Allows searching for a symbol, a definition, functions, strings,
files, etc.
Integration with editors like vim and emacs.
Dedicated graphical front-end: KScope
http://cscope.sourceforge.net/

Cscope screenshot
[Tab]: move the cursor between search results and commands
[Ctrl] [D]: exit cscope

LXR: Linux Cross Reference
Generic source indexing tool and code browser
Web server based, very easy and fast to use
Very easy to find the declaration, implementation or usage of
symbols
Supports C and C++
Supports huge code projects such as the Linux kernel (431
MB of source code in version 3.0).
Takes a little time and patience to setup (configuration,
indexing, web server configuration)
You don’t need to set up LXR by yourself. Use our
http://lxr.free-electrons.com server!
http://sourceforge.net/projects/lxr

LXR screenshot

Practical lab - Kernel Source Code - Exploring
Explore kernel sources manually
Use automated tools to explore the
source code

Kernel Source Code
Kernel conﬁguration

Kernel configuration and build system
The kernel configuration and build system is based on
multiple Makefiles
One only interacts with the main Makefile, present at the
top directory of the kernel source tree
Interaction takes place
using the make tool, which parses the Makefile
through various targets, defining which action should be done
(configuration, compilation, installation, etc.). Run
make help to see all available targets.
Example
cd linux-3.6.x/
make <target>

Kernel configuration (1)
The kernel contains thousands of device drivers, filesystem
drivers, network protocols and other configurable items
Thousands of options are available, that are used to
selectively compile parts of the kernel source code
The kernel configuration is the process of defining the set of
options with which you want your kernel to be compiled
The set of options depends
On your hardware (for device drivers, etc.)
On the capabilities you would like to give to your kernel
(network capabilities, filesystems, real-time, etc.)

Kernel configuration (2)
The configuration is stored in the .config file at the root of
kernel sources
Simple text file, key=value style
As options have dependencies, typically never edited by hand,
but through graphical or text interfaces:
make xconfig, make gconfig (graphical)
make menuconfig, make nconfig (text)
You can switch from one to another, they all load/save the
same .config file, and show the same set of options
To modify a kernel in a GNU/Linux distribution: the
configuration files are usually released in /boot/, together
with kernel images: /boot/config-3.2.0-31-generic

Kernel or module?
The kernel image is a single file, resulting from the linking
of all object files that correspond to features enabled in the
configuration
This is the file that gets loaded in memory by the bootloader
All included features are therefore available as soon as the
kernel starts, at a time where no filesystem exists
Some features (device drivers, filesystems, etc.) can however
be compiled as modules
These are plugins that can be loaded/unloaded dynamically to
add/remove features to the kernel
Each module is stored as a separate file in the filesystem,
and therefore access to a filesystem is mandatory to use
modules
This is not possible in the early boot procedure of the kernel,
because no filesystem is available

Kernel option types
There are diﬀerent types of options
bool options, they are either
true (to include the feature in the kernel) or
false (to exclude the feature from the kernel)
tristate options, they are either
true (to include the feature in the kernel image) or
module (to include the feature as a kernel module) or
false (to exclude the feature)
int options, to specify integer values
hex options, to specify hexadecimal values
string options, to specify string values

Kernel option dependencies
There are dependencies between kernel options
For example, enabling a network driver requires the network
stack to be enabled
Two types of dependencies
depends on dependencies. In this case, option A that depends
on option B is not visible until option B is enabled
select dependencies. In this case, with option A depending
on option B, when option A is enabled, option B is
automatically enabled
make xconfig allows to see all options, even the ones that
cannot be selected because of missing dependencies. In this
case, they are displayed in gray

make xconfig
make xconfig
The most common graphical interface to configure the kernel.
Make sure you read
help -> introduction: useful options!
File browser: easier to load configuration files
Search interface to look for parameters
Required Debian / Ubuntu packages: libqt4-dev g++
(libqt3-mt-dev for older kernel releases)

make xconﬁg screenshot

make xconﬁg search interface
Looks for a keyword in the parameter name. Allows to select or
unselect found parameters.

Kernel conﬁguration options

Corresponding .config file excerpt
Options are grouped by sections and are prefixed with CONFIG_.
#
# CD-ROM/DVD Filesystems
#
CONFIG_ISO9660_FS=m
CONFIG_JOLIET=y
CONFIG_ZISOFS=y
CONFIG_UDF_FS=y
CONFIG_UDF_NLS=y
#
# DOS/FAT/NT Filesystems
#
# CONFIG_MSDOS_FS is not set
# CONFIG_VFAT_FS is not set
CONFIG_NTFS_FS=m
# CONFIG_NTFS_DEBUG is not set
CONFIG_NTFS_RW=y

make gconﬁg
make gconfig
GTK based graphical
conﬁguration interface.
Functionality similar to that
of make xconfig.
Just lacking a search
functionality.
Required Debian packages:
libglade2-dev

make menuconﬁg
make menuconfig
Useful when no graphics are
available. Pretty convenient
too!
Same interface found in
other tools: BusyBox,
Buildroot...
libncurses-dev

make nconﬁg
make nconfig
A newer, similar text
interface
More user friendly (for
example, easier to access
help information).
libncurses-dev

make oldconfig
make oldconfig
Needed very often!
Useful to upgrade a .config file from an earlier kernel release
Issues warnings for configuration parameters that no longer
exist in the new kernel.
Asks for values for new parameters (while xconfig and
menuconfig silently set default values for new parameters).
If you edit a .config file by hand, it’s strongly recommended to
run make oldconfig afterwards!

Undoing configuration changes
A frequent problem:
After changing several kernel configuration settings, your
kernel no longer works.
If you don’t remember all the changes you made, you can get
back to your previous configuration:
$ cp .config.old .config
All the configuration interfaces of the kernel (xconfig,
menuconfig, oldconfig...) keep this .config.old backup
copy.

Configuration per architecture
The set of configuration options is architecture dependent
Some configuration options are very architecture-specific
Most of the configuration options (global kernel options,
network subsystem, filesystems, most of the device drivers) are
visible in all architectures.
By default, the kernel build system assumes that the kernel is
being built for the host architecture, i.e. native compilation
The architecture is not defined inside the configuration, but at
a higher level
We will see later how to override this behaviour, to allow the
configuration of kernels for a different architecture

Kernel Source Code
Compiling and installing the kernel
for the host system

Kernel compilation
make
in the main kernel source directory
Remember to run multiple jobs in parallel if you have multiple
CPU cores. Example: make -j 4
No need to run as root!
Generates
vmlinux, the raw uncompressed kernel image, in the ELF
format, useful for debugging purposes, but cannot be booted
arch/<arch>/boot/*Image, the final, usually compressed,
kernel image that can be booted
bzImage for x86, zImage for ARM, vmImage.gz for Blackfin,
etc.
arch/<arch>/boot/dts/*.dtb, compiled Device Tree files
(on some architectures)
All kernel modules, spread over the kernel source tree, as .ko
files.

Kernel installation
make install
Does the installation for the host system by default, so needs
to be run as root. Generally not used when compiling for an
embedded system, as it installs files on the development
workstation.
Installs
/boot/vmlinuz-<version>
Compressed kernel image. Same as the one in
arch/<arch>/boot
/boot/System.map-<version>
Stores kernel symbol addresses
/boot/config-<version>
Kernel configuration for this version
Typically re-runs the bootloader configuration utility to take
the new kernel into account.

Module installation
make modules_install
Does the installation for the host system by default, so needs
to be run as root
Installs all modules in /lib/modules/<version>/
kernel/
Module .ko (Kernel Object) ﬁles, in the same directory
structure as in the sources.
modules.alias
Module aliases for module loading utilities. Example line:
alias sound-service-?-0 snd_mixer_oss
modules.dep
Module dependencies
modules.symbols
Tells which module a given symbol belongs to.

Kernel cleanup targets
Clean-up generated files (to force
re-compilation):
make clean
Remove all generated files. Needed when
switching from one architecture to another.
Caution: it also removes your .config file!
make mrproper
Also remove editor backup and patch reject files
(mainly to generate patches):
make distclean

Kernel Source Code
Cross-compiling the kernel

Cross-compiling the kernel
When you compile a Linux kernel for another CPU architecture
Much faster than compiling natively, when the target system
is much slower than your GNU/Linux workstation.
Much easier as development tools for your GNU/Linux
workstation are much easier to find.
To make the difference with a native compiler, cross-compiler
executables are prefixed by the name of the target system,
architecture and sometimes library. Examples:
mips-linux-gcc, the prefix is mips-linux-
arm-linux-gnueabi-gcc, the prefix is arm-linux-gnueabi-

Specifying cross-compilation (1)
The CPU architecture and cross-compiler prefix are defined through
the ARCH and CROSS_COMPILE variables in the toplevel Makefile.
ARCH is the name of the architecture. It is defined by the
name of the subdirectory in arch/ in the kernel sources
Example: arm if you want to compile a kernel for the arm
architecture.
CROSS_COMPILE is the prefix of the cross compilation tools
Example: arm-linux- if your compiler is arm-linux-gcc

Specifying cross-compilation (2)
Two solutions to define ARCH and CROSS_COMPILE:
Pass ARCH and CROSS_COMPILE on the make command line:
make ARCH=arm CROSS_COMPILE=arm-linux- ...
Drawback: it is easy to forget to pass these variables when
you run any make command, causing your build and
configuration to be screwed up.
Define ARCH and CROSS_COMPILE as environment variables:
export ARCH=arm
export CROSS_COMPILE=arm-linux-
Drawback: it only works inside the current shell or terminal.
You could put these settings in a file that you source every
time you start working on the project. If you only work on a
single architecture with always the same toolchain, you could
even put these settings in your ~/.bashrc file to make them
permanent and visible from any terminal.

Predefined configuration files
Default configuration files available, per board or per-CPU
family
They are stored in arch/<arch>/configs/, and are just
minimal .config files
This is the most common way of configuring a kernel for
embedded platforms
Run make help to find if one is available for your platform
To load a default configuration file, just run
make acme_defconfig
This will overwrite your existing .config file!
To create your own default configuration file
make savedefconfig, to create a minimal configuration file
mv defconfig arch/<arch>/configs/myown_defconfig

Configuring the kernel
After loading a default configuration file, you can adjust the
configuration to your needs with the normal xconfig,
gconfig or menuconfig interfaces
You can also start the configuration from scratch without
loading a default configuration file
As the architecture is different from your host architecture
Some options will be different from the native configuration
(processor and architecture specific options, specific drivers,
etc.)
Many options will be identical (filesystems, network protocols,
architecture-independent drivers, etc.)
Make sure you have the support for the right CPU, the right
board and the right device drivers.

Device Tree
Many embedded architectures have a lot of non-discoverable
hardware.
Depending on the architecture, such hardware is either
described using C code directly within the kernel, or using a
special hardware description language in a Device Tree.
ARM, PowerPC, OpenRISC, ARC, Microblaze are examples of
architectures using the Device Tree.
A Device Tree Source, written by kernel developers, is
compiled into a binary Device Tree Blob, passed at boot time
to the kernel.
There is one diﬀerent Device Tree for each board/platform
supported by the kernel, available in
arch/arm/boot/dts/<board>.dtb.
The bootloader must load both the kernel image and the
Device Tree Blob in memory before starting the kernel.

Building and installing the kernel
Run make
Copy the final kernel image to the target storage
can be uImage, zImage, vmlinux, bzImage in
arch/<arch>/boot
copying the Device Tree Blob might be necessary as well, they
are available in arch/<arch>/boot/dts
make install is rarely used in embedded development, as
the kernel image is a single file, easy to handle
It is however possible to customize the make install behaviour
in arch/<arch>/boot/install.sh
make modules_install is used even in embedded
development, as it installs many modules and description files
make INSTALL_MOD_PATH=<dir>/ modules_install
The INSTALL_MOD_PATH variable is needed to install the
modules in the target root filesystem instead of your host root
filesystem.

Booting with U-Boot
Recent versions of U-Boot can boot the zImage binary.
Older versions require a special kernel image format: uImage
uImage is generated from zImage using the mkimage tool. It
is done automatically by the kernel make uImage target.
On some ARM platforms, make uImage requires passing a
LOADADDR environment variable, which indicates at which
physical memory address the kernel will be executed.
In addition to the kernel image, U-Boot can also pass a
Device Tree Blob to the kernel.
The typical boot process is therefore:
1. Load zImage or uImage at address X in memory
2. Load <board>.dtb at address Y in memory
3. Start the kernel with bootz X - Y or bootm X - Y
The - in the middle indicates no initramfs

Kernel command line
In addition to the compile time configuration, the kernel
behaviour can be adjusted with no recompilation using the
kernel command line
The kernel command line is a string that defines various
arguments to the kernel
It is very important for system configuration
root= for the root filesystem (covered later)
console= for the destination of kernel messages
Many more exist. The most important ones are documented in
Documentation/kernel-parameters.txt in kernel sources.
This kernel command line is either
Passed by the bootloader. In U-Boot, the contents of the
bootargs environment variable is automatically passed to the
kernel
Built into the kernel, using the CONFIG_CMDLINE option.

Practical lab - Kernel compiling and booting
1st lab: board and bootloader setup:
Prepare the board and access its
serial port
Conﬁgure its bootloader to use
TFTP
2nd lab: kernel compiling and booting:
Set up a cross-compiling
environment
Cross-compile a kernel for an ARM
target platform
Boot this kernel from a directory
on your workstation, accessed by
the board through NFS

Kernel Source Code
Using kernel modules

Advantages of modules
Modules make it easy to develop drivers without rebooting:
load, test, unload, rebuild, load...
Useful to keep the kernel image size to the minimum
(essential in GNU/Linux distributions for PCs).
Also useful to reduce boot time: you don’t spend time
initializing devices and kernel features that you only need later.
Caution: once loaded, have full control and privileges in the
system. No particular protection. That’s why only the root
user can load and unload modules.

Module dependencies
Some kernel modules can depend on other modules, which
need to be loaded ﬁrst.
Example: the usb-storage module depends on the
scsi_mod, libusual and usbcore modules.
Dependencies are described in
/lib/modules/<kernel-version>/modules.dep
This ﬁle is generated when you run make modules_install.

Kernel log
When a new module is loaded, related information is available in
the kernel log.
The kernel keeps its messages in a circular buﬀer (so that it
doesn’t consume more memory with many messages)
Kernel log messages are available through the dmesg
command (diagnostic message)
Kernel log messages are also displayed in the system console
(console messages can be ﬁltered by level using the loglevel
kernel parameter, or completely disabled with the quiet
parameter).
Note that you can write to the kernel log from user space too:
echo "<n>Debug info" > /dev/kmsg

Module utilities (1)
modinfo <module_name>
modinfo <module_path>.ko
Gets information about a module: parameters, license,
description and dependencies.
Very useful before deciding to load a module or not.
sudo insmod <module_path>.ko
Tries to load the given module. The full path to the module
object ﬁle must be given.

Understanding module loading issues
When loading a module fails, insmod often doesn’t give you
enough details!
Details are often available in the kernel log.
Example:
$ sudo insmod ./intr_monitor.ko
insmod: error inserting './intr_monitor.ko': -1 Device or resource busy
$ dmesg
[17549774.552000] Failed to register handler for irq channel 2

sudo modprobe <module_name>
Most common usage of modprobe: tries to load all the
modules the given module depends on, and then this module.
Lots of other options are available. modprobe automatically
looks in /lib/modules/<version>/ for the object ﬁle
corresponding to the given module name.
lsmod
Displays the list of loaded modules
Compare its output with the contents of /proc/modules!

sudo rmmod <module_name>
Tries to remove the given module.
Will only be allowed if the module is no longer in use (for
example, no more processes opening a device ﬁle)
sudo modprobe -r <module_name>
Tries to remove the given module and all dependent modules
(which are no longer needed after removing the module)

Passing parameters to modules
Find available parameters:
modinfo snd-intel8x0m
Through insmod:
sudo insmod ./snd-intel8x0m.ko index=-2
Through modprobe:
Set parameters in /etc/modprobe.conf or in any ﬁle in
/etc/modprobe.d/:
options snd-intel8x0m index=-2
Through the kernel command line, when the driver is built
statically into the kernel:
snd-intel8x0m.index=-2
snd-intel8x0m is the driver name
index is the driver parameter name
-2 is the driver parameter value

Check module parameter values
How to ﬁnd the current values for the parameters of a loaded
module?
Check /sys/module/<name>/parameters.
There is one ﬁle per parameter, containing the parameter
value.

Useful reading
Linux Kernel in a Nutshell, Dec 2006
By Greg Kroah-Hartman, O’Reilly
http://www.kroah.com/lkn/
A good reference book and guide on
conﬁguring, compiling and managing the
Linux kernel sources.
Freely available on-line!
Great companion to the printed book for
easy electronic searches!
Available as single PDF ﬁle on
community/kernel/lkn/
Our rating: 2 stars

Developing Kernel Modules
Developing Kernel
Modules
Free Electrons
Embedded Linux
Developers
Free Electrons

Hello Module 1/2
/* hello.c */
#include <linux/init.h>
#include <linux/module.h>
#include <linux/kernel.h>
static int __init hello_init(void)
{
pr_alert("Good morrow to this fair assembly.n");
return 0;
}
static void __exit hello_exit(void)
{
pr_alert("Alas, poor world, what treasure hast thou lost!n");
}
module_init(hello_init);
module_exit(hello_exit);
MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("Greeting module");
MODULE_AUTHOR("William Shakespeare");

Hello Module 2/2
__init
removed after initialization (static kernel or module.)
__exit
discarded when module compiled statically into the kernel, or
when module unloading support is not enabled.
Example available on
http://git.free-electrons.com/training-
materials/plain/code/hello/hello.c

Hello Module Explanations
Headers speciﬁc to the Linux kernel: linux/xxx.h
No access to the usual C library, we’re doing kernel
programming
An initialization function
Called when the module is loaded, returns an error code (0 on
success, negative value on failure)
Declared by the module_init() macro: the name of the
function doesn’t matter, even though <modulename>_init()
is a convention.
A cleanup function
Called when the module is unloaded
Declared by the module_exit() macro.
Metadata information declared using MODULE_LICENSE(),
MODULE_DESCRIPTION() and MODULE_AUTHOR()

Symbols Exported to Modules 1/2
From a kernel module, only a limited number of kernel
functions can be called
Functions and variables have to be explicitly exported by the
kernel to be visible to a kernel module
Two macros are used in the kernel to export functions and
variables:
EXPORT_SYMBOL(symbolname), which exports a function or
variable to all modules
EXPORT_SYMBOL_GPL(symbolname), which exports a function
or variable only to GPL modules
A normal driver should not need any non-exported function.

Symbols exported to modules 2/2

Module License
Several usages
Used to restrict the kernel functions that the module can use if
it isn’t a GPL licensed module
Diﬀerence between EXPORT_SYMBOL() and
EXPORT_SYMBOL_GPL()
Used by kernel developers to identify issues coming from
proprietary drivers, which they can’t do anything about
(“Tainted” kernel notice in kernel crashes and oopses).
Useful for users to check that their system is 100% free (check
/proc/sys/kernel/tainted)
Values
GPL compatible (see include/linux/license.h: GPL,
GPL v2, GPL and additional rights, Dual MIT/GPL,
Dual BSD/GPL, Dual MPL/GPL)
Proprietary

Compiling a Module
Two solutions
Out of tree
When the code is outside of the kernel source tree, in a
different directory
Advantage: Might be easier to handle than modifications to
the kernel itself
Drawbacks: Not integrated to the kernel
configuration/compilation process, needs to be built
separately, the driver cannot be built statically
Inside the kernel tree
Well integrated into the kernel configuration/compilation
process
Driver can be built statically if needed

Compiling an out-of-tree Module 1/2
The below Makefile should be reusable for any single-file
out-of-tree Linux module
The source file is hello.c
Just run make to build the hello.ko file
ifneq ($(KERNELRELEASE),)
obj-m := hello.o
else
KDIR := /path/to/kernel/sources
all:
<tab>$(MAKE) -C $(KDIR) M=$$PWD
endif
For KDIR, you can either set:
full kernel source directory
(configured + make modules_prepare)
or just kernel headers directory (make headers_install)

Compiling an out-of-tree Module 2/2
The module Makefile is interpreted with KERNELRELEASE
undefined, so it calls the kernel Makefile, passing the module
directory in the M variable
The kernel Makefile knows how to compile a module, and
thanks to the M variable, knows where the Makefile for our
module is. The module Makefile is interpreted with
KERNELRELEASE defined, so the kernel sees the obj-m
definition.

Modules and Kernel Version
To be compiled, a kernel module needs access to the kernel
headers, containing the definitions of functions, types and
constants.
Two solutions
Full kernel sources
Only kernel headers (linux-headers-* packages in
Debian/Ubuntu distributions)
The sources or headers must be configured
Many macros or functions depend on the configuration
A kernel module compiled against version X of kernel headers
will not load in kernel version Y
modprobe / insmod will say Invalid module format

New Driver in Kernel Sources 1/2
To add a new driver to the kernel sources:
Add your new source file to the appropriate source directory.
Example: drivers/usb/serial/navman.c
Single file drivers in the common case, even if the file is several
thousand lines of code big. Only really big drivers are split in
several files or have their own directory.
Describe the configuration interface for your new driver by
adding the following lines to the Kconfig file in this directory:
config USB_SERIAL_NAVMAN
tristate "USB Navman GPS device"
depends on USB_SERIAL
help
To compile this driver as a module, choose M
here: the module will be called navman.

New Driver in Kernel Sources 2/2
Add a line in the Makefile file based on the Kconfig setting:
obj-$(CONFIG_USB_SERIAL_NAVMAN) += navman.o
It tells the kernel build system to build navman.c when the
USB_SERIAL_NAVMAN option is enabled. It works both if
compiled statically or as a module.
Run make xconfig and see your new options!
Run make and your new files are compiled!
See Documentation/kbuild/ for details and more elaborate
examples like drivers with several source files, or drivers in their
own subdirectory, etc.

Hello Module with Parameters 1/2
/* hello_param.c */
#include <linux/init.h>
#include <linux/module.h>
MODULE_LICENSE("GPL");
/* A couple of parameters that can be passed in: how many
times we say hello, and to whom */
static char *whom = "world";
module_param(whom, charp, 0);
static int howmany = 1;
module_param(howmany, int, 0);

Hello Module with Parameters 2/2
static int __init hello_init(void)
{
int i;
for (i = 0; i < howmany; i++)
pr_alert("(%d) Hello, %sn", i, whom);
return 0;
}
static void __exit hello_exit(void)
{
pr_alert("Goodbye, cruel %sn", whom);
}
module_init(hello_init);
module_exit(hello_exit);
Thanks to Jonathan Corbet for the example!
Source code available on:
http://git.free-electrons.com/training-
materials/plain/code/hello-param/hello_param.c

Declaring a module parameter
module_param(
name, /* name of an already defined variable */
type, /* either byte, short, ushort, int, uint, long, ulong,
charp, bool or invbool. (checked at run time!) */
perm /* for /sys/module/<module_name>/parameters/<param>,
0: no such module parameter value file */
);
/* Example */
static int irq=5;
module_param(irq, int, S_IRUGO);
Modules parameter arrays are also possible with
module_param_array().

Practical lab - Writing Modules
Create, compile and load your ﬁrst
module
Add module parameters
Access kernel internals from your
module

Useful general-purpose kernel APIs
Useful
general-purpose
kernel APIs
Free Electrons
Embedded Linux
Developers
Free Electrons

Memory/string utilities
In linux/string.h
Memory-related: memset(), memcpy(), memmove(),
memscan(), memcmp(), memchr()
String-related: strcpy(), strcat(), strcmp(), strchr(),
strrchr(), strlen() and variants
Allocate and copy a string: kstrdup(), kstrndup()
Allocate and copy a memory area: kmemdup()
In linux/kernel.h
String to int conversion: simple_strtoul(),
simple_strtol(), simple_strtoull(),
simple_strtoll()
Other string functions: sprintf(), sscanf()

Linked lists
Convenient linked-list facility in linux/list.h
Used in thousands of places in the kernel
Add a struct list_head member to the structure whose
instances will be part of the linked list. It is usually named
node when each instance needs to only be part of a single list.
Deﬁne the list with the LIST_HEAD() macro for a global list,
or deﬁne a struct list_head element and initialize it with
INIT_LIST_HEAD() for lists embedded in a structure.
Then use the list_*() API to manipulate the list
Add elements: list_add(), list_add_tail()
Remove, move or replace elements: list_del(),
list_move(), list_move_tail(), list_replace()
Test the list: list_empty()
Iterate over the list: list_for_each_*() family of macros

Linked Lists Examples (1)
From include/linux/atmel_tc.h
/*
* Definition of a list element, with a
* struct list_head member
*/
struct atmel_tc
{
/* some members */
struct list_head node;
};

Linked Lists Examples (2)
From drivers/misc/atmel_tclib.c
/* Define the global list */
static LIST_HEAD(tc_list);
static int __init tc_probe(struct platform_device *pdev) {
struct atmel_tc *tc;
tc = kzalloc(sizeof(struct atmel_tc), GFP_KERNEL);
/* Add an element to the list */
list_add_tail(&tc->node, &tc_list);
}
struct atmel_tc *atmel_tc_alloc(unsigned block, const char *name)
{
struct atmel_tc *tc;
/* Iterate over the list elements */
list_for_each_entry(tc, &tc_list, node) {
/* Do something with tc */
}
[...]
}

Linux device and driver model
Linux device and
driver model
Free Electrons
Embedded Linux
Developers
Free Electrons

Introduction

The need for a device model?
The Linux kernel runs on a wide range of architectures and
hardware platforms, and therefore needs to maximize the
reusability of code between platforms.
For example, we want the same USB device driver to be
usable on a x86 PC, or an ARM platform, even though the
USB controllers used on these platforms are diﬀerent.
This requires a clean organization of the code, with the device
drivers separated from the controller drivers, the hardware
description separated from the drivers themselves, etc.
This is what the Linux kernel Device Model allows, in
addition to other advantages covered in this section.

Kernel and Device Drivers
In Linux, a driver is always
interfacing with:
a framework that allows the
driver to expose the
hardware features in a
generic way.
a bus infrastructure, part
of the device model, to
detect/communicate with
the hardware.
This section focuses on the
device model, while kernel
frameworks are covered later in
this training.

Device Model data structures
The device model is organized around three main data
structures:
The struct bus_type structure, which represent one type of
bus (USB, PCI, I2C, etc.)
The struct device_driver structure, which represents one
driver capable of handling certain devices on a certain bus.
The struct device structure, which represents one device
connected to a bus
The kernel uses inheritance to create more specialized versions
of struct device_driver and struct device for each bus
subsystem.
In order to explore the device model, we will
First look at a popular bus that oﬀers dynamic enumeration,
the USB bus
Continue by studying how buses that do not oﬀer dynamic
enumerations are handled.

Bus Drivers
The first component of the device model is the bus driver
One bus driver for each type of bus: USB, PCI, SPI, MMC,
I2C, etc.
It is responsible for
Registering the bus type (struct bus_type)
Allowing the registration of adapter drivers (USB controllers,
I2C adapters, etc.), able to detect the connected devices, and
providing a communication mechanism with the devices
Allowing the registration of device drivers (USB devices, I2C
devices, PCI devices, etc.), managing the devices
Matching the device drivers against the devices detected by
the adapter drivers.
Provides an API to both adapter drivers and device drivers
Defining driver and device specific structures, mainly
struct usb_driver and struct usb_interface

Example of the USB bus

Example: USB Bus 1/2

Example: USB Bus 2/2
Core infrastructure (bus driver)
drivers/usb/core
struct bus_type is deﬁned in
drivers/usb/core/driver.c and registered in
drivers/usb/core/usb.c
Adapter drivers
drivers/usb/host
For EHCI, UHCI, OHCI, XHCI, and their implementations on
various systems (Atmel, IXP, Xilinx, OMAP, Samsung, PXA,
etc.)
Device drivers
Everywhere in the kernel tree, classiﬁed by their type

Example of Device Driver
To illustrate how drivers are implemented to work with the
device model, we will study the source code of a driver for a
USB network card
It is USB device, so it has to be a USB device driver
It is a network device, so it has to be a network device
Most drivers rely on a bus infrastructure (here, USB) and
register themselves in a framework (here, network)
We will only look at the device driver side, and not the
adapter driver side
The driver we will look at is drivers/net/usb/rtl8150.c

Device Identifiers
Defines the set of devices that this driver can manage, so that
the USB core knows for which devices this driver should be
used
The MODULE_DEVICE_TABLE() macro allows depmod to
extract at compile time the relation between device identifiers
and drivers, so that drivers can be loaded automatically by
udev. See
/lib/modules/$(uname -r)/modules.{alias,usbmap}
static struct usb_device_id rtl8150_table[] = {
{ USB_DEVICE(VENDOR_ID_REALTEK, PRODUCT_ID_RTL8150) },
{ USB_DEVICE(VENDOR_ID_MELCO, PRODUCT_ID_LUAKTX) },
{ USB_DEVICE(VENDOR_ID_MICRONET, PRODUCT_ID_SP128AR) },
{ USB_DEVICE(VENDOR_ID_LONGSHINE, PRODUCT_ID_LCS8138TX) },
[...]
{}
};
MODULE_DEVICE_TABLE(usb, rtl8150_table);

Instantiation of usb driver
struct usb_driver is a structure deﬁned by the USB core.
Each USB device driver must instantiate it, and register itself
to the USB core using this structure
This structure inherits from struct device_driver, which
is deﬁned by the device model.
static struct usb_driver rtl8150_driver = {
.name = "rtl8150",
.probe = rtl8150_probe,
.disconnect = rtl8150_disconnect,
.id_table = rtl8150_table,
.suspend = rtl8150_suspend,
.resume = rtl8150_resume
};

Driver (Un)Registration
When the driver is loaded or unloaded, it must register or
unregister itself from the USB core
Done using usb_register() and usb_deregister(),
provided by the USB core.
static int __init usb_rtl8150_init(void)
{
return usb_register(&rtl8150_driver);
}
static void __exit usb_rtl8150_exit(void)
{
usb_deregister(&rtl8150_driver);
}
module_init(usb_rtl8150_init);
module_exit(usb_rtl8150_exit);
Note: this code has now been replaced by a shorter
module_usb_driver() macro call.

At Initialization
The USB adapter driver that corresponds to the USB
controller of the system registers itself to the USB core
The rtl8150 USB device driver registers itself to the USB
core
The USB core now knows the association between the
vendor/product IDs of rtl8150 and the struct usb_driver
structure of this driver

When a Device is Detected

Probe Method
The probe() method receives as argument a structure
describing the device, usually specialized by the bus
infrastructure (struct pci_dev, struct usb_interface,
etc.)
This function is responsible for
Initializing the device, mapping I/O memory, registering the
interrupt handlers. The bus infrastructure provides methods to
get the addresses, interrupt numbers and other device-speciﬁc
information.
Registering the device to the proper kernel framework, for
example the network infrastructure.

Probe Method Example
static int rtl8150_probe(struct usb_interface *intf,
const struct usb_device_id *id)
{
rtl8150_t *dev;
struct net_device *netdev;
netdev = alloc_etherdev(sizeof(rtl8150_t));
[...]
dev = netdev_priv(netdev);
tasklet_init(&dev->tl, rx_fixup, (unsigned long)dev);
spin_lock_init(&dev->rx_pool_lock);
[...]
netdev->netdev_ops = &rtl8150_netdev_ops;
alloc_all_urbs(dev);
[...]
usb_set_intfdata(intf, dev);
SET_NETDEV_DEV(netdev, &intf->dev);
register_netdev(netdev);
return 0;
}

The Model is Recursive

Platform drivers

Non-discoverable buses
On embedded systems, devices are often not connected
through a bus allowing enumeration, hotplugging, and
providing unique identiﬁers for devices.
For example, the devices on I2C buses or SPI buses, or the
devices directly part of the system-on-chip.
However, we still want all of these devices to be part of the
device model.
Such devices, instead of being dynamically detected, must be
statically described in either:
The kernel source code
The Device Tree, a hardware description ﬁle used on some
architectures.

Platform devices
Amongst the non-discoverable devices, a huge family are the
devices that are directly part of a system-on-chip: UART
controllers, Ethernet controllers, SPI or I2C controllers,
graphic or audio devices, etc.
In the Linux kernel, a special bus, called the platform bus has
been created to handle such devices.
It supports platform drivers that handle platform devices.
It works like any other bus (USB, PCI), except that devices are
enumerated statically instead of being discovered dynamically.

Implementation of a Platform Driver
The driver implements a struct platform_driver structure
(example taken from drivers/serial/imx.c)
static struct platform_driver serial_imx_driver = {
.probe = serial_imx_probe,
.remove = serial_imx_remove,
.driver = {
.name = "imx-uart",
.owner = THIS_MODULE,
},
};
And registers its driver to the platform driver infrastructure
static int __init imx_serial_init(void) {
ret = platform_driver_register(&serial_imx_driver);
}
static void __exit imx_serial_cleanup(void) {
platform_driver_unregister(&serial_imx_driver);
}

Platform Device Instantiation: old style (1/2)
As platform devices cannot be detected dynamically, they are
defined statically
By direct instantiation of struct platform_device
structures, as done on some ARM platforms. Definition done
in the board-specific or SoC specific code.
By using a device tree, as done on Power PC (and on some
ARM platforms) from which struct platform_device
structures are created
Example on ARM, where the instantiation is done in
arch/arm/mach-imx/mx1ads.c
static struct platform_device imx_uart1_device = {
.name = "imx-uart",
.id = 0,
.num_resources = ARRAY_SIZE(imx_uart1_resources),
.resource = imx_uart1_resources,
.dev = {
.platform_data = &uart_pdata,
}
};

Platform device instantiation: old style (2/2)
The device is part of a list
static struct platform_device *devices[] __initdata = {
&cs89x0_device,
&imx_uart1_device,
&imx_uart2_device,
};
And the list of devices is added to the system during board
initialization
static void __init mx1ads_init(void)
{
[...]
platform_add_devices(devices, ARRAY_SIZE(devices));
}
MACHINE_START(MX1ADS, "Freescale MX1ADS")
[...]
.init_machine = mx1ads_init,
MACHINE_END

The Resource Mechanism
Each device managed by a particular driver typically uses
diﬀerent hardware resources: addresses for the I/O registers,
DMA channels, IRQ lines, etc.
Such information can be represented using
struct resource, and an array of struct resource is
associated to a struct platform_device
Allows a driver to be instantiated for multiple devices
functioning similarly, but with diﬀerent addresses, IRQs, etc.

Declaring resources
static struct resource imx_uart1_resources[] = {
[0] = {
.start = 0x00206000,
.end = 0x002060FF,
.flags = IORESOURCE_MEM,
},
[1] = {
.start = (UART1_MINT_RX),
.end = (UART1_MINT_RX),
.flags = IORESOURCE_IRQ,
},
};

Using Resources
When a struct platform_device is added to the system
using platform_add_device(), the probe() method of the
platform driver gets called
This method is responsible for initializing the hardware,
registering the device to the proper framework (in our case,
the serial driver framework)
The platform driver has access to the I/O resources:
res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
base = ioremap(res->start, PAGE_SIZE);
sport->rxirq = platform_get_irq(pdev, 0);

platform data Mechanism
In addition to the well-defined resources, many drivers require
driver-specific information for each platform device
Such information can be passed using the platform_data
field of struct device (from which
struct platform_device inherits)
As it is a void * pointer, it can be used to pass any type of
information.
Typically, each driver defines a structure to pass information
through struct platform_data

platform data example 1/2
The i.MX serial port driver deﬁnes the following structure to
be passed through struct platform_data
struct imxuart_platform_data {
int (*init)(struct platform_device *pdev);
void (*exit)(struct platform_device *pdev);
unsigned int flags;
void (*irda_enable)(int enable);
unsigned int irda_inv_rx:1;
unsigned int irda_inv_tx:1;
unsigned short transceiver_delay;
};
The MX1ADS board code instantiates such a structure
static struct imxuart_platform_data uart1_pdata = {
.flags = IMXUART_HAVE_RTSCTS,
};

platform data Example 2/2
The uart_pdata structure is associated to the
struct platform_device structure in the MX1ADS board
ﬁle (the real code is slightly more complicated)
struct platform_device mx1ads_uart1 = {
.name = "imx-uart",
.dev {
.platform_data = &uart1_pdata,
},
.resource = imx_uart1_resources,
[...]
};
The driver can access the platform data:
static int serial_imx_probe(struct platform_device *pdev)
{
struct imxuart_platform_data *pdata;
pdata = pdev->dev.platform_data;
if (pdata && (pdata->flags & IMXUART_HAVE_RTSCTS))
sport->have_rtscts = 1;
[...]

Device Tree
On many embedded architectures, manual instantiation of
platform devices was considered to be too verbose and not
easily maintainable.
Such architectures are moving, or have moved, to use the
Device Tree.
It is a tree of nodes that models the hierarchy of devices in
the system, from the devices inside the processor to the
devices on the board.
Each node can have a number of properties describing various
properties of the devices: addresses, interrupts, clocks, etc.
At boot time, the kernel is given a compiled version, the
Device Tree Blob, which is parsed to instantiate all the
devices described in the DT.
On ARM, they are located in arch/arm/boot/dts.

Device Tree example
uart0: serial@44e09000 {
compatible = "ti,omap3-uart";
ti,hwmods = "uart1";
clock-frequency = <48000000>;
reg = <0x44e09000 0x2000>;
interrupts = <72>;
status = "disabled";
};
serial@44e09000 is the node name
uart0 is an alias, that can be referred to in other parts of the
DT as &uart0
other lines are properties. Their values are usually strings, list
of integers, or references to other nodes.

Device Tree inheritance (1/2)
Each particular hardware platform has its own device tree.
However, several hardware platforms use the same processor,
and often various processors in the same family share a
number of similarities.
To allow this, a device tree file can include another one. The
trees described by the including file overlays the tree described
by the included file. This can be done:
Either by using the /include/ statement provided by the
Device Tree language.
Either by using the #include statement, which requires
calling the C preprocessor before parsing the Device Tree.
Linux currently uses either one technique or the other,
(different from one ARM subarchitecture to another, for
example).

Device Tree inheritance (2/2)

Device Tree: compatible string
With the device tree, a device is bound with the
corresponding driver using the compatible string.
The of_match_table ﬁeld of struct device_driver lists
the compatible strings supported by the driver.
#if defined(CONFIG_OF)
static const struct of_device_id omap_serial_of_match[] = {
{ .compatible = "ti,omap2-uart" },
{},
};
MODULE_DEVICE_TABLE(of, omap_serial_of_match);
#endif
static struct platform_driver serial_omap_driver = {
.probe = serial_omap_probe,
.remove = serial_omap_remove,
.driver = {
.name = DRIVER_NAME,
.pm = &serial_omap_dev_pm_ops,
.of_match_table = of_match_ptr(omap_serial_of_match),
},
};

Device Tree Resources
The drivers will use the same mechanism that we saw
previously to retrieve basic information: interrupts numbers,
physical addresses, etc.
The available resources list will be built up by the kernel at
boot time from the device tree, so that you don’t need to
make any unnecessary lookups to the DT when loading your
driver.
Any additional information will be speciﬁc to a driver or the
class it belongs to, deﬁning the bindings

Device Tree bindings
The compatible string and the associated properties define
what is called a device tree binding.
Device tree bindings are all documented in
Documentation/devicetree/bindings .
Since the Device Tree is normally part of the kernel ABI, the
bindings must remain compatible over-time.
A new kernel must be capable of using an old Device Tree.
This requires a very careful design of the bindings. They are all
reviewed on the devicetree@vger.kernel.org mailing list.
A Device Tree binding should contain only a description of the
hardware and not configuration.
An interrupt number can be part of the Device Tree as it
describes the hardware.
But not whether DMA should be used for a device or not, as it
is a configuration choice.

sysfs
The bus, device, drivers, etc. structures are internal to the
kernel
The sysfs virtual filesystem offers a mechanism to export
such information to user space
Used for example by udev to provide automatic module
loading, firmware loading, device file creation, etc.
sysfs is usually mounted in /sys
/sys/bus/ contains the list of buses
/sys/devices/ contains the list of devices
/sys/class enumerates devices by class (net, input,
block...), whatever the bus they are connected to. Very useful!
Take your time to explore /sys on your workstation.

References
Device Tree for Dummies, Thomas Petazzoni (Apr. 2014):
http://j.mp/1jQU6NR
Kernel documentation
Documentation/driver-model/
Documentation/devicetree/
Documentation/filesystems/sysfs.txt
The kernel source code
Full of examples of other drivers!

Introduction to the I2C subsystem
Introduction to
the I2C subsystem
Free Electrons
Embedded Linux
Developers
Free Electrons

What is I2C?
A very commonly used low-speed bus to connect on-board
devices to the processor.
Uses only two wires: SDA for the data, SCL for the clock.
It is a master/slave bus: only the master can initiate
transactions, and slaves can only reply to transactions
initiated by masters.
In a Linux system, the I2C controller embedded in the
processor is typically the master, controlling the bus.
Each slave device is identiﬁed by a unique I2C address. Each
transaction initiated by the master contains this address,
which allows the relevant slave to recognize that it should
reply to this particular transaction.

An I2C bus example

The I2C subsystem
Like all bus subsystems, the I2C subsystem is responsible for:
Providing an API to implement I2C controller drivers
Providing an API to implement I2C device drivers, in kernel
space
Providing an API to implement I2C device drivers, in user
space
The core of the I2C subsystem is located in drivers/i2c.
The I2C controller drivers are located in
drivers/i2c/busses.
The I2C device drivers are located throughout drivers/,
depending on the type of device (ex: drivers/input for
input devices).

Registering an I2C device driver
Like all bus subsystems, the I2C subsystem deﬁnes a
struct i2c_driver that inherits from
struct device_driver, and which must be instantiated and
registered by each I2C device driver.
As usual, this structure points to the ->probe() and
->remove() functions.
It also contains an id_table ﬁeld that must point to a list of
device IDs (which is a list of tuples containing a string and
some private driver data). It is used for non-DT based probing
of I2C devices.
The i2c_add_driver() and i2c_del_driver() functions
are used to register/unregister the driver.
If the driver doesn’t do anything else in its init()/exit()
functions, it is advised to use the module_i2c_driver()
macro instead.

Registering an I2C device driver: example
static const struct i2c_device_id <driver>_id[] = {
{ "<device-name>", 0 },
{ }
};
MODULE_DEVICE_TABLE(i2c, <driver>_id);
#ifdef CONFIG_OF
static const struct of_device_id <driver>_dt_ids[] = {
{ .compatible = "<vendor>,<device-name>", },
{ }
};
MODULE_DEVICE_TABLE(of, <driver>_dt_ids);
#endif
static struct i2c_driver <driver>_driver = {
.probe = <driver>_probe,
.remove = <driver>_remove,
.id_table = <driver>_id,
.driver = {
.name = "<driver-name>",
.of_match_table = of_match_ptr(<driver>_dt_ids),
},
};
module_i2c_driver(<driver>_driver);

Registering an I2C device: non-DT
On non-DT platforms, the struct i2c_board_info
structure allows to describe how an I2C device is connected to
a board.
Such structures are normally deﬁned with the
I2C_BOARD_INFO() helper macro.
Takes as argument the device name and the slave address of
the device on the bus.
An array of such structures is registed on a per-bus basis using
i2c_register_board_info(), when the platform is
initialized.

Registering an I2C device, non-DT example
static struct i2c_board_info <board>_i2c_devices[] __initdata = {
{
I2C_BOARD_INFO("cs42l51", 0x4a),
},
};
void board_init(void)
{
/*
* Here should be the registration of all devices, including
* the I2C controller device.
*/
i2c_register_board_info(0, <board>_i2c_devices,
ARRAY_SIZE(<board>_i2c_devices));
/* More devices registered here */
}

Registering an I2C device, in the DT
In the Device Tree, the I2C controller device is typically
defined in the .dtsi file that describes the processor.
Normally defined with status = "disabled".
At the board/platform level:
the I2C controller device is enabled (status = "okay")
the I2C bus frequency is defined, using the clock-frequency
property.
the I2C devices on the bus are described as children of the I2C
controller node, where the reg property gives the I2C slave
address on the bus.

Registering an I2C device, DT example (1/2)
Deﬁnition of the I2C controller, .dtsi ﬁle
i2c@7000c000 {
compatible = "nvidia,tegra20-i2c";
reg = <0x7000c000 0x100>;
interrupts = <GIC_SPI 38 IRQ_TYPE_LEVEL_HIGH>;
#address-cells = <1>;
#size-cells = <0>;
clocks = <&tegra_car TEGRA20_CLK_I2C1>,
<&tegra_car TEGRA20_CLK_PLL_P_OUT3>;
clock-names = "div-clk", "fast-clk";
status = "disabled";
};

Registering an I2C device, DT example (2/2)
Deﬁnition of the I2C device, .dts ﬁle
i2c@7000c000 {
status = "okay";
alc5632: alc5632@1e {
compatible = "realtek,alc5632";
reg = <0x1e>;
gpio-controller;
#gpio-cells = <2>;
};
};

probe() and remove()
The ->probe() function is responsible for initializing the
device and registering it in the appropriate kernel framework.
It receives as argument:
A struct i2c_client pointer, which represents the I2C
device itself. This structure inherits from struct device.
A struct i2c_device_id pointer, which points to the I2C
device ID entry that matched the device that is being probed.
The ->remove() function is responsible for unregistering the
device from the kernel framework and shut it down. It receives
as argument:
The same struct i2c_client pointer that was passed as
argument to ->probe()

Probe/remove example
static int <driver>_probe(struct i2c_client *client,
const struct i2c_device_id *id)
{
/* initialize device */
/* register to a kernel framework */
i2c_set_clientdata(client, <private data>);
return 0;
}
static int <driver>_remove(struct i2c_client *client)
{
<private data> = i2c_get_clientdata(client);
/* unregister device from kernel framework */
/* shut down the device */
return 0;
}

Practical lab - Linux device model for an I2C driver
Modify the Device Tree to
instantiate an I2C device.
Implement a driver that registers as
an I2C driver.
Make sure that the probe/remove
functions are called when there is a
device/driver match.
Explore the sysfs entries related to
your driver and device.

Communicating with the I2C device: raw API
The most basic API to communicate with the I2C device provides
functions to either send or receive data:
int i2c_master_send(struct i2c_client *client,
const char *buf, int count);
Sends the contents of buf to the client.
int i2c_master_recv(struct i2c_client *client,
char *buf, int count);
Receives count bytes from the client, and store them into
buf.

Communicating with the I2C device: message transfer
The message transfer API allows to describe transfers that
consists of several messages, with each message being a
transaction in one direction:
int i2c_transfer(struct i2c_adapter *adap, struct
i2c_msg *msg, int num);
The struct i2c_adapter pointer can be found by using
client->adapter
The struct i2c_msg structure deﬁnes the length, location,
and direction of the message.

I2C: message transfer example
struct i2c_msg msg[2];
int error;
u8 start_reg;
u8 buf[10];
msg[0].addr = client->addr;
msg[0].flags = 0;
msg[0].len = 1;
msg[0].buf = &start_reg;
start_reg = 0x10;
msg[1].addr = client->addr;
msg[1].flags = I2C_M_RD;
msg[1].len = sizeof(buf);
msg[1].buf = buf;
error = i2c_transfer(client->adapter, msg, 2);

SMBus calls
SMBus is a subset of the I2C protocol.
It deﬁnes a standard set of transactions, for example to read
or write a register into a device.
Linux provides SMBus functions that should be used instead
of the raw API, if the I2C device supports this standard type of
transactions. The driver can then be used on both SMBus and
I2C adapters (can’t use I2C commands on SMBus adapters).
Example: the i2c_smbus_read_byte_data() function allows
to read one byte of data from a device register.
It does the following operations:
S Addr Wr [A] Comm [A] S Addr Rd [A] [Data] NA P
Which means it ﬁrst writes a one byte data command
(Comm), and then reads back one byte of data ([Data]).
See Documentation/i2c/smbus-protocol for details.

List of SMBus functions
Read/write one byte
s32 i2c_smbus_read_byte(const struct i2c_client *client);
s32 i2c_smbus_write_byte(const struct i2c_client *client, u8 value);
Write a command byte, and read or write one byte
s32 i2c_smbus_read_byte_data(const struct i2c_client *client, u8 command);
s32 i2c_smbus_write_byte_data(const struct i2c_client *client, u8 command, u8
value);
Write a command byte, and read or write one word
s32 i2c_smbus_read_word_data(const struct i2c_client *client, u8 command);
s32 i2c_smbus_write_word_data(const struct i2c_client *client, u8 command, u16
value);
Write a command byte, and read or write a block of data
(max 32 bytes)
s32 i2c_smbus_read_block_data(const struct i2c_client *client, u8 command, u8
*values);
s32 i2c_smbus_write_block_data(const struct i2c_client *client, u8 command, u8
length, const u8 *values);
Write a command byte, and read or write a block of data (no
limit)
s32 i2c_smbus_read_i2c_block_data(const struct i2c_client *client, u8 command, u8
length, u8 *values);
s32 i2c_smbus_write_i2c_block_data(const struct i2c_client *client, u8 command, u8
length, const u8 *values);

I2C functionality
Not all I2C controllers support all functionalities.
The I2C controller drivers therefore tell the I2C core which
functionalities they support.
An I2C device driver must check that the functionalities they
need are provided by the I2C controller in use on the system.
The i2c_check_functionality() function allows to make
such a check.
Examples of functionalities: I2C_FUNC_I2C to be able to use
the raw I2C functions, I2C_FUNC_SMBUS_BYTE_DATA to be
able to use SMBus commands to write a command and
read/write one byte of data.
See include/uapi/linux/i2c.h for the full list of existing
functionalities.

References
http://en.wikipedia.org/wiki/I2C, general presentation
of the I2C protocol
Documentation/i2c/ , details about the Linux support for
I2C
writing-clients, how to write I2C device drivers
instantiating-devices, how to instantiate devices
smbus-protocol, details on the SMBus functions
functionality, how the functionality mechanism works
and many more documentation ﬁles
http://free-electrons.com/pub/video/2012/elce/
elce-2012-anders-board-bringup-i2c.webm, excellent
talk: You, me and I2C from David Anders at ELCE 2012.

Introduction to pin muxing
Introduction to pin
muxing
Free Electrons
Embedded Linux
Developers
Free Electrons

What is pin muxing?
Modern SoCs (System on Chip) include more and more
hardware blocks, many of which need to interface with the
outside world using pins.
However, the physical size of the chips remains small, and
therefore the number of available pins is limited.
For this reason, not all of the internal hardware block features
can be exposed on the pins simultaneously.
The pins are multiplexed: they expose either the functionality
of hardware block A or the functionality of hardware block B.
This multiplexing is usually software conﬁgurable.

Pin muxing diagram

Pin muxing in the Linux kernel
Since Linux 3.2, a pinctrl subsystem has been added.
This subsystem, located in drivers/pinctrl provides a
generic subsystem to handle pin muxing. It offers:
A pin muxing consumer interface, for device drivers.
A pin muxing driver interface, to implement the
system-on-chip specific drivers that configure the muxing.
Most pinctrl drivers provide a Device Tree binding, and the
pin muxing must be described in the Device Tree.
The exact Device Tree binding depends on each driver. Each
binding is documented in
Documentation/devicetree/bindings/pinctrl .

pinctrl subsystem diagram

Device Tree binding for consumer devices
The devices that require certains pins to be muxed will use the
pinctrl-<x> and pinctrl-names Device Tree properties.
The pinctrl-0, pinctrl-1, pinctrl-<x> properties link to
a pin conﬁguration for a given state of the device.
The pinctrl-names property associates a name to each
state. The name default is special, and is automatically
selected by a device driver, without having to make an explicit
pinctrl function call.
In most cases, the following is suﬃcient:
i2c@11000 {
pinctrl-0 = <&pmx_twsi0>;
pinctrl-names = "default";
...
};
See Documentation/devicetree/bindings/pinctrl/
pinctrl-bindings.txt for details.

Defining pinctrl configurations
The different pinctrl configurations must be defined as child
nodes of the main pinctrl device (which controls the muxing
of pins).
The configurations may be defined at:
the SoC level (.dtsi file), for pin configurations that are often
shared between multiple boards
at the board level (.dts file) for configurations that are board
specific.
The pinctrl-<x> property of the consumer device points to
the pin configuration it needs through a DT phandle.
The description of the configurations is specific to each pinctrl
driver. See
Documentation/devicetree/bindings/pinctrl for the
DT bindings documentation.

Example on OMAP/AM33xx
On OMAP/AM33xx, the
pinctrl-single driver is
used. It is common between
multiple SoCs and simply
allows to configure pins by
writing a value to a register.
In each pin configuration,
a pinctrl-single,pins
value gives a list of
(register, value) pairs
needed to configure the
pins.
To know the correct values,
one must use the SoC and
board datasheets.
am33xx_pinmux: pinmux@44e10800 {
i2c0_pins: pinmux_i2c0_pins {
pinctrl-single,pins = <
/* i2c0_sda.i2c0_sda */
0x188 (PIN_INPUT_PULLUP | MUX_MODE0)
/* i2c0_scl.i2c0_scl */
0x18c (PIN_INPUT_PULLUP | MUX_MODE0)
>;
};
};
i2c0: i2c@44e0b000 {
pinctrl-0 = <&i2c0_pins>;
status = "okay";
tps: tps@2d {
reg = <0x2d>;
};
};

Example on Allwinner SoC

Practical lab - Communicate with the Nunchuk
Conﬁgure the pinmuxing for the
I2C bus used to communicate with
the Nunchuk
Validate that the I2C
communication works with user
space tools.
Extend the I2C driver started in the
previous lab to communicate with
the Nunchuk.

Kernel frameworks for device drivers
Kernel frameworks
for device drivers
Free Electrons
Embedded Linux
Developers
Free Electrons

Kernel and Device Drivers
In Linux, a driver is always
interfacing with:
a framework that allows the
driver to expose the
hardware features to user
space applications.
a bus infrastructure, part
of the device model, to
detect/communicate with
the hardware.
This section focuses on the
kernel frameworks, while the
device model was covered earlier
in this training.

User space vision of devices

Types of devices
Under Linux, there are essentially three types of devices:
Network devices. They are represented as network
interfaces, visible in user space using ifconfig.
Block devices. They are used to provide user space
applications access to raw storage devices (hard disks, USB
keys). They are visible to the applications as device ﬁles in
/dev.
Character devices. They are used to provide user space
applications access to all other types of devices (input, sound,
graphics, serial, etc.). They are also visible to the applications
as device ﬁles in /dev.
→ Most devices are character devices, so we will study these in
more details.

Major and minor numbers
Within the kernel, all block and character devices are
identiﬁed using a major and a minor number.
The major number typically indicates the family of the device.
The minor number typically indicates the number of the
device (when they are for example several serial ports)
Most major and minor numbers are statically allocated, and
identical across all Linux systems.
They are deﬁned in Documentation/devices.txt .

Devices: everything is a file
A very important Unix design decision was to represent most
of the “system objects” as files
It allows applications to manipulate all “system objects” with
the normal file API (open, read, write, close, etc.)
So, devices had to be represented as files to the applications
This is done through a special artifact called a device file
It is a special type of file, that associates a file name visible to
user space applications to the triplet (type, major, minor) that
the kernel understands
All device files are by convention stored in the /dev directory

Device files examples
Example of device files in a Linux system
$ ls -l /dev/ttyS0 /dev/tty1 /dev/sda1 /dev/sda2 /dev/zero
brw-rw---- 1 root disk 8, 1 2011-05-27 08:56 /dev/sda1
brw-rw---- 1 root disk 8, 2 2011-05-27 08:56 /dev/sda2
crw------- 1 root root 4, 1 2011-05-27 08:57 /dev/tty1
crw-rw---- 1 root dialout 4, 64 2011-05-27 08:56 /dev/ttyS0
crw-rw-rw- 1 root root 1, 5 2011-05-27 08:56 /dev/zero
Example C code that uses the usual file API to write data to a
serial port
int fd;
fd = open("/dev/ttyS0", O_RDWR);
write(fd, "Hello", 5);
close(fd);

Creating device files
On a basic Linux system, the device files have to be created
manually using the mknod command
mknod /dev/<device> [c|b] major minor
Needs root privileges
Coherency between device files and devices handled by the
kernel is left to the system developer
On more elaborate Linux systems, mechanisms can be added
to create/remove them automatically when devices appear
and disappear
devtmpfs virtual filesystem
udev daemon, solution used by desktop and server Linux
systems
mdev program, a lighter solution than udev

Character drivers

A character driver in the kernel
From the point of view of an application, a character device is
essentially a file.
The driver of a character device must therefore implement
operations that let applications think the device is a file:
open, close, read, write, etc.
In order to achieve this, a character driver must implement
the operations described in the struct file_operations
structure and register them.
The Linux filesystem layer will ensure that the driver’s
operations are called when a user space application makes the
corresponding system call.

From user space to the kernel: character devices

File operations
Here are the most important operations for a character driver.
All of them are optional.
#include <linux/fs.h>
struct file_operations {
ssize_t (*read) (struct file *, char __user *,
size_t, loff_t *);
ssize_t (*write) (struct file *, const char __user *,
size_t, loff_t *);
long (*unlocked_ioctl) (struct file *, unsigned int,
unsigned long);
int (*mmap) (struct file *, struct vm_area_struct *);
int (*open) (struct inode *, struct file *);
int (*release) (struct inode *, struct file *);
};

open() and release()
int foo_open(struct inode *i, struct file *f)
Called when user space opens the device file.
struct inode is a structure that uniquely represents a file in
the system (be it a regular file, a directory, a symbolic link, a
character or block device)
struct file is a structure created every time a file is opened.
Several file structures can point to the same inode structure.
Contains information like the current position, the opening
mode, etc.
Has a void *private_data pointer that one can freely use.
A pointer to the file structure is passed to all other
operations
int foo_release(struct inode *i, struct file *f)
Called when user space closes the file.

read()
ssize_t foo_read(struct file *f, char __user *buf,
size_t sz, loff_t *off)
Called when user space uses the read() system call on the
device.
Must read data from the device, write at most sz bytes in the
user space buffer buf, and update the current position in the
file off. f is a pointer to the same file structure that was
passed in the open() operation
Must return the number of bytes read.
On UNIX, read() operations typically block when there isn’t
enough data to read from the device

write()
ssize_t foo_write(struct file *f,
const char __user *buf, size_t sz, loff_t *off)
Called when user space uses the write() system call on the
device
The opposite of read, must read at most sz bytes from buf,
write it to the device, update off and return the number of
bytes written.

Exchanging data with user space 1/3
Kernel code isn’t allowed to directly access user space
memory, using memcpy() or direct pointer dereferencing
Doing so does not work on some architectures
If the address passed by the application was invalid, the
application would segfault.
To keep the kernel code portable and have proper error
handling, your driver must use special kernel functions to
exchange data with user space.

A single value
get_user(v, p);
The kernel variable v gets the value pointed by the user space
pointer p
put_user(v, p);
The value pointed by the user space pointer p is set to the
contents of the kernel variable v.
A buﬀer
unsigned long copy_to_user(void __user *to,
const void *from, unsigned long n);
unsigned long copy_from_user(void *to,
const void __user *from, unsigned long n);
The return value must be checked. Zero on success, non-zero
on failure. If non-zero, the convention is to return -EFAULT.

Zero copy access to user memory
Having to copy data to or from an intermediate kernel buﬀer
can become expensive when the amount of data to transfer is
large (video).
Zero copy options are possible:
mmap() system call to allow user space to directly access
memory mapped I/O space. See our mmap() chapter.
get_user_pages_fast() to get a mapping to user pages
without having to copy them. See http://j.mp/1sML7lP
(Kernel API doc). This API is more complex to use though.

unlocked ioctl()
long unlocked_ioctl(struct file *f,
unsigned int cmd, unsigned long arg)
Associated to the ioctl() system call.
Called unlocked because it didn’t hold the Big Kernel Lock
(gone now).
Allows to extend the driver capabilities beyond the limited
read/write API.
For example: changing the speed of a serial port, setting video
output format, querying a device serial number...
cmd is a number identifying the operation to perform
arg is the optional argument passed as third argument of the
ioctl() system call. Can be an integer, an address, etc.
The semantic of cmd and arg is driver-speciﬁc.

ioctl() example: kernel side
static long phantom_ioctl(struct file *file, unsigned int cmd,
unsigned long arg)
{
struct phm_reg r;
void __user *argp = (void __user *)arg;
switch (cmd) {
case PHN_SET_REG:
if (copy_from_user(&r, argp, sizeof(r)))
return -EFAULT;
/* Do something */
break;
case PHN_GET_REG:
if (copy_to_user(argp, &r, sizeof(r)))
return -EFAULT;
/* Do something */
break;
default:
return -ENOTTY;
}
return 0; }
Selected excerpt from drivers/misc/phantom.c

Ioctl() Example: Application Side
int main(void)
{
int fd, ret;
struct phm_reg reg;
fd = open("/dev/phantom");
assert(fd > 0);
reg.field1 = 42;
reg.field2 = 67;
ret = ioctl(fd, PHN_SET_REG, & reg);
assert(ret == 0);
return 0;
}

The concept of kernel frameworks

Beyond character drivers: kernel frameworks
Many device drivers are not implemented directly as character
drivers
They are implemented under a framework, speciﬁc to a given
device type (framebuﬀer, V4L, serial, etc.)
The framework allows to factorize the common parts of drivers
for the same type of devices
From user space, they are still seen as character devices by the
applications
The framework allows to provide a coherent user space
interface (ioctl, etc.) for every type of device, regardless of
the driver

Kernel Frameworks

Example of the framebuﬀer
framework

Example: Framebuffer Framework
Kernel option CONFIG_FB
menuconfig FB
tristate "Support for frame buffer devices"
Implemented in drivers/video/
fb.c, fbmem.c, fbmon.c, fbcmap.c, fbsysfs.c, modedb.c,
fbcvt.c
Implements a single character driver and defines the
user/kernel API
First part of include/linux/fb.h
Defines the set of operations a framebuffer driver must
implement and helper functions for the drivers
struct fb_ops
Second part of include/linux/fb.h

Framebuffer Driver Skeleton
Skeleton driver in drivers/video/skeletonfb.c
Implements the set of framebuffer specific operations defined
by the struct fb_ops structure
xxxfb_open()
xxxfb_read()
xxxfb_write()
xxxfb_release()
xxxfb_checkvar()
xxxfb_setpar()
xxxfb_setcolreg()
xxxfb_blank()
xxxfb_pan_display()
xxxfb_fillrect()
xxxfb_copyarea()
xxxfb_imageblit()
xxxfb_cursor()
xxxfb_rotate()
xxxfb_sync()
xxxfb_ioctl()
xxxfb_mmap()

After the implementation of the operations, deﬁnition of a
struct fb_ops structure
static struct fb_ops xxxfb_ops = {
.fb_open = xxxfb_open,
.fb_read = xxxfb_read,
.fb_write = xxxfb_write,
.fb_release = xxxfb_release,
.fb_check_var = xxxfb_check_var,
.fb_set_par = xxxfb_set_par,
.fb_setcolreg = xxxfb_setcolreg,
.fb_blank = xxxfb_blank,
.fb_pan_display = xxxfb_pan_display,
.fb_fillrect = xxxfb_fillrect, /* Needed !!! */
.fb_copyarea = xxxfb_copyarea, /* Needed !!! */
.fb_imageblit = xxxfb_imageblit, /* Needed !!! */
.fb_cursor = xxxfb_cursor, /* Optional !!! */
.fb_rotate = xxxfb_rotate,
.fb_sync = xxxfb_sync,
.fb_ioctl = xxxfb_ioctl,
.fb_mmap = xxxfb_mmap,
};

In the probe() function, registration of the framebuﬀer
device and operations
static int xxxfb_probe (struct pci_dev *dev,
const struct pci_device_id *ent)
{
struct fb_info *info;
[...]
info = framebuffer_alloc(sizeof(struct xxx_par), device);
[...]
info->fbops = &xxxfb_ops;
[...]
if (register_framebuffer(info) > 0)
return -EINVAL;
[...]
}
register_framebuffer() will create the character device
that can be used by user space applications with the generic
framebuﬀer API.

Driver-specific Data Structure
Each framework defines a structure that a device driver must
register to be recognized as a device in this framework
struct uart_port for serial ports, struct netdev for
network devices, struct fb_info for framebuffers, etc.
In addition to this structure, the driver usually needs to store
additional information about its device
This is typically done
By subclassing the appropriate framework structure
By storing a reference to the appropriate framework structure
Or by including your information in the framework structure

Driver-speciﬁc Data Structure Examples 1/2
i.MX serial driver: struct imx_port is a subclass of
struct uart_port
struct imx_port {
struct uart_port port;
struct timer_list timer;
unsigned int old_status;
int txirq, rxirq, rtsirq;
unsigned int have_rtscts:1;
[...]
};
ds1305 RTC driver: struct ds1305 has a reference to
struct rtc_device
struct ds1305 {
struct spi_device *spi;
struct rtc_device *rtc;
[...]
};

Driver-speciﬁc Data Structure Examples 2/2
rtl8150 network driver: struct rtl8150 has a reference to
struct net_device and is allocated within that framework
structure.
struct rtl8150 {
unsigned long flags;
struct usb_device *udev;
struct tasklet_struct tl;
[...]
};

Link Between Structures 1/4
The framework typically contains a struct device * pointer
that the driver must point to the corresponding
struct device
It’s the relation between the logical device (for example a
network interface) and the physical device (for example the
USB network adapter)
The device structure also contains a void * pointer that the
driver can freely use.
It’s often used to link back the device to the higher-level
structure from the framework.
It allows, for example, from the struct platform_device
structure, to ﬁnd the structure describing the logical device

static int serial_imx_probe(struct platform_device *pdev)
{
struct imx_port *sport;
[...]
/* setup the link between uart_port and the struct
* device inside the platform_device */
sport->port.dev = &pdev->dev;
[...]
/* setup the link between the struct device inside
* the platform device to the imx_port structure */
platform_set_drvdata(pdev, sport);
[...]
uart_add_one_port(&imx_reg, &sport->port);
}
static int serial_imx_remove(struct platform_device *pdev)
{
/* retrieve the imx_port from the platform_device */
struct imx_port *sport = platform_get_drvdata(pdev);
[...]
uart_remove_one_port(&imx_reg, &sport->port);
[...]
}

static int ds1305_probe(struct spi_device *spi)
{
struct ds1305 *ds1305;
[...]
/* set up driver data */
ds1305 = devm_kzalloc(&spi->dev, sizeof(*ds1305), GFP_KERNEL);
if (!ds1305)
return -ENOMEM;
ds1305->spi = spi;
spi_set_drvdata(spi, ds1305);
[...]
/* register RTC ... from here on, ds1305->ctrl needs locking */
ds1305->rtc = devm_rtc_device_register(&spi->dev, "ds1305",
&ds1305_ops, THIS_MODULE);
[...]
}
static int ds1305_remove(struct spi_device *spi)
{
struct ds1305 *ds1305 = spi_get_drvdata(spi);
[...]
}

static int rtl8150_probe(struct usb_interface *intf,
const struct usb_device_id *id)
{
struct usb_device *udev = interface_to_usbdev(intf);
rtl8150_t *dev;
netdev = alloc_etherdev(sizeof(rtl8150_t));
dev = netdev_priv(netdev);
[...]
dev->udev = udev;
dev->netdev = netdev;
[...]
usb_set_intfdata(intf, dev);
SET_NETDEV_DEV(netdev, &intf->dev);
[...]
}
static void rtl8150_disconnect(struct usb_interface *intf)
{
rtl8150_t *dev = usb_get_intfdata(intf);
[...]
}

The input subsystem
The input
subsystem
Free Electrons
Embedded Linux
Developers
Free Electrons

What is input subsystem?
The input subsystem takes care of all the input events coming
from the human user.
Initially written to support the USB HID (Human Interface
Device) devices, it quickly grew up to handle all kind of inputs
(using USB or not): keyboards, mice, joysticks, touchscreens,
etc.
The input subsystem is split in two parts:
Device drivers: they talk to the hardware (for example via
USB), and provide events (keystrokes, mouse movements,
touchscreen coordinates) to the input core
Event handlers: they get events from drivers and pass them
where needed via various interfaces (most of the time through
evdev)
In user space it is usually used by the graphic stack such as
X.Org, Wayland or Android’s InputManager.

Input subsystem diagram

Input subsystem overview
Kernel option CONFIG_INPUT
menuconfig INPUT
tristate "Generic input layer (needed for keyboard,
mouse, ...)"
Implemented in drivers/input/
input.c, input-polldev.c, evbug.c
Implements a single character driver and deﬁnes the
user/kernel API
include/uapi/linux/input.h
Deﬁnes the set of operations a input driver must implement
and helper functions for the drivers
struct input_dev for the device driver part
struct input_handler for the event handler part
include/linux/input.h

Input subsystem API 1/3
An input device is described by a very long
struct input_dev structure, an excerpt is:
struct input_dev {
const char *name;
[...]
unsigned long evbit[BITS_TO_LONGS(EV_CNT)];
unsigned long keybit[BITS_TO_LONGS(KEY_CNT)];
[...]
int (*getkeycode)(struct input_dev *dev,
struct input_keymap_entry *ke);
[...]
int (*open)(struct input_dev *dev);
[...]
int (*event)(struct input_dev *dev, unsigned int type,
unsigned int code, int value);
[...]
};
Before being used this struct must be allocated and initialized:
struct input_dev *input_allocate_device(void);
After unregistering struct input_dev, it must be freed:
void input_free_device(struct input_dev *dev);

Depending on the type of event that will be generated, the
input bit fields evbit and keybit must be configured: For
example, for a button we only generate EV_KEY type events,
and from these only BTN_0 events code:
set_bit(EV_KEY, myinput_dev.evbit);
set_bit(BTN_0, myinput_dev.keybit);
set_bit() is an atomic operation allowing to set a particular
bit to 1 (explained later).
Once the input device is allocated and filled, the function to
register it is:
int input_register_device(struct input_dev *);
When the driver is unloaded, the input device will be
unregistered using:
void input_unregister_device(struct input_dev *);

The events are sent by the driver to the event handler using
input_event(struct input_dev *dev, unsigned int
type, unsigned int code, int value);
The event types are documented in
Documentation/input/event-codes.txt
An event is composed by one or several input data changes
(packet of input data changes) such as the button state, the
relative or absolute position along an axis, etc..
After submitting potentially multiple events, the input core
must be notiﬁed by calling:
void input_sync(struct input_dev *dev):
The input subsystem provides other wrappers such as
input_report_key(), input_report_abs(), ...

Polled input subclass
The input subsystem provides a subclass supporting simple
input devices that do not raise interrupts but have to be
periodically scanned or polled to detect changes in their state.
A polled input device is described by a
struct input_polled_dev structure:
struct input_polled_dev {
void *private;
void (*open)(struct input_polled_dev *dev);
void (*close)(struct input_polled_dev *dev);
void (*poll)(struct input_polled_dev *dev);
unsigned int poll_interval; /* msec */
unsigned int poll_interval_max; /* msec */
unsigned int poll_interval_min; /* msec */
struct input_dev *input;
/* private: */
struct delayed_work work;
}

Polled input subsystem API
Allocating/freeing the struct input_polled_dev structure
is done using
input_allocate_polled_device()
input_free_polled_device()
Among the handlers of the struct input_polled_dev only
the poll() method is mandatory, this function polls the
device and posts input events.
The fields id, name, evkey and keybit of the input field
must be initialized too.
If none of the poll_interval fields are filled then the default
poll interval is 500ms.
The device registration/unregistration is done with:
input_register_polled_device(struct input_polled_
dev *dev).
input_unregister_polled_device(struct input_
polled_dev *dev)

evdev user space interface
The main user space interface to input devices is the event
interface
Each input device is represented as a /dev/input/event<X>
character device
A user space application can use blocking and non-blocking
reads, but also select() (to get notiﬁed of events) after
opening this device.
Each read will return struct input_event structures of the
following format:
struct input_event {
struct timeval time;
unsigned short type;
unsigned short code;
unsigned int value;
};
A very useful application for input device testing is evtest,
from http://cgit.freedesktop.org/evtest/

Practical lab - Expose the Nunchuk to user space
Extend the Nunchuk driver to
expose the Nunchuk features to
user space applications, as an input
device.
Test the operation of the Nunchuk
using sample user space
applications.

Memory Management
Memory
Management
Free Electrons
Embedded Linux
Developers
Free Electrons

Physical and Virtual Memory

Virtual Memory Organization
1GB reserved for kernel-space
Contains kernel code and core data
structures, identical in all address spaces
Most memory can be a direct mapping
of physical memory at a fixed offset
Complete 3GB exclusive mapping
available for each user space process
Process code and data (program, stack,
...)
Memory-mapped files
Not necessarily mapped to physical
memory (demand fault paging used for
dynamic mapping to physical memory
pages)
Differs from one address space to
another

Physical / virtual memory mapping

Accessing more physical memory
Only less than 1GB memory addressable directly through
kernel virtual address space
If more physical memory is present on the platform, part of
the memory will not be accessible by kernel space, but can be
used by user space
To allow the kernel to access more physical memory:
Change 1GB/3GB memory split (2GB/2GB)
(CONFIG_VMSPLIT_3G) ⇒ reduces total memory available for
each process
Change for a 64 bit architecture ;-) See
Documentation/x86/x86_64/mm.txt for an example.
Activate highmem support if available for your architecture:
Allows kernel to map parts of its non-directly accessible
memory
Mapping must be requested explicitly
Limited addresses ranges reserved for this usage
See http://lwn.net/Articles/75174/ for useful
explanations

Accessing even more physical memory!
If your 32 bit platform hosts more than 4GB, they just cannot
be mapped
PAE (Physical Address Expansion) may be supported by your
architecture
Adds some address extension bits used to index memory areas
Allows accessing up to 64 GB of physical memory on x86
Note that each user space process is still limited to a 3 GB
memory space

Notes on user space memory
New user space memory is allocated either from the already
allocated process memory, or using the mmap system call
Note that memory allocated may not be physically allocated:
Kernel uses demand fault paging to allocate the physical page
(the physical page is allocated when access to the virtual
address generates a page fault)
... or may have been swapped out, which also induces a page
fault
User space memory allocation is allowed to over-commit
memory (more than available physical memory) ⇒ can lead to
out of memory
OOM killer kicks in and selects a process to kill to retrieve
some memory. That’s better than letting the system freeze.

Back to kernel memory
Kernel memory allocators (see following slides) allocate
physical pages, and kernel allocated memory cannot be
swapped out, so no fault handling required for kernel memory.
Most kernel memory allocation functions also return a kernel
virtual address to be used within the kernel space.
Kernel memory low-level allocator manages pages. This is the
ﬁnest granularity (usually 4 KB, architecture dependent).
However, the kernel memory management handles smaller
memory allocations through its allocator (see SLAB allocators
– used by kmalloc()).

Allocators in the Kernel

Page Allocator
Appropriate for medium-size allocations
A page is usually 4K, but can be made greater in some
architectures (sh, mips: 4, 8, 16 or 64 KB, but not
conﬁgurable in x86 or arm).
Buddy allocator strategy, so only allocations of power of two
number of pages are possible: 1 page, 2 pages, 4 pages, 8
pages, 16 pages, etc.
Typical maximum size is 8192 KB, but it might depend on the
kernel conﬁguration.
The allocated area is virtually contiguous (of course), but also
physically contiguous. It is allocated in the identity-mapped
part of the kernel memory space.
This means that large areas may not be available or hard to
retrieve due to physical memory fragmentation.

Page Allocator API: Get free pages
unsigned long get_zeroed_page(int flags)
Returns the virtual address of a free page, initialized to zero
flags: see the next pages for details.
unsigned long __get_free_page(int flags)
Same, but doesn’t initialize the contents
unsigned long __get_free_pages(int flags,
unsigned int order)
Returns the starting virtual address of an area of several
contiguous pages in physical RAM, with order being
log2(number_of_pages).Can be computed from the size
with the get_order() function.

Page Allocator API: Free Pages
void free_page(unsigned long addr)
Frees one page.
void free_pages(unsigned long addr,
unsigned int order)
Frees multiple pages. Need to use the same order as in
allocation.

Page Allocator Flags
The most common ones are:
GFP_KERNEL
Standard kernel memory allocation. The allocation may block
in order to ﬁnd enough available memory. Fine for most
needs, except in interrupt handler context.
GFP_ATOMIC
RAM allocated from code which is not allowed to block
(interrupt handlers or critical sections). Never blocks, allows
to access emergency pools, but can fail if no free memory is
readily available.
GFP_DMA
Allocates memory in an area of the physical memory usable
for DMA transfers. See our DMA chapter.
Others are deﬁned in include/linux/gfp.h

SLAB Allocator 1/2
The SLAB allocator allows to create caches, which contains a
set of objects of the same size
The object size can be smaller or greater than the page size
The SLAB allocator takes care of growing or reducing the size
of the cache as needed, depending on the number of allocated
objects. It uses the page allocator to allocate and free pages.
SLAB caches are used for data structures that are present in
many many instances in the kernel: directory entries, ﬁle
objects, network packet descriptors, process descriptors, etc.
See /proc/slabinfo
They are rarely used for individual drivers.
See include/linux/slab.h for the API

SLAB Allocator 2/2

Different SLAB Allocators
There are three different, but API compatible,
implementations of a SLAB allocator in the Linux kernel. A
particular implementation is chosen at configuration time.
SLAB: legacy, well proven allocator.
Still the default in most ARM defconfig files.
SLOB: much simpler. More space efficient but doesn’t scale
well. Saves a few hundreds of KB in small systems (depends
on CONFIG_EXPERT)
Linux 3.17 on ARM: used in 5 defconfig files
SLUB: more recent and simpler than SLAB, scaling much
better (in particular for huge systems) and creating less
fragmentation.
Linux 3.17 on ARM: used in 0 defconfig files

kmalloc Allocator
The kmalloc allocator is the general purpose memory allocator
in the Linux kernel
For small sizes, it relies on generic SLAB caches, named
kmalloc-XXX in /proc/slabinfo
For larger sizes, it relies on the page allocator
The allocated area is guaranteed to be physically contiguous
The allocated area size is rounded up to the size of the
smallest SLAB cache in which it can fit (while using the SLAB
allocator directly allows to have more flexibility)
It uses the same flags as the page allocator (GFP_KERNEL,
GFP_ATOMIC, GFP_DMA, etc.) with the same semantics.
Maximum sizes, on x86 and arm (see http://j.mp/YIGq6W):
- Per allocation: 4 MB
- Total allocations: 128 MB
Should be used as the primary allocator unless there is a
strong reason to use another one.

kmalloc API 1/2
#include <linux/slab.h>
void *kmalloc(size_t size, int flags);
Allocate size bytes, and return a pointer to the area (virtual
address)
size: number of bytes to allocate
flags: same ﬂags as the page allocator
void kfree(const void *objp);
Free an allocated area
Example: (drivers/infiniband/core/cache.c)
struct ib_update_work *work;
work = kmalloc(sizeof *work, GFP_ATOMIC);
...
kfree(work);

kmalloc API 2/2
void *kzalloc(size_t size, gfp_t flags);
Allocates a zero-initialized buffer
void *kcalloc(size_t n, size_t size, gfp_t flags);
Allocates memory for an array of n elements of size size, and
zeroes its contents.
void *krealloc(const void *p, size_t new_size, gfp_t flags);
Changes the size of the buffer pointed by p to new_size, by
reallocating a new buffer and copying the data, unless
new_size fits within the alignment of the existing buffer.

devm kmalloc functions
Automatically free the allocated buﬀers when the
corresponding device or module is unprobed.
Need to have a reference to a struct device.
void *devm_kmalloc(struct device *dev, size_t size, int flags);
void *devm_kzalloc(struct device *dev, size_t size, int flags);
void *devm_kcalloc(struct device *dev, size_t n, size_t size, gfp_t flags)
void *devm_kfree(struct device *dev, void *p);
Useful to immediately free an allocated buﬀer
See Documentation/driver-model/devres.txt for details
about managed device resources.

vmalloc Allocator
The vmalloc() allocator can be used to obtain virtually
contiguous memory zones, but not physically contiguous. The
requested memory size is rounded up to the next page.
The allocated area is in the kernel space part of the address
space, but outside of the identically-mapped area
Allocations of fairly large areas is possible (almost as big as
total available memory, see http://j.mp/YIGq6W again),
since physical memory fragmentation is not an issue, but areas
cannot be used for DMA, as DMA usually requires physically
contiguous buﬀers.
API in include/linux/vmalloc.h
void *vmalloc(unsigned long size);
Returns a virtual address
void vfree(void *addr);

Kernel memory debugging
Kmemcheck
Dynamic checker for access to uninitialized memory.
Only available on x86 so far (Linux 3.17 status), but will help
to improve architecture independent code anyway.
See Documentation/kmemcheck.txt for details.
Kmemleak
Dynamic checker for memory leaks
This feature is available for all architectures.
See Documentation/kmemleak.txt for details.
Both have a signiﬁcant overhead. Only use them in development!

I/O Memory and Ports
I/O Memory and
Ports
Free Electrons
Embedded Linux
Developers
Free Electrons

Port I/O vs. Memory-Mapped I/O
MMIO
Same address bus to address memory and I/O devices
Access to the I/O devices using regular instructions
Most widely used I/O method across the diﬀerent architectures
supported by Linux
PIO
Diﬀerent address spaces for memory and I/O devices
Uses a special class of CPU instructions to access I/O devices
Example on x86: IN and OUT instructions

MMIO vs PIO

Requesting I/O ports
Tells the kernel which driver is using which I/O ports
Allows to prevent other drivers from using the same I/O ports,
but is purely voluntary.
struct resource *request_region(
unsigned long start,
unsigned long len,
char *name);
Tries to reserve the given region and returns NULL if
unsuccessful.
request_region(0x0170, 8, "ide1");
void release_region(
unsigned long len);

/proc/ioports example (x86)
0000-001f : dma1
0020-0021 : pic1
0040-0043 : timer0
0050-0053 : timer1
0070-0077 : rtc
0080-008f : dma page reg
00a0-00a1 : pic2
00c0-00df : dma2
00f0-00ff : fpu
0170-0177 : ide1
01f0-01f7 : ide0
0376-0376 : ide1
03f6-03f6 : ide0
03f8-03ff : serial
0800-087f : 0000:00:1f.0
...

Accessing I/O ports
Functions to read/write bytes (b), word (w) and longs (l) to
I/O ports:
unsigned in[bwl](unsigned long port)
void out[bwl](value, unsigned long port)
And the strings variants: often more eﬃcient than the
corresponding C loop, if the processor supports such
operations!
void ins[bwl](unsigned port, void *addr,
unsigned long count)
void outs[bwl](unsigned port, void *addr,
unsigned long count)
Examples
read 8 bits
oldlcr = inb(baseio + UART_LCR)
write 8 bits
outb(MOXA_MUST_ENTER_ENCHANCE, baseio + UART_LCR)

Requesting I/O memory
Functions equivalent to request_region() and
release_region(), but for I/O memory.
struct resource *request_mem_region(
unsigned long len,
char *name);
void release_mem_region(
unsigned long len);

/proc/iomem example
00000000-0009efff : System RAM
0009f000-0009ffff : reserved
000a0000-000bffff : Video RAM area
000c0000-000cffff : Video ROM
000f0000-000fffff : System ROM
00100000-3ffadfff : System RAM
00100000-0030afff : Kernel code
0030b000-003b4bff : Kernel data
3ffae000-3fffffff : reserved
40000000-400003ff : 0000:00:1f.1
40001000-40001fff : 0000:02:01.0
40400000-407fffff : PCI CardBus #03
40800000-40bfffff : PCI CardBus #03
a0000000-a0000fff : pcmcia_socket0
e8000000-efffffff : PCI Bus #01
...

Mapping I/O memory in virtual memory
Load/store instructions work with virtual addresses
To access I/O memory, drivers need to have a virtual address
that the processor can handle, because I/O memory is not
mapped by default in virtual memory.
The ioremap function satisﬁes this need:
#include <asm/io.h>
void __iomem *ioremap(phys_addr_t phys_addr,
unsigned long size);
void iounmap(void __iomem *addr);
Caution: check that ioremap() doesn’t return a NULL
address!

ioremap()
ioremap(0xFFEBC00, 4096) = 0xCDEFA000

Managed API
Using request_mem_region() and ioremap() in device drivers is
now deprecated. You should use the below ”managed” functions
instead, which simplify driver coding and error handling:
devm_ioremap()
devm_iounmap()
devm_ioremap_resource()
Takes care of both the request and remapping operations!

Accessing MMIO devices
Directly reading from or writing to addresses returned by
ioremap() (pointer dereferencing) may not work on some
architectures.
To do PCI-style, little-endian accesses, conversion being done
automatically
unsigned read[bwl](void *addr);
void write[bwl](unsigned val, void *addr);
To do raw access, without endianness conversion
unsigned __raw_read[bwl](void *addr);
void __raw_write[bwl](unsigned val, void *addr);
Example
32 bits write
__raw_writel(1 << KS8695_IRQ_UART_TX,
membase + KS8695_INTST);

Avoiding I/O access issues
Caching on I/O ports or memory already disabled
Use the macros, they do the right thing for your architecture
The compiler and/or CPU can reorder memory accesses,
which might cause troubles for your devices is they expect one
register to be read/written before another one.
Memory barriers are available to prevent this reordering
rmb() is a read memory barrier, prevents reads to cross the
barrier
wmb() is a write memory barrier
mb() is a read-write memory barrier
Starts to be a problem with CPUs that reorder instructions
and SMP.
See Documentation/memory-barriers.txt for details

/dev/mem
Used to provide user space applications with direct access to
physical addresses.
Usage: open /dev/mem and read or write at given oﬀset.
What you read or write is the value at the corresponding
physical address.
Used by applications such as the X server to write directly to
device memory.
On x86, arm, arm64, tile, powerpc, unicore32, s390:
CONFIG_STRICT_DEVMEM option to restrict /dev/mem
non-RAM addresses, for security reasons (Linux 3.10 status).

Practical lab - I/O Memory and Ports
Add UART devices to the board
device tree
Access I/O registers to control the
device and send ﬁrst characters to
it.

The misc subsystem
The misc
subsystem
Free Electrons
Embedded Linux
Developers
Free Electrons

Why a misc subsystem?
The kernel offers a large number of frameworks covering a
wide range of device types: input, network, video, audio, etc.
These frameworks allow to factorize common functionality
between drivers and offer a consistent API to user space
applications.
However, there are some devices that really do not fit in any
of the existing frameworks.
Highly customized devices implemented in a FPGA, or other
weird devices for which implementing a complete framework is
not useful.
The drivers for such devices could be implemented directly as
raw character drivers.
But there is a subsystem that makes this work a little bit
easier: the misc subsystem.
It is really only a thin layer above the character driver API.

Misc subsystem diagram

Misc subsystem API (1/2)
The misc subsystem API mainly provides two functions, to
register and unregister a single misc device:
int misc_register(struct miscdevice * misc);
int misc_deregister(struct miscdevice *misc);
A misc device is described by a struct miscdevice
structure:
struct miscdevice {
int minor;
const char *name;
const struct file_operations *fops;
struct list_head list;
struct device *parent;
struct device *this_device;
const char *nodename;
umode_t mode;
};

Misc subsystem API (2/2)
The main ﬁelds to be ﬁlled in struct miscdevice are:
minor, the minor number for the device, or
MISC_DYNAMIC_MINOR to get a minor number automatically
assigned.
name, name of the device, which will be used to create the
device node if devtmpfs is used.
fops, pointer to a struct file_operations structure, that
describes which functions implement the read, write, ioctl,
etc. operations.
parent, the struct device that represents the hardware
device exposed by this driver.

User space API for misc devices
misc devices are regular character devices
The operations they support in user space depends on the
operations the kernel driver implements:
The open() and close() system calls to open/close the
device.
The read() and write() system calls to read/write to/from
the device.
The ioctl() system call to call some driver-speciﬁc
operations.

Practical lab - Output-only serial port driver
Extend the driver started in the
previous lab by registering it into
the misc subsystem.
Implement serial output
functionality through the misc
subsystem.
Test serial output using user space
applications.

Processes, scheduling and interrupts
Processes,
scheduling and
interrupts
Free Electrons
Embedded Linux
Developers
Free Electrons

Processes and scheduling

Process, thread?
Confusion about the terms process, thread and task
In Unix, a process is created using fork() and is composed of
An address space, which contains the program code, data,
stack, shared libraries, etc.
One thread, that starts executing the main() function.
Upon creation, a process contains one thread
Additional threads can be created inside an existing process,
using pthread_create()
They run in the same address space as the initial thread of the
process
They start executing a function passed as argument to
pthread_create()

Process, thread: kernel point of view
The kernel represents each thread running in the system by a
structure of type struct task_struct
From a scheduling point of view, it makes no diﬀerence
between the initial thread of a process and all additional
threads created dynamically using pthread_create()

A thread life

Execution of system calls
The execution of system calls takes place in the context of the
thread requesting them.

Sleeping

Sleeping
Sleeping is needed when a process (user space or kernel space) is
waiting for data.

How to sleep 1/3
Must declare a wait queue
A wait queue will be used to store the list of threads waiting
for an event
Static queue declaration
useful to declare as a global variable
DECLARE_WAIT_QUEUE_HEAD(module_queue);
Or dynamic queue declaration
Useful to embed the wait queue inside another data structure
wait_queue_head_t queue;
init_waitqueue_head(&queue);

How to sleep 2/3
Several ways to make a kernel process sleep
void wait_event(queue, condition);
Sleeps until the task is woken up and the given C expression is
true. Caution: can’t be interrupted (can’t kill the user space
process!)
int wait_event_killable(queue, condition);
Can be interrupted, but only by a fatal signal (SIGKILL).
Returns -ERESTARSYS if interrupted.
int wait_event_interruptible(queue, condition);
Can be interrupted by any signal. Returns -ERESTARTSYS if
interrupted.

How to sleep 3/3
int wait_event_timeout(queue, condition, timeout);
Also stops sleeping when the task is woken up and the timeout
expired. Returns 0 if the timeout elapsed, non-zero if the
condition was met.
int wait_event_interruptible_timeout(queue,
condition, timeout);
Same as above, interruptible. Returns 0 if the timeout elapsed,
-ERESTARTSYS if interrupted, positive value if the condition
was met.

How to Sleep - Example
ret = wait_event_interruptible
(sonypi_device.fifo_proc_list,
kfifo_len(sonypi_device.fifo) != 0);
if (ret)
return ret;

Waking up!
Typically done by interrupt handlers when data sleeping
processes are waiting for become available.
wake_up(&queue);
Wakes up all processes in the wait queue
wake_up_interruptible(&queue);
Wakes up all processes waiting in an interruptible sleep on the
given queue

Exclusive vs. non-exclusive
wait_event_interruptible() puts a task in a
non-exclusive wait.
All non-exclusive tasks are woken up by wake_up() /
wake_up_interruptible()
wait_event_interruptible_exclusive() puts a task in
an exclusive wait.
wake_up() / wake_up_interruptible() wakes up all
non-exclusive tasks and only one exclusive task
wake_up_all() / wake_up_interruptible_all() wakes up
all non-exclusive and all exclusive tasks
Exclusive sleeps are useful to avoid waking up multiple tasks
when only one will be able to “consume” the event.
Non-exclusive sleeps are useful when the event can “beneﬁt”
to multiple tasks.

Sleeping and waking up - Implementation 1/2

Sleeping and waking up - Implementation 2/2
The scheduler doesn’t keep evaluating the sleeping condition!
wait_event(queue, condition);
The process is put in the TASK_UNINTERRUPTIBLE state.
wake_up(&queue);
All processes waiting in queue are woken up, so they get
scheduled later and have the opportunity to evaluate the
condition again and go back to sleep if it is not met.
See include/linux/wait.h for implementation details.

Interrupt Management

Registering an interrupt handler 1/2
The ”managed” API is recommended:
int devm_request_irq(struct device *dev,
unsigned int irq,
irq_handler_t handler,
unsigned long irq_flags,
const char *devname,
void *dev_id);
Register an interrupt handler.
device for automatic freeing at device or module release time.
irq is the requested IRQ channel. For platform devices, use
platform_get_irq() to retrieve the interrupt number.
handler is a pointer to the IRQ handler
irq_flags are option masks (see next slide)
devname is the registered name
dev_id is a pointer to some data. It cannot be NULL as it is
used as an identiﬁer for free_irq() when using shared IRQs.

Releasing an interrupt handler
void devm_free_irq(struct device *dev,
unsigned int irq, void *dev_id);
Explicitly release an interrupt handler. Done automatically in
normal situations.
Deﬁned in include/linux/interrupt.h

Registering an interrupt handler 2/2
Main irq_flags bit values
(can be combined, 0 when no ﬂags are needed):
IRQF_SHARED
The interrupt channel can be shared by several devices.
Requires a hardware status register telling whether an IRQ was
raised or not.

Interrupt handler constraints
No guarantee in which address space the system will be in
when the interrupt occurs: can’t transfer data to and from
user space.
Interrupt handler execution is managed by the CPU, not by
the scheduler. Handlers can’t run actions that may sleep,
because there is nothing to resume their execution. In
particular, need to allocate memory with GFP_ATOMIC.
Interrupt handlers are run with all interrupts disabled on the
local CPU (see http://lwn.net/Articles/380931).
Therefore, they have to complete their job quickly enough, to
avoiding blocking interrupts for too long.

/proc/interrupts on a Panda board
CPU0 CPU1
39: 4 0 GIC TWL6030-PIH
41: 0 0 GIC l3-dbg-irq
42: 0 0 GIC l3-app-irq
43: 0 0 GIC prcm
44: 20294 0 GIC DMA
52: 0 0 GIC gpmc
...
IPI0: 0 0 Timer broadcast interrupts
IPI1: 23095 25663 Rescheduling interrupts
IPI2: 0 0 Function call interrupts
IPI3: 231 173 Single function call interrupts
IPI4: 0 0 CPU stop interrupts
LOC: 196407 136995 Local timer interrupts
Err: 0
Note: interrupt numbers shown on the left-most column are virtual numbers
when the Device Tree is used. The real physical interrupt numbers are either
shown as an additional column, or can be seen in
/sys/kernel/debug/irq_domain_mapping.

Interrupt handler prototype
irqreturn_t foo_interrupt(int irq, void *dev_id)
irq, the IRQ number
dev_id, the opaque pointer that was passed to
devm_request_irq()
Return value
IRQ_HANDLED: recognized and handled interrupt
IRQ_NONE: not on a device managed by the module. Useful to
share interrupt channels and/or report spurious interrupts to
the kernel.

Typical interrupt handler’s job
Acknowledge the interrupt to the device (otherwise no more
interrupts will be generated, or the interrupt will keep ﬁring
over and over again)
Read/write data from/to the device
Wake up any waiting process waiting for the completion of an
operation, typically using wait queues
wake_up_interruptible(&module_queue);

Threaded interrupts
In 2.6.30, support for threaded interrupts has been added to the
Linux kernel
The interrupt handler is executed inside a thread.
Allows to block during the interrupt handler, which is often
needed for I2C/SPI devices as the interrupt handler needs to
communicate with them.
Allows to set a priority for the interrupt handler execution,
which is useful for real-time usage of Linux
int devm_request_threaded_irq(
struct device *dev,
unsigned int irq,
irq_handler_t handler, irq_handler_t thread_fn
unsigned long flags, const char *name, void *dev);
handler, “hard IRQ” handler
thread_fn, executed in a thread

Top half and bottom half processing
Splitting the execution of interrupt handlers in 2 parts
Top half
This is the real interrupt handler, which should complete as
quickly as possible since all interrupts are disabled. If possible,
take the data out of the device and schedule a bottom half to
handle it.
Bottom half
Is the general Linux name for various mechanisms which allow
to postpone the handling of interrupt-related work.
Implemented in Linux as softirqs, tasklets or workqueues.

Top half and bottom half diagram

Softirqs
Softirqs are a form of bottom half processing
The softirqs handlers are executed with all interrupts enabled,
and a given softirq handler can run simultaneously on multiple
CPUs
They are executed once all interrupt handlers have completed,
before the kernel resumes scheduling processes, so sleeping is
not allowed.
The number of softirqs is ﬁxed in the system, so softirqs are
not directly used by drivers, but by complete kernel
subsystems (network, etc.)
The list of softirqs is deﬁned in
include/linux/interrupt.h: HI, TIMER, NET_TX, NET_RX,
BLOCK, BLOCK_IOPOLL, TASKLET, SCHED, HRTIMER, RCU
The HI and TASKLET softirqs are used to execute tasklets

Tasklets
Tasklets are executed within the HI and TASKLET softirqs.
They are executed with all interrupts enabled, but a given
tasklet is guaranteed to execute on a single CPU at a time.
A tasklet can be declared statically with the
DECLARE_TASKLET() macro or dynamically with the
tasklet_init() function. A tasklet is simply implemented
as a function. Tasklets can easily be used by individual device
drivers, as opposed to softirqs.
The interrupt handler can schedule the execution of a tasklet
with
tasklet_schedule() to get it executed in the TASKLET
softirq
tasklet_hi_schedule() to get it executed in the HI softirq
(higher priority)

Tasklet Example: simpliﬁed atmel serial.c 1/2
/* The tasklet function */
static void atmel_tasklet_func(unsigned long data) {
struct uart_port *port = (struct uart_port *)data;
[...]
}
/* Registering the tasklet */
init function(...) {
[...]
tasklet_init(&atmel_port->tasklet,
atmel_tasklet_func, (unsigned long)port);
[...]
}

Tasklet Example: simpliﬁed atmel serial.c 2/2
/* Removing the tasklet */
cleanup function(...) {
[...]
tasklet_kill(&atmel_port->tasklet);
[...]
}
/* Triggering execution of the tasklet */
somewhere function(...) {
tasklet_schedule(&atmel_port->tasklet);
}

Workqueues
Workqueues are a general mechanism for deferring work. It is
not limited in usage to handling interrupts.
The function registered as workqueue is executed in a thread,
which means:
All interrupts are enabled
Sleeping is allowed
A workqueue is registered with INIT_WORK() and typically
triggered with queue_work()
The complete API, in include/linux/workqueue.h
provides many other possibilities (creating its own workqueue
threads, etc.)

Interrupt management summary
Device driver
When the device ﬁle is ﬁrst opened, register an interrupt
handler for the device’s interrupt channel.
Interrupt handler
Called when an interrupt is raised.
Acknowledge the interrupt
If needed, schedule a tasklet taking care of handling data.
Otherwise, wake up processes waiting for the data.
Tasklet
Process the data
Wake up processes waiting for the data
Device driver
When the device is no longer opened by any process,
unregister the interrupt handler.

Practical lab - Interrupts
Adding read capability to the
character driver developed earlier.
Register an interrupt handler.
Waiting for data to be available in
the read ﬁle operation.
Waking up the code when data are
available from the device.

Concurrent Access to Resources: Locking
Concurrent Access
to Resources:
Locking
Free Electrons
Embedded Linux
Developers
Free Electrons

Sources of concurrency issues
In terms of concurrency, the kernel has the same constraint as
a multi-threaded program: its state is global and visible in all
executions contexts
Concurrency arises because of
Interrupts, which interrupts the current thread to execute an
interrupt handler. They may be using shared resources.
Kernel preemption, if enabled, causes the kernel to switch from
the execution of one system call to another. They may be
using shared resources.
Multiprocessing, in which case code is really executed in
parallel on diﬀerent processors, and they may be using shared
resources as well.
The solution is to keep as much local state as possible and for
the shared resources, use locking.

Concurrency protection with locks

Linux mutexes
The kernel’s main locking primitive
The process requesting the lock blocks when the lock is
already held. Mutexes can therefore only be used in contexts
where sleeping is allowed.
Mutex deﬁnition:
#include <linux/mutex.h>
Initializing a mutex statically:
DEFINE_MUTEX(name);
Or initializing a mutex dynamically:
void mutex_init(struct mutex *lock);

Locking and Unlocking Mutexes 1/2
void mutex_lock(struct mutex *lock);
Tries to lock the mutex, sleeps otherwise.
Caution: can’t be interrupted, resulting in processes you
cannot kill!
int mutex_lock_killable(struct mutex *lock);
Same, but can be interrupted by a fatal (SIGKILL) signal. If
interrupted, returns a non zero value and doesn’t hold the
lock. Test the return value!!!
int mutex_lock_interruptible(struct mutex *lock);
Same, but can be interrupted by any signal.

Locking and Unlocking Mutexes 2/2
int mutex_trylock(struct mutex *lock);
Never waits. Returns a non zero value if the mutex is not
available.
int mutex_is_locked(struct mutex *lock);
Just tells whether the mutex is locked or not.
void mutex_unlock(struct mutex *lock);
Releases the lock. Do it as soon as you leave the critical
section.

Spinlocks
Locks to be used for code that is not allowed to sleep
(interrupt handlers), or that doesn’t want to sleep (critical
sections). Be very careful not to call functions which can
sleep!
Originally intended for multiprocessor systems
Spinlocks never sleep and keep spinning in a loop until the
lock is available.
Spinlocks cause kernel preemption to be disabled on the CPU
executing them.
The critical section protected by a spinlock is not allowed to
sleep.

Initializing Spinlocks
Statically
DEFINE_SPINLOCK(my_lock);
Dynamically
void spin_lock_init(spinlock_t *lock);

Using Spinlocks 1/2
Several variants, depending on where the spinlock is called:
void spin_lock(spinlock_t *lock);
void spin_unlock(spinlock_t *lock);
Doesn’t disable interrupts. Used for locking in process context
(critical sections in which you do not want to sleep).
void spin_lock_irqsave(spinlock_t *lock,
unsigned long flags);
void spin_unlock_irqrestore(spinlock_t *lock,
unsigned long flags);
Disables / restores IRQs on the local CPU.
Typically used when the lock can be accessed in both process
and interrupt context, to prevent preemption by interrupts.

Using Spinlocks 2/2
void spin_lock_bh(spinlock_t *lock);
void spin_unlock_bh(spinlock_t *lock);
Disables software interrupts, but not hardware ones.
Useful to protect shared data accessed in process context and
in a soft interrupt (bottom half).
No need to disable hardware interrupts in this case.
Note that reader / writer spinlocks also exist.

Spinlock example
Spinlock structure embedded into struct uart_port
struct uart_port {
spinlock_t lock;
/* Other fields */
};
Spinlock taken/released with protection against interrupts
static unsigned int ulite_tx_empty
(struct uart_port *port) {
unsigned long flags;
spin_lock_irqsave(&port->lock, flags);
/* Do something */
spin_unlock_irqrestore(&port->lock, flags);
}

Deadlock Situations
They can lock up your system. Make sure they never happen!
Don’t call a function that can try to get access to the same
lock
Holding multiple locks is risky!

Kernel lock validator
From Ingo Molnar and Arjan van de Ven
Adds instrumentation to kernel locking code
Detect violations of locking rules during system life, such as:
Locks acquired in diﬀerent order (keeps track of locking
sequences and compares them).
Spinlocks acquired in interrupt handlers and also in process
context when interrupts are enabled.
Not suitable for production systems but acceptable overhead in
development.
See Documentation/locking/lockdep-design.txt for
details

Alternatives to Locking
As we have just seen, locking can have a strong negative
impact on system performance. In some situations, you could
do without it.
By using lock-free algorithms like Read Copy Update (RCU).
RCU API available in the kernel (See
http://en.wikipedia.org/wiki/RCU).
When available, use atomic operations.

Atomic Variables 1/2
Useful when the shared resource is an integer value
Even an instruction like n++ is not guaranteed to be atomic
on all processors!
Atomic operations deﬁnitions
#include <asm/atomic.h>
atomic_t
Contains a signed integer (at least 24 bits)
Atomic operations (main ones)
Set or read the counter:
void atomic_set(atomic_t *v, int i);
int atomic_read(atomic_t *v);
Operations without return value:
void atomic_inc(atomic_t *v);
void atomic_dec(atomic_t *v);
void atomic_add(int i, atomic_t *v);
void atomic_sub(int i, atomic_t *v);

Atomic Variables 2/2
Similar functions testing the result:
int atomic_inc_and_test(...);
int atomic_dec_and_test(...);
int atomic_sub_and_test(...);
Functions returning the new value:
int atomic_inc_return(...);
int atomic_dec_return(...);
int atomic_add_return(...);
int atomic_sub_return(...);

Atomic Bit Operations
Supply very fast, atomic operations
On most platforms, apply to an unsigned long type.
Apply to a void type on a few others.
Set, clear, toggle a given bit:
void set_bit(int nr, unsigned long * addr);
void clear_bit(int nr, unsigned long * addr);
void change_bit(int nr, unsigned long * addr);
Test bit value:
int test_bit(int nr, unsigned long *addr);
Test and modify (return the previous value):
int test_and_set_bit(...);
int test_and_clear_bit(...);
int test_and_change_bit(...);

Practical lab - Locking
Add locking to the driver to
prevent concurrent accesses to
shared resources

Kernel Debugging
Kernel Debugging
Free Electrons
Embedded Linux
Developers
Free Electrons

Debugging Using Messages
Three APIs are available
The old printk(), no longer recommended for new debugging
messages
The pr_*() family of functions: pr_emerg(), pr_alert(),
pr_crit(), pr_err(), pr_warning(), pr_notice(),
pr_info(), pr_cont()
and the special pr_debug() (see next page)
They take a classic format string with arguments
Defined in include/linux/printk.h
The dev_*() family of functions: dev_emerg(),
dev_alert(), dev_crit(), dev_err(), dev_warning(),
dev_notice(), dev_info()
and the special dev_dbg() (see next page)
They take a pointer to struct device as first argument, and
then a format string with arguments
Defined in include/linux/device.h
To be used in drivers integrated with the Linux device model

pr debug() and dev dbg()
When the driver is compiled with DEBUG defined, all these
messages are compiled and printed at the debug level. DEBUG
can be defined by #define DEBUG at the beginning of the
driver, or using ccflags-$(CONFIG_DRIVER) += -DDEBUG
in the Makefile
When the kernel is compiled with CONFIG_DYNAMIC_DEBUG,
then these messages can dynamically be enabled on a per-file,
per-module or per-message basis
See Documentation/dynamic-debug-howto.txt for details
Very powerful feature to only get the debug messages you’re
interested in.
When DEBUG is not defined and CONFIG_DYNAMIC_DEBUG is
not enabled, these messages are not compiled in.

Configuring The Priority
Each message is associated to a priority, ranging from 0 for
emergency to 7 for debug.
All the messages, regardless of their priority, are stored in the
kernel log ring buffer
Typically accessed using the dmesg command
Some of the messages may appear on the console, depending
on their priority and the configuration of
The loglevel kernel parameter, which defines the priority
above which messages are displayed on the console. See
Documentation/kernel-parameters.txt for details.
The value of /proc/sys/kernel/printk, which allows to
change at runtime the priority above which messages are
displayed on the console. See
Documentation/sysctl/kernel.txt for details.

DebugFS
A virtual filesystem to export debugging information to user
space.
Kernel configuration: DEBUG_FS
Kernel hacking -> Debug Filesystem
The debugging interface disappears when Debugfs is
configured out.
You can mount it as follows:
sudo mount -t debugfs none /sys/kernel/debug
First described on http://lwn.net/Articles/115405/
API documented in the Linux Kernel Filesystem API:
Documentation/DocBook/filesystems/

DebugFS API
Create a sub-directory for your driver:
struct dentry *debugfs_create_dir(const char *name,
struct dentry *parent);
Expose an integer as a file in DebugFS:
struct dentry *debugfs_create_{u,x}{8,16,32}
(const char *name, mode_t mode, struct dentry *parent,
u8 *value);
u for decimal representation
x for hexadecimal representation
Expose a binary blob as a file in DebugFS:
struct dentry *debugfs_create_blob(const char *name,
mode_t mode, struct dentry *parent,
struct debugfs_blob_wrapper *blob);
Also possible to support writable DebugFS files or customize
the output using the more generic debugfs_create_file()
function.

Deprecated Debugging Mechanisms
Some additional debugging mechanisms, whose usage is now
considered deprecated
Adding special ioctl() commands for debugging purposes.
DebugFS is preferred.
Adding special entries in the proc ﬁlesystem. DebugFS is
preferred.
Adding special entries in the sysfs ﬁlesystem. DebugFS is
preferred.
Using printk(). The pr_*() and dev_*() functions are
preferred.

Using Magic SysRq
Allows to run multiple debug / rescue commands even when
the kernel seems to be in deep trouble
On PC: [Alt] + [SysRq] + <character>
On embedded: break character on the serial line +
<character>
Example commands:
n: makes RT processes nice-able.
w: shows the kernel stack of all sleeping processes
t: shows the kernel stack of all running processes
b: reboot the system
You can even register your own!
Detailed in Documentation/sysrq.txt

kgdb - A Kernel Debugger
The execution of the kernel is fully controlled by gdb from
another machine, connected through a serial line.
Can do almost everything, including inserting breakpoints in
interrupt handlers.
Feature supported for the most popular CPU architectures

Using kgdb 1/2
Details available in the kernel documentation:
Documentation/DocBook/kgdb/
Recommended to turn on CONFIG_FRAME_POINTER to aid in
producing more reliable stack backtraces in gdb.
You must include a kgdb I/O driver. One of them is kgdb
over serial console (kgdboc: kgdb over console, enabled by
CONFIG_KGDB_SERIAL_CONSOLE)
Conﬁgure kgdboc at boot time by passing to the kernel:
kgdboc=<tty-device>,<bauds>.
For example: kgdboc=ttyS0,115200

Using kgdb 2/2
Then also pass kgdbwait to the kernel: it makes kgdb wait
for a debugger connection.
Boot your kernel, and when the console is initialized, interrupt
the kernel with Alt + SysRq + g.
On your workstation, start gdb as follows:
gdb ./vmlinux
(gdb) set remotebaud 115200
(gdb) target remote /dev/ttyS0
Once connected, you can debug a kernel the way you would
debug an application program.

Debugging with a JTAG Interface
Two types of JTAG dongles
The ones offering a gdb compatible interface, over a serial port
or an Ethernet connection. gdb can directly connect to them.
The ones not offering a gdb compatible interface are generally
supported by OpenOCD (Open On Chip Debugger):
http://openocd.sourceforge.net/
OpenOCD is the bridge between the gdb debugging language
and the JTAG interface of the target CPU.
See the very complete documentation: http://openocd.
sourceforge.net/documentation/online-docs/
For each board, you’ll need an OpenOCD configuration file
(ask your supplier)

More Kernel Debugging Tips
Enable CONFIG_KALLSYMS_ALL
General Setup -
> Configure standard kernel features
To get oops messages with symbol names instead of raw
addresses
This obsoletes the ksymoops tool
On ARM, if your kernel doesn’t boot or hangs without any
message, you can activate early debugging options
(CONFIG_DEBUG_LL and CONFIG_EARLYPRINTK), and add
earlyprintk to the kernel command line.
Techniques to locate the C instruction which caused an oops:
http://j.mp/18oMRHx

Kernel Crash Analysis with kexec/kdump
kexec system call: makes it
possible to call a new kernel,
without rebooting and going
through the BIOS / ﬁrmware.
Idea: after a kernel panic, make
the kernel automatically execute a
new, clean kernel from a reserved
location in RAM, to perform
post-mortem analysis of the
memory of the crashed kernel.
See Documentation/kdump/
kdump.txt in the kernel sources
for details.

Practical lab - Kernel debugging
Use the dynamic printk feature.
Add debugfs entries
Load a broken driver and see it
crash
Analyze the error information
dumped by the kernel.
Disassemble the code and locate
the exact C instruction which
caused the failure.

Porting the Linux Kernel to an ARM Board
Porting the Linux
Kernel to an ARM
Board
Free Electrons
Embedded Linux
Developers
Free Electrons

Porting the Linux kernel
The Linux kernel supports a lot of different CPU architectures
Each of them is maintained by a different group of
contributors
See the MAINTAINERS file for details
The organization of the source code and the methods to port
the Linux kernel to a new board are therefore very
architecture-dependent
For example, some architectures use the Device Tree, some do
not.
This presentation is focused on the ARM architecture only

Architecture, CPU and Machine
In the source tree, each architecture has its own directory
arch/arm for the ARM architecture
This directory contains generic ARM code
boot, common, configs, kernel, lib, mm, nwfpe, vfp,
oprofile, tools
And many directories for different SoC families
mach-* directories: mach-pxa for PXA CPUs, mach-imx for
Freescale iMX CPUs, etc.
Before the ARM cleanup, these directories contained support
for the SoC family (GPIO, clocks, pinmux, power
management, interrupt controller, etc.) and for the various
boards.
Nowadays, they contain a lot less code, essentially a small
SoC description file, power management and SMP code.
Some CPU types share some code, in directories named
plat-*
Device Tree source files in arch/arm/boot/dts.

Before the Device Tree and ARM cleanup
Until 2011, the ARM architecture wasn’t using the Device
Tree, and a large portion of the SoC support was located in
arch/arm/mach-<foo>.
Each board supported by the kernel was associated to an
unique machine ID.
The entire list of machine ID can be downloaded at
http://www.arm.linux.org.uk/developer/machines/
download.php and one could freely register an additional one.
The Linux kernel was deﬁning a machine structure for each
board, which associates the machine ID with a set of
information and callbacks.
The bootloader had to pass the machine ID to the kernel in a
speciﬁc ARM register.

The Device Tree and the ARM cleanup
As the ARM architecture gained signiﬁcantly in popularity,
some major refactoring was needed.
First, the Device Tree was introduced on ARM: instead of
using C code to describe SoCs and boards, a specialized
language is used.
Second, many driver infrastructures were created to replace
custom code in arch/arm/mach-<foo>:
The common clock framework in drivers/clk
The pinctrl subsystem in drivers/pinctrl
The irqchip subsystem in drivers/irqchip
The clocksource subsystem in drivers/clocksource
The amount of code in mach-<foo> has now signiﬁcantly
reduced.

Adding the support for a new ARM board
Provided the SoC used on your board is supported by the Linux
kernel:
1. Create a Device Tree ﬁle in arch/arm/boot/dts, generally
named <soc-name>-<board-name>.dts, and make it
include the relevant SoC .dtsi ﬁle.
Your Device Tree will describe all the SoC peripherals that are
enabled, the pin muxing, as well as all the devices on the
board.
2. Modify arch/arm/boot/dts/Makefile to make sure your
Device Tree gets built as a DTB during the kernel build.
3. If needed, develop the missing device drivers for the devices
that are on your board outside the SoC.

Example of the Freescale iMX28 SoCs
The hardware platform used in this training is based on the
AM335x processor from Texas Instruments.
This platform inherits from the OMAP family of TI, for which
kernel support has been around for a long time.
Due to this, and the complexity of the platform, the AM335x
and OMAP support in the kernel hasn’t fully migrated yet to
all the infrastructures created during the ARM cleanup.
Therefore, to illustrate this section, we will take the example
of the Freescale iMX28 platform, on which Free Electrons has
worked speciﬁcally.

Studying the Crystalfontz CFA-10036 platform
Crystalfontz CFA-10036
Uses the Freescale iMX28 SoC, from the
MXS family.
128MB of RAM
1 serial port, 1 LED
1 I2C bus, equipped with an OLED display
1 SD-Card slot

Crystalfontz CFA-10036 Device Tree, header
Mandatory Device Tree language definition
/dts-v1/
Include the .dtsi file describing the SoC
#include "im28.dtsi"
Start the root of the tree
/ {
A human-readable string to describe the machine
model = "Crystalfontz CFA-10036 Board";
A list of compatible strings, from the most specific one to the
most general one. Can be used by kernel code to do a SoC or
board-specific check.
compatible = "crystalfontz,cfa10036", "fsl,imx28";

Crystalfontz CFA-10036 Device Tree, chosen/memory
Definition of the default kernel command line. Some
additional operating-system specific entries can be added in
chosen:
chosen {
bootargs = "console=ttyS0,115200 earlyprintk";
};
Definition of the size and location of the RAM:
memory {
device_type = "memory";
reg = <0x40000000 0x8000000>; /* 128 MB */
};

Crystalfontz CFA-10036, bus/UART
Start of the internal SoC peripherals.
apb@80000000 {
apbh@80000000 {
apbx@80040000 {
The CFA-10036 has one debug UART, so the corresponding
controller is enabled:
duart: serial@80074000 {
pinctrl-0 = <&duart_pins_b>;
status = "okay";
};

Crystalfontz CFA-10036 Device Tree, Muxing
Deﬁnition of a few pins that will be muxed as GPIO, for LEDs
and reset.
pinctrl@80018000 {
ssd1306_cfa10036: ssd1306-10036@0 {
reg = <0>;
fsl,pinmux-ids = <
0x2073 /* MX28_PAD_SSP0_D7__GPIO_2_7 */
>;
fsl,drive-strength = <0>;
fsl,voltage = <1>;
fsl,pull-up = <0>;
};
led_pins_cfa10036: leds-10036@0 {
reg = <0>;
fsl,pinmux-ids = <
0x3043 /* MX28_PAD_AUART1_RX__GPIO_3_4 */
>;
fsl,drive-strength = <0>;
fsl,voltage = <1>;
fsl,pull-up = <0>;
};
};

Crystalfontz CFA-10036 Device Tree, LED
One LED is connected to this platform. Note the reference to
the led_pins_cfa10036 muxing conﬁguration.
leds {
compatible = "gpio-leds";
pinctrl-0 = <&led_pins_cfa10036>;
power {
gpios = <&gpio3 4 1>;
default-state = "on";
};
};

Crystalfontz CFA-10036 Device Tree, SD Card/USB
The platform also has a USB port
usb0: usb@80080000 {
pinctrl-0 = <&usb0_otg_cfa10036>;
status = "okay";
};
and an SD Card slot:
ssp0: ssp@80010000 {
compatible = "fsl,imx28-mmc";
pinctrl-0 = <&mmc0_4bit_pins_a
&mmc0_cd_cfg &mmc0_sck_cfg>;
bus-width = <4>;
status = "okay";
};

Crystalfontz CFA-10036 Device Tree, I2C bus
An I2C bus, with a Solomon SSD1306 OLED display
connected on it:
i2c0: i2c@80058000 {
pinctrl-0 = <&i2c0_pins_b>;
status = "okay";
ssd1306: oled@3c {
compatible = "solomon,ssd1306fb-i2c";
pinctrl-0 = <&ssd1306_cfa10036>;
reg = <0x3c>;
reset-gpios = <&gpio2 7 0>;
solomon,height = <32>;
solomon,width = <128>;
solomon,page-offset = <0>;
};
};

Crystalfontz CFA-10036 Device Tree, Breakout Boards
The CFA-10036 can be plugged in other breakout boards, and
the device tree also allows us to describe this, using includes.
For example, the CFA-10057:
#include "imx28-cfa10036.dts"
This allows to have a layered description. This can also be
done for boards that have a lot in common, like the
BeagleBone and the BeagleBone Black, or the AT91
SAMA5D3-based boards.

Crystalfontz CFA-10036: build the DTB
To ensure that the Device Tree Blob gets built for this board
Device Tree Source, one need to ensure it is listed in
arch/arm/boot/dts/Makefile:
dtb-$(CONFIG_ARCH_MXS) += imx28-cfa10036.dtb
imx28-cfa10037.dtb
imx28-cfa10049.dtb
imx28-cfa10055.dtb
imx28-cfa10056.dtb
imx28-cfa10057.dtb
imx28-cfa10058.dtb
imx28-evk.dtb

Understanding the SoC support
Let’s consider another ARM platform here, the Marvell
Armada 370/XP.
For this platform, the core of the SoC support is located in
arch/arm/mach-mvebu
The armada-370-xp.c (see code on the next slide) contains
the ”entry point” of the SoC definition, the
DT_MACHINE_START .. MACHINE_END definition:
Defines the list of platform compatible strings that will match
this platform, in this case marvell,armada-370-xp. This
allows the kernel to know which DT_MACHINE structure to use
depending on the DTB that is passed at boot time.
Defines various callbacks for the platform initialization, the
most important one being the .init_machine callback, which
calls of_platform_populate(). This function travels
through the Device Tree and instantiates all the devices.

arch/arm/mach-mvebu/armada-370-xp.c
static void __init armada_370_xp_dt_init(void)
{
of_platform_populate(NULL, of_default_bus_match_table, NULL, NULL);
}
static const char * const armada_370_xp_dt_compat[] = {
"marvell,armada-370-xp",
NULL,
};
DT_MACHINE_START(ARMADA_XP_DT, "Marvell Armada 370/XP (Device Tree)")
.smp = smp_ops(armada_xp_smp_ops),
.init_machine = armada_370_xp_dt_init,
.map_io = armada_370_xp_map_io,
.init_time = armada_370_xp_timer_and_clk_init,
.restart = mvebu_restart,
.dt_compat = armada_370_xp_dt_compat,
MACHINE_END

Components of the minimal SoC support
The minimal SoC support consists in
An SoC entry point ﬁle, arch/arm/mach-mvebu/armada-370-xp.c
At least one SoC .dtsi DT and one board .dts DT, in
arch/arm/boot/dts
A interrupt controller driver, drivers/irqchip/irq-armada-370-xp.c
A timer driver, drivers/clocksource/time-armada-370-xp.c
An earlyprintk implementation to get early messages from the console,
arch/arm/Kconfig.debug and arch/arm/include/debug
A serial port driver in drivers/tty/serial. For Armada 370/XP, the
8250 driver drivers/tty/serial/8250 is used.
This allows to boot a minimal system up to user space, using a
root ﬁlesystem in initramfs.

Extending the minimal SoC support
Once the minimal SoC support is in place, the following core
components should be added:
Support for the clocks. Usually requires some clock drivers, as
well as DT representations of the clocks. See
drivers/clk/mvebu for Armada 370/XP clock drivers.
Support for pin muxing, through the pinctrl subsystem. See
drivers/pinctrl/mvebu for the Armada 370/XP drivers.
Support for GPIOs, through the GPIO subsystem. See
drivers/gpio/gpio-mvebu.c for the Armada 370/XP
GPIO driver.
Support for SMP, through struct smp_operations. See
arch/arm/mach-mvebu/platsmp.c.

Adding device drivers
Once the core pieces of the SoC support have been implemented,
the remaining part is to add drivers for the diﬀerent hardware
blocks:
Ethernet driver, in drivers/net/ethernet/mvneta.c
SATA driver, in drivers/ata/sata_mv.c
I2C driver, in drivers/i2c/busses/i2c-mv64xxx.c
SPI driver, in drivers/spi/spi-orion.c
PCIe driver, in drivers/pci/host/pci-mvebu.c
USB driver, in drivers/usb/host/ehci-orion.c
etc.

Power Management
Power
Management
Free Electrons
Embedded Linux
Developers
Free Electrons

PM Building Blocks
Several power management building blocks
Suspend and resume
CPUidle
Runtime power management
Frequency and voltage scaling
Applications
Independent building blocks that can be improved gradually
during development

Clock Framework (1)
Generic framework to manage clocks used by devices in the
system
Allows to reference count clock users and to shutdown the
unused clocks to save power
Simple API described in
Documentation/DocBook/kernel-api/clk.html .
clk_get() to get a reference to a clock
clk_enable() to start the clock
clk_disable() to stop the clock
clk_put() to free the clock source
clk_get_rate() to get the current rate

Clock Framework (2)
The common clock framework
Allows to declare the available clocks and their association to
devices in the Device Tree (preferred) or statically in the
source code (old method)
Provides a debugfs representation of the clock tree
Is implemented in drivers/clk

Diagram overview of the common clock framework

Clock Framework (3)
The interface of the CCF divided into two halves:
Common Clock Framework core
Common definition of struct clk
Common implementation of the clk.h API (defined in
drivers/clk/clk.c)
struct clk_ops: operations invoked by the clk API
implementation
Not supposed to be modified when adding a new driver
Hardware-specific
Callbacks registered with struct clk_ops and the
corresponding hardware-specific structures
Has to be written for each new hardware clock

Clock Framework (4)
Hardware clock operations: device tree
The device tree is the mandatory way to declare a clock and
to get its resources, as for any other driver using DT we have
to:
Parse the device tree to setup the clock: the resources but
also the properties are retrieved.
Declare the compatible clocks and associate it with an
initialization function using CLK_OF_DECLARE

Suspend and Resume
Infrastructure in the kernel to support suspend and resume
Platform hooks
prepare(), enter(), finish(), valid() in a
struct platform_suspend_ops structure
Registered using the suspend_set_ops() function
See arch/arm/mach-at91/pm.c
Device drivers
suspend() and resume() hooks in the *_driver structures
(struct platform_driver, struct usb_driver, etc.)
See drivers/net/macb.c

Board-speciﬁc Power Management
Typically takes care of battery and charging management.
Also deﬁnes presuspend and postsuspend handlers.
Example: arch/arm/mach-pxa/spitz_pm.c

arch/arm/mach-cpu/sleep.S
Assembly code implementing CPU speciﬁc suspend and
resume code.
Note: only found on arm, just 3 other occurrences in other
architectures, with other paths.
First scenario: only a suspend function. The code goes in
sleep state (after enabling DRAM self-refresh), and continues
with resume code.
Second scenario: suspend and resume functions. Resume
functions called by the bootloader.
Examples to look at:
arch/arm/mach-omap2/sleep24xx.S (1st case)
arch/arm/mach-pxa/sleep.S (2nd case)

Triggering Suspend
Whatever the power management implementation, CPU
speciﬁc struct suspend_ops functions are called by the
enter_state() function.
enter_state() also takes care of executing the suspend and
resume functions for your devices.
The execution of this function can be triggered from user
space. To suspend to RAM:
echo mem > /sys/power/state
Can also use the s2ram program from
http://suspend.sourceforge.net/
Read kernel/power/suspend.c

Runtime Power Management
According to the kernel conﬁguration interface: Enable
functionality allowing I/O devices to be put into energy-saving
(low power) states at run time (or autosuspended) after a
speciﬁed period of inactivity and woken up in response to a
hardware-generated wake-up event or a driver’s request.
New hooks must be added to the drivers:
runtime_suspend(), runtime_resume(), runtime_idle()
API and details on Documentation/power/runtime_pm.txt
See also Kevin Hilman’s presentation at ELC Europe 2010:
http://elinux.org/images/c/cd/ELC-2010-khilman-
Runtime-PM.odp

Saving Power in the Idle Loop
The idle loop is what you run when there’s nothing left to run
in the system.
Implemented in all architectures in
arch/<arch>/kernel/process.c
Example to read: look for cpu_idle in
arch/arm/kernel/process.c
Each ARM cpu deﬁnes its own arch_idle function.
The CPU can run power saving HLT instructions, enter NAP
mode, and even disable the timers (tickless systems).
See also http://en.wikipedia.org/wiki/Idle_loop

Managing Idle
Adding support for multiple idle levels
Modern CPUs have several sleep states offering different power
savings with associated wake up latencies
Since 2.6.21, the dynamic tick feature allows to remove the
periodic tick to save power, and to know when the next event
is scheduled, for smarter sleeps.
CPUidle infrastructure to change sleep states
Platform-specific driver defining sleep states and transition
operations
Platform-independent governors (ladder and menu)
Available for x86/ACPI, not supported yet by all ARM cpus.
(look for cpuidle* files under arch/arm/)
See Documentation/cpuidle/ in kernel sources.

PowerTOP
https://01.org/powertop/
With dynamic ticks, allows to ﬁx parts of kernel code and
applications that wake up the system too often.
PowerTOP allows to track the worst oﬀenders
Now available on ARM cpus implementing CPUidle
Also gives you useful hints for reducing power.

Frequency and Voltage Scaling (1)
Frequency and voltage scaling possible through the cpufreq
kernel infrastructure.
Generic infrastructure: drivers/cpufreq/cpufreq.c and
include/linux/cpufreq.h
Generic governors, responsible for deciding frequency and
voltage transitions
performance: maximum frequency
powersave: minimum frequency
ondemand: measures CPU consumption to adjust frequency
conservative: often better than ondemand. Only increases
frequency gradually when the CPU gets loaded.
userspace: leaves the decision to a user space daemon.
This infrastructure can be controlled from
/sys/devices/system/cpu/cpu<n>/cpufreq/

Frequency and Voltage Scaling (2)
CPU support code in architecture dependent ﬁles. Example to
read: arch/arm/plat-omap/cpu-omap.c
Must implement the operations of the cpufreq_driver
structure and register them using
cpufreq_register_driver()
init() for initialization
exit() for cleanup
verify() to verify the user-chosen policy
setpolicy() or target() to actually perform the frequency
change
See Documentation/cpu-freq/ for useful explanations

PM Quality Of Service interface
Kernel and user mode interface for registering performance
expectations by drivers, subsystems and user space
applications.
Two diﬀerent PM QoS frameworks are available:
PM QoS classes for CPU DMA latency, network latency and
and network throughput.
The per-device PM QoS framework API to manage per-device
latency.
According to these requirements, PM QoS allows kernel
drivers to adjust their power management
See Documentation/power/pm_qos_interface.txt
Still in very early deployment (only used in about 15 drivers in
3.12).

Regulator Framework
Modern embedded hardware have hardware responsible for
voltage and current regulation
The regulator framework allows to take advantage of this
hardware to save power when parts of the system are unused
A consumer interface for device drivers (i.e users)
Regulator driver interface for regulator drivers
Machine interface for board conﬁguration
sysfs interface for user space
Merged in Linux 2.6.27.
See Documentation/power/regulator/ in kernel sources.
See Liam Girdwood’s presentation at ELC 2008
http://free-electrons.com/blog/elc-2008-
report#girdwood

BSP Work for a New Board
In case you just need to create a BSP for your board, and your
CPU already has full PM support, you should just need to:
Create clock definitions and bind your devices to them.
Implement PM handlers (suspend, resume) in the drivers for
your board specific devices.
Implement runtime PM handlers in your drivers.
Implement board specific power management if needed (mainly
battery management)
Implement regulator framework hooks for your board if needed.
All other parts of the PM infrastructure should be already
there: suspend / resume, cpuidle, cpu frequency and voltage
scaling.

Useful Resources
Documentation/power/ in the Linux kernel sources.
Will give you many useful details.
http://wiki.linaro.org/WorkingGroups/PowerManagement/
Ongoing developments on the ARM platform.
Tips and ideas for prolonging battery life
http://j.mp/fVdxKh

The kernel development and contribution process
The kernel
development and
contribution
process
Free Electrons
Embedded Linux
Developers
Free Electrons

Linux versioning scheme and
development process

Until 2.6 (1)
One stable major branch every 2 or 3 years
Identiﬁed by an even middle number
Examples: 1.0.x, 2.0.x, 2.2.x, 2.4.x
One development branch to integrate new functionalities and
major changes
Identiﬁed by an odd middle number
Examples: 2.1.x, 2.3.x, 2.5.x
After some time, a development version becomes the new base
version for the stable branch
Minor releases once in while: 2.2.23, 2.5.12, etc.

Until 2.6 (2)

Changes since Linux 2.6
Since 2.6.0, kernel developers have been able to introduce
lots of new features one by one on a steady pace, without
having to make disruptive changes to existing subsystems.
Since then, there has been no need to create a new
development branch massively breaking compatibility with the
stable branch.
Thanks to this, more features are released to users at a
faster pace.

3.x stable branch
From 2003 to 2011, the official kernel versions were named
2.6.x.
Linux 3.0 was released in July 2011
This is only a change to the numbering scheme
Official kernel versions are now named 3.x (3.0, 3.1, 3.2,
etc.)
Stabilized versions are named 3.x.y (3.0.2, 3.4.3, etc.)
It effectively only removes a digit compared to the previous
numbering scheme

New development model
Using merge and bug ﬁxing windows

New development model - Details
After the release of a 3.x version (for example), a two-weeks
merge window opens, during which major additions are
merged.
The merge window is closed by the release of test version
3.(x+1)-rc1
The bug fixing period opens, for 6 to 10 weeks.
At regular intervals during the bug fixing period,
3.(x+1)-rcY test versions are released.
When considered sufficiently stable, kernel 3.(x+1) is
released, and the process starts again.

More stability for the kernel source tree
Issue: bug and security ﬁxes only released
for most recent stable kernel versions.
Some people need to have a recent kernel,
but with long term support for security
updates.
You could get long term support from a
commercial embedded Linux provider.
You could reuse sources for the kernel
used in Ubuntu Long Term Support
releases (5 years of free security updates).
The http://kernel.org front page
shows which versions will be supported for
some time (up to 2 or 3 years), and which
ones won’t be supported any more
(”EOL: End Of Life”)

What’s new in each Linux release?
The official list of changes for each Linux release is just a
huge list of individual patches!
commit aa6e52a35d388e730f4df0ec2ec48294590cc459
Author: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Date: Wed Jul 13 11:29:17 2011 +0200
at91: at91-ohci: support overcurrent notification
Several USB power switches (AIC1526 or MIC2026) have a digital output
that is used to notify that an overcurrent situation is taking
place. This digital outputs are typically connected to GPIO inputs of
the processor and can be used to be notified of these overcurrent
situations.
Therefore, we add a new overcurrent_pin[] array in the at91_usbh_data
structure so that boards can tell the AT91 OHCI driver which pins are
used for the overcurrent notification, and an overcurrent_supported
boolean to tell the driver whether overcurrent is supported or not.
The code has been largely borrowed from ohci-da8xx.c and
ohci-s3c2410.c.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Very difficult to find out the key changes and to get the global
picture out of individual changes.
Fortunately, there are some useful resources available
http://wiki.kernelnewbies.org/LinuxChanges
http://lwn.net
http://linuxfr.org, for French readers

Contributing to the Linux kernel

Solving Issues
If you face an issue, and it doesn’t look speciﬁc to your work
but rather to the tools you are using, it is very likely that
someone else already faced it.
Search the Internet for similar error reports.
You have great chances of ﬁnding a solution or workaround, or
at least an explanation for your issue.
Otherwise, reporting the issue is up to you!

Getting Help
If you have a support contract, ask your vendor.
Otherwise, don’t hesitate to share your questions and issues
Either contact the Linux mailing list for your architecture (like
linux-arm-kernel or linuxsh-dev...).
Or contact the mailing list for the subsystem you’re dealing
with (linux-usb-devel, linux-mtd...). Don’t ask the maintainer
directly!
Most mailing lists come with a FAQ page. Make sure you read
it before contacting the mailing list.
Useful IRC resources are available too (for example on
http://kernelnewbies.org).
Refrain from contacting the Linux Kernel mailing list, unless
you’re an experienced developer and need advice.

Reporting Linux Bugs
First make sure you’re using the latest version
Make sure you investigate the issue as much as you can: see
Documentation/BUG-HUNTING
Check for previous bugs reports. Use web search engines,
accessing public mailing list archives.
If the subsystem you report a bug on has a mailing list, use it.
Otherwise, contact the oﬃcial maintainer (see the
MAINTAINERS ﬁle). Always give as many useful details as
possible.

How to Become a Kernel Developer?
Recommended resources
See Documentation/SubmittingPatches for guidelines and
http://kernelnewbies.org/UpstreamMerge for very
helpful advice to have your changes merged upstream (by Rik
van Riel).
Watch the Write and Submit your ﬁrst Linux kernel Patch talk
by Greg. K.H:
http://www.youtube.com/watch?v=LLBrBBImJt4
How to Participate in the Linux Community (by Jonathan
Corbet) A Guide To The Kernel Development Process
http://j.mp/tX2Ld6

Contribute to the Linux Kernel (1)
Clone Linus Torvalds’ tree:
Keep your tree up to date
git pull
Look at the master branch and check whether your issue /
change hasn’t been solved / implemented yet. Also check the
maintainer’s git tree and mailing list (see the MAINTAINERS
ﬁle).You may miss submissions that are not in mainline yet.
If the maintainer has its own git tree, create a remote branch
tracking this tree. This is much better than creating another
clone (doesn’t duplicate common stuﬀ):
git remote add linux-omap git://git.kernel.org/
pub/scm/linux/kernel/git/tmlind/linux-omap.git
git fetch linux-omap

Either create a new branch starting from the current commit
in the master branch:
git checkout -b feature
Or, if more appropriate, create a new branch starting from the
maintainer’s master branch:
git checkout -b feature linux-omap/master (remote
tree / remote branch)
In your new branch, implement your changes.
Test your changes (must at least compile them).
Run git add to add any new files to the index.
Check that each file you modified is ready for submission:
scripts/checkpatch.pl --strict --file <file>
If needed, fix indenting rule violations:
indent -linux <file>

Configure git send-email
Make sure you already have configured your name and e-mail
address (should be done before the first commit).
git config --global user.name ’My Name’
git config --global user.email me@mydomain.net
Configure your SMTP settings. Example for a Google Mail
account:
git config --
global sendemail.smtpserver smtp.googlemail.com
git config --global sendemail.smtpserverport 587
git config --global sendemail.smtpencryption tls
git config --
global sendemail.smtpuser jdoe@gmail.com
git config --global sendemail.smtppass xxx

Group your changes by sets of logical changes, corresponding
to the set of patches that you wish to submit.
Commit and sign these groups of changes (signing required by
Linux developers).
git commit -s
Make sure your first description line is a useful summary and
starts with the name of the modified subsystem. This first
description line will appear in your e-mails
The easiest way is to look at previous commit summaries on
the main file you modify
git log --pretty=oneline <path-to-file>
Examples subject lines ([PATCH] omitted):
Documentation: prctl/seccomp_filter
PCI: release busn when removing bus
ARM: add support for xz kernel decompression

Remove previously generated patches
rm 00*.patch
Have git generate patches corresponding to your branch
If your branch is based on mainline
git format-patch master..<your branch>
If your branch is based on a remote branch
git format-patch <remote>/<branch>..<your branch>
You can run a last check on all your patches (easy)
scripts/checkpatch.pl --strict 00*.patch
Now, send your patches to yourself
git send-email --compose --
to me@mydomain.com 00*.patch
If you have just one patch, or a trivial patch, you can remove
the empty line after In-Reply-To:. This way, you won’t add
a summary e-mail introducing your changes (recommended
otherwise).

Check that you received your e-mail properly, and that it looks
good.
Now, ﬁnd the maintainers for your patches
scripts/get_maintainer.pl ~/patches/00*.patch
Russell King <linux@arm.linux.org.uk> (maintainer:ARM PORT)
Nicolas Pitre <nicolas.pitre@linaro.org>
(commit_signer:1/1=100%)
linux-arm-kernel@lists.infradead.org (open list:ARM PORT)
linux-kernel@vger.kernel.org (open list)
Now, send your patches to each of these people and lists
git send-email --compose --to linux@arm.linux.
org.uk --to nicolas.pitre@linaro.org --to linux-
arm-kernel@lists.infradead.org --to linux-
kernel@vger.kernel.org 00*.patch
Wait for replies about your changes, take the comments into
account, and resubmit if needed, until your changes are
eventually accepted.

If you use git format-patch to produce your patches, you
will need to update your branch and may need to group your
changes in a diﬀerent way (one patch per commit).
Here’s what we recommend
Update your master branch
git checkout master; git pull
Back to your branch, implement the changes taking
community feedback into account. Commit these changes.
Still in your branch: reorganize your commits and commit
messages
git rebase --interactive origin/master
git rebase allows to rebase (replay) your changes starting
from the latest commits in master. In interactive mode, it also
allows you to merge, edit and even reorder commits, in an
interactive way.
Third, generate the new patches with git format-patch.

Kernel Resources
Kernel Resources
Free Electrons
Embedded Linux
Developers
Free Electrons

Kernel Development News
Linux Weekly News
http://lwn.net/
The weekly digest oﬀ all Linux and free software information
sources
In depth technical discussions about the kernel
Subscribe to ﬁnance the editors ($7 / month)
Articles available for non subscribers after 1 week.

Useful Reading (1)
Essential Linux Device Drivers, April 2008
http://elinuxdd.com/
By Sreekrishnan Venkateswaran, an
embedded IBM engineer with more than
10 years of experience
Covers a wide range of topics not covered
by LDD: serial drivers, input drivers, I2C,
PCMCIA and Compact Flash, PCI, USB,
video drivers, audio drivers, block drivers,
network drivers, Bluetooth, IrDA, MTD,
drivers in user space, kernel debugging,
etc.
Probably the most wide ranging and
complete Linux device driver book I’ve
read – Alan Cox

Useful Reading (2)
Writing Linux Device drivers, September 2009
http://www.coopj.com/
Self published by Jerry Cooperstein
Available like any other book (Amazon
and others)
Though not as thorough as the previous
book on speciﬁc drivers, still a good
complement on multiple aspects of kernel
and device driver development.
Based on Linux 2.6.31
Multiple exercises. Updated solutions for
2.6.36.

Useful Reading (3)
Linux Device Drivers, 4th edition, July 2015
(estimated)
http://shop.oreilly.com/product/
0636920030867.do
By Jonathan Corbet, Alessandro Rubini,
Greg Kroah-Hartman, Jessica McKellar,
O’Reilly
Expected to be a great book, if as good
as the previous edition (Free PDF:
community/kernel/ldd3/), which is
now out of date.

Useful Reading (4)
Linux Kernel Development, 3rd Edition,
Jun 2010
Robert Love, Novell Press
http://free-electrons.com/redir/
lkd3-book.html
A very synthetic and pleasant way to
learn about kernel subsystems (beyond
the needs of device driver writers)
The Linux Programming Interface, Oct
2010
Michael Kerrisk, No Starch Press
http://man7.org/tlpi/
A gold mine about the kernel interface
and how to use it

Useful Online Resources
Kernel documentation (Documentation/ in kernel sources)
Available on line:
http://free-electrons.com/kerneldoc/ (with HTML
documentation extracted from source code)
Linux kernel mailing list FAQ
http://www.tux.org/lkml/
Complete Linux kernel FAQ
Read this before asking a question to the mailing list
Kernel Newbies
http://kernelnewbies.org/
Glossary, articles, presentations, HOWTOs, recommended
reading, useful tools for people getting familiar with Linux
kernel or driver development.
Kernel glossary
http://kernelnewbies.org/KernelGlossary

International Conferences
Embedded Linux Conference:
http://embeddedlinuxconference.com/
Organized by the Linux Foundation:
in California (San Francisco, in the spring)
in Europe (October-November)
Very interesting kernel and user space topics for embedded
systems developers.
Presentation slides freely available
Linux Plumbers: http://linuxplumbersconf.org
Conference on the low-level plumbing of Linux: kernel, audio,
power management, device management, multimedia, etc.
linux.conf.au: http://linux.org.au/conf/
In Australia / New Zealand
Features a few presentations by key kernel hackers.
Don’t miss our free conference videos on http://free-
electrons.com/community/videos/conferences/

ARM resources
ARM Linux project: http://www.arm.linux.org.uk/
Developer documentation:
http://www.arm.linux.org.uk/developer/
linux-arm-kernel mailing list:
http://lists.infradead.org/mailman/listinfo/linux-
arm-kernel
FAQ:
http://www.arm.linux.org.uk/armlinux/mlfaq.php
Linaro: http://linaro.org
Many optimizations and resources for recent ARM CPUs
(toolchains, kernels, debugging utilities...).
ARM Limited: http://www.linux-arm.com/
Wiki with links to useful developer resources
See our Embedded Linux course for details about toolchains:
http://free-electrons.com/training/embedded-linux/

Last slides
Last slides
Free Electrons
Embedded Linux
Developers
Free Electrons

Last slide
Thank you!
And may the Source be with you

Backup slides
Backup slides
Free Electrons
Embedded Linux
Developers
Free Electrons

Backup slides
DMA

DMA Integration

Peripheral DMA

DMA Controllers

DMA descriptors

Backup slides
DMA Usage

Constraints with a DMA
A DMA deals with physical addresses, so:
Programming a DMA requires retrieving a physical address at
some point (virtual addresses are usually used)
The memory accessed by the DMA shall be physically
contiguous
The CPU can access memory through a data cache
Using the cache can be more efficient (faster accesses to the
cache than the bus)
But the DMA does not access to the CPU cache, so one need
to take care of cache coherency (cache content vs memory
content)
Either flush or invalidate the cache lines corresponding to the
buffer accessed by DMA and processor at strategic times

DMA Memory Constraints
Need to use contiguous memory in physical space.
Can use any memory allocated by kmalloc() (up to 128 KB)
or __get_free_pages() (up to 8MB).
Can use block I/O and networking buﬀers, designed to
support DMA.
Can not use vmalloc() memory (would have to setup DMA
on each individual physical page).

Memory Synchronization Issues
Memory caching could interfere with DMA
Before DMA to device
Need to make sure that all writes to DMA buffer are
committed.
After DMA from device
Before drivers read from DMA buffer, need to make sure that
memory caches are flushed.
Bidirectional DMA
Need to flush caches before and after the DMA transfer.

Linux DMA API
The kernel DMA utilities can take care of:
Either allocating a buffer in a cache coherent area,
Or making sure caches are flushed when required,
Managing the DMA mappings and IOMMU (if any).
See Documentation/DMA-API.txt for details about the
Linux DMA generic API.
Most subsystems (such as PCI or USB) supply their own DMA
API, derived from the generic one. May be sufficient for most
needs.

Coherent or Streaming DMA Mappings
Coherent mappings
The kernel allocates a suitable buffer and sets the mapping for
the driver.
Can simultaneously be accessed by the CPU and device.
So, has to be in a cache coherent memory area.
Usually allocated for the whole time the module is loaded.
Can be expensive to setup and use on some platforms.
Streaming mappings
The kernel just sets the mapping for a buffer provided by the
driver.
Use a buffer already allocated by the driver.
Mapping set up for each transfer. Keeps DMA registers free on
the hardware.
The recommended solution.

Allocating Coherent Mappings
The kernel takes care of both buﬀer allocation and mapping
#include <asm/dma-mapping.h>
void * /* Output: buffer address */
dma_alloc_coherent(
struct device *dev, /* device structure */
size_t size, /* Needed buffer size in bytes */
dma_addr_t *handle, /* Output: DMA bus address */
gfp_t gfp /* Standard GFP flags */
);
void dma_free_coherent(struct device *dev,
size_t size, void *cpu_addr, dma_addr_t handle);

Setting up streaming mappings
Works on buﬀers already allocated by the driver
#include <linux/dmapool.h>
dma_addr_t dma_map_single(
struct device *, /* device structure */
void *, /* input: buffer to use */
size_t, /* buffer size */
enum dma_data_direction /* Either DMA_BIDIRECTIONAL,
* DMA_TO_DEVICE or
* DMA_FROM_DEVICE */
);
void dma_unmap_single(struct device *dev, dma_addr_t handdle,
size_t size, enum dma_data_direction dir);

DMA Streaming Mapping Notes
When the mapping is active: only the device should access
the buffer (potential cache issues otherwise).
The CPU can access the buffer only after unmapping!
Another reason: if required, this API can create an
intermediate bounce buffer (used if the given buffer is not
usable for DMA).
The Linux API also supports scatter / gather DMA streaming
mappings.

Backup slides
DMA transfers

Starting DMA transfers
If the device you’re writing a driver for is doing peripheral
DMA, no external API is involved.
If it relies on an external DMA controller, you’ll need to
Ask the hardware to use DMA, so that it will drive its request
line
Use Linux DMAEngine framework, especially its slave API

DMAEngine Slave API
In order to start a DMA transfer, you need to call the
following functions from your driver
1. Request a channel for exclusive use with
dma_request_channel, or one of its variants
2. Configure it for our use case, by filling a dma_slave_config
structures with various parameters (source and destination
adresses, accesses width, etc.) and passing it as an argument
to dmaengine_slave_config
3. Start a new transaction with dmaengine_prep_slave_single
or dmaengine_prep_slave_sg
4. Put the transaction in the driver pending queue using
dmaengine_submit
5. And finally ask the driver to process all pending transactions
using dmaengine_issue_pending
Of course, this needs to be done in addition to the DMA
mapping seen previously
Some framework abstract it away from you, such as SPI and
ASoC

Backup slides
mmap

mmap
Possibility to have parts of the virtual address space of a
program mapped to the contents of a file
Particularly useful when the file is a device file
Allows to access device I/O memory and ports without having
to go through (expensive) read, write or ioctl calls
One can access to current mapped files by two means:
/proc/<pid>/maps
pmap <pid>

/proc/<pid>/maps
start-end perm offset major:minor inode mapped file name
...
7f4516d04000-7f4516d06000 rw-s 1152a2000 00:05 8406 /dev/dri/card0
7f4516d07000-7f4516d0b000 rw-s 120f9e000 00:05 8406 /dev/dri/card0
...
7f4518728000-7f451874f000 r-xp 00000000 08:01 268909 /lib/x86_64-linux-gnu/libexpat.so.1.5.2
7f451874f000-7f451894f000 ---p 00027000 08:01 268909 /lib/x86_64-linux-gnu/libexpat.so.1.5.2
7f451894f000-7f4518951000 r--p 00027000 08:01 268909 /lib/x86_64-linux-gnu/libexpat.so.1.5.2
7f4518951000-7f4518952000 rw-p 00029000 08:01 268909 /lib/x86_64-linux-gnu/libexpat.so.1.5.2
...
7f451da4f000-7f451dc3f000 r-xp 00000000 08:01 1549 /usr/bin/Xorg
7f451de3e000-7f451de41000 r--p 001ef000 08:01 1549 /usr/bin/Xorg
7f451de41000-7f451de4c000 rw-p 001f2000 08:01 1549 /usr/bin/Xorg
...

mmap Overview

How to Implement mmap - User Space
Open the device ﬁle
Call the mmap system call (see man mmap for details):
void * mmap(
void *start, /* Often 0, preferred starting address */
size_t length, /* Length of the mapped area */
int prot, /* Permissions: read, write, execute */
int flags, /* Options: shared mapping, private copy... */
int fd, /* Open file descriptor */
off_t offset /* Offset in the file */
);
You get a virtual address you can write to or read from.

How to Implement mmap - Kernel Space
Character driver: implement an mmap ﬁle operation and add it
to the driver ﬁle operations:
int (*mmap) (
struct file *, /* Open file structure */
struct vm_area_struct * /* Kernel VMA structure */
);
Initialize the mapping.
Can be done in most cases with the remap_pfn_range()
function, which takes care of most of the job.

remap pfn range()
pfn: page frame number
The most signiﬁcant bits of the page address (without the
bits corresponding to the page size).
#include <linux/mm.h>
int remap_pfn_range(
struct vm_area_struct *, /* VMA struct */
unsigned long virt_addr, /* Starting user
* virtual address */
unsigned long pfn, /* pfn of the starting
* physical address */
unsigned long size, /* Mapping size */
pgprot_t prot /* Page permissions */
);

Simple mmap Implementation
static int acme_mmap
(struct file * file, struct vm_area_struct *vma)
{
size = vma->vm_end - vma->vm_start;
if (size > ACME_SIZE)
return -EINVAL;
if (remap_pfn_range(vma,
vma->vm_start,
ACME_PHYS >> PAGE_SHIFT,
size,
vma->vm_page_prot))
return -EAGAIN;
return 0;
}

devmem2
http://free-electrons.com/pub/mirror/devmem2.c, by
Jan-Derk Bakker
Very useful tool to directly peek (read) or poke (write) I/O
addresses mapped in physical address space from a shell
command line!
Very useful for early interaction experiments with a device,
without having to code and compile a driver.
Uses mmap to /dev/mem.
Examples (b: byte, h: half, w: word)
devmem2 0x000c0004 h (reading)
devmem2 0x000c0008 w 0xffffffff (writing)
devmem is now available in BusyBox, making it even easier to
use.

mmap Summary
The device driver is loaded. It defines an mmap file operation.
A user space process calls the mmap system call.
The mmap file operation is called.
It initializes the mapping using the device physical address.
The process gets a starting address to read from and write to
(depending on permissions).
The MMU automatically takes care of converting the process
virtual addresses into physical ones.
Direct access to the hardware without any expensive read or
write system calls

Backup slides
Introduction to Git

What is Git?
A version control system, like CVS, SVN, Perforce or
ClearCase
Originally developed for the Linux kernel development, now
used by a large number of projects, including U-Boot,
GNOME, Buildroot, uClibc and many more
Contrary to CVS or SVN, Git is a distributed version control
system
No central repository
Everybody has a local repository
Local branches are possible, and very important
Easy exchange of code between developers
Well-suited to the collaborative development model used in
open-source projects

Install and Setup
Git is available as a package in your distribution
sudo apt-get install git
Everything is available through the git command
git has many commands, called using git <command>, where
<command> can be clone, checkout, branch, etc.
Help can be found for a given command using
git help <command>
Set up your name and e-mail address
They will be referenced in each of your commits
git config --global user.name ’My Name’
git config --global user.email me@mydomain.net

Clone a Repository
To start working on a project, you use Git’s clone operation.
With CVS or SVN, you would have used the checkout
operation, to get a working copy of the project (latest version)
With Git, you get a full copy of the repository, including the
history, which allows to perform most of the operations oﬄine.
Cloning Linus Torvalds’ Linux kernel repository
git:// is a special Git protocol. Most repositories can also
be accessed using http://, but this is slower.
After cloning, in linux/, you have the repository and a
working copy of the master branch.

Explore the History
git log will list all the commits. The latest commit is the
ﬁrst.
commit 4371ee353c3fc41aad9458b8e8e627eb508bc9a3
Author: Florian Fainelli <florian@openwrt.org>
Date: Mon Jun 1 02:43:17 2009 -0700
MAINTAINERS: take maintainership of the cpmac Ethernet driver
This patch adds me as the maintainer of the CPMAC (AR7)
Ethernet driver.
Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
git log -p will list the commits with the corresponding diﬀ
The history in Git is not linear like in CVS or SVN, but it is a
graph of commits
Makes it a little bit more complicated to understand at the
beginning
But this is what allows the powerful features of Git
(distributed, branching, merging)

Visualize the History: gitk
gitk is a graphical tool that represents the history of the
current Git repository
Can be installed from the gitk package

Visualize the History: cgit
Another great tool is cgit, a web interface to Git. For the
kernel, it is available at http://git.kernel.org/

Update your Repository
The repository that has been cloned at the beginning will
change over time
Updating your local repository to reﬂect the changes of the
remote repository will be necessary from time to time
git pull
Internally, does two things
Fetch the new changes from the remote repository
(git fetch)
Merge them in the current branch (git merge)

Tags
The list of existing tags can be found using
git tag -l
To check out a working copy of the repository at a given tag
git checkout <tagname>
To get the list of changes between a given tag and the latest
available version
git log v2.6.30..master
List of changes with diﬀ on a given ﬁle between two tags
git log -p v2.6.29..v2.6.30 MAINTAINERS
With gitk
gitk v2.6.30..master

Branches
To start working on something, the best is to make a branch
It is local-only, nobody except you sees the branch
It is fast
It allows to split your work on diﬀerent topics, try something
and throw it away
It is cheap, so even if you think you’re doing something small
and quick, do a branch
Unlike other version control systems, Git encourages the use
of branches. Don’t hesitate to use them.

Branches
Create a branch
git branch <branchname>
Move to this branch
git checkout <branchname>
Both at once (create and switch to branch)
git checkout -b <branchname>
List of local branches
git branch
List of all branches, including remote branches
git branch -a

Making Changes
Edit a file with your favorite text editor
Get the status of your working copy
git status
Git has a feature called the index, which allows you to stage
your commits before committing them. It allows to commit
only part of your modifications, by file or even by chunk.
On each modified file
git add <filename>
Then commit. No need to be on-line or connected to commit
Linux requires the -s option to sign your changes
git commit -s
If all modified files should be part of the commit
git commit -as

Sharing Changes: E-mail
The simplest way of sharing a few changes is to send patches
by e-mail
The first step is to generate the patches
git format-patch -n master..<yourbranch>
Will generate one patch for each of the commits done on
<yourbranch>
The patch files will be 0001-...., 0002-...., etc.
The second step is to send these patches by e-mail
git send-email --compose --
to email@domain.com 00*.patch
Required Ubuntu package: git-email
In a later slide, we will see how to use git config to set the
SMTP server, port, user and password.

Sharing Changes: Your Own Repository
If you do a lot of changes and want to ease collaboration with
others, the best is to have your own public repository
Use a git hosting service on the Internet:
Gitorious (https://gitorious.org/)
Open Source server. Easiest. For public repositories.
GitHub (https://github.com/)
For public repositories. Have to pay for private repositories.
Publish on your own web server
Easy to implement.
Just needs git software on the server and ssh access.
Drawback: only supports http cloning (less eﬃcient)
Set up your own git server
Most ﬂexible solution.
Today’s best solutions are gitolite
(https://github.com/sitaramc/gitolite) for the server
and cgit for the web interface
(http://git.zx2c4.com/cgit/about/).

Sharing changes: HTTP Hosting
Create a bare version of your repository
cd /tmp
git clone --bare ~/project project.git
touch project.git/git-daemon-export-ok
Transfer the contents of project.git to a publicly-visible
place (reachable read-only by HTTP for everybody, and
read-write by you through SSH)
Tell people to clone
http://yourhost.com/path/to/project.git
Push your changes using
git push ssh://yourhost.com/path/toproject.git
srcbranch:destbranch

Tracking Remote Trees
In addition to the oﬃcial Linus Torvalds tree, you might want
to use other development or experimental trees
The OMAP tree at git://git.kernel.org/pub/scm/
linux/kernel/git/tmlind/linux-omap.git
The stable realtime tree at git://git.kernel.org/pub/
scm/linux/kernel/git/rt/linux-stable-rt.git
The git remote command allows to manage remote trees
git remote add rt git://git.kernel.org/pub/scm/
linux/kernel/git/rt/linux-stable-rt.git
Get the contents of the tree
git fetch rt
Switch to one of the branches
git checkout rt/master

About Git
We have just seen the very basic features of Git.
A lot more interesting features are available (rebasing,
bisection, merging and more)
References
Git Manual
http://schacon.github.com/git/user-manual.html
Git Book
http://git-scm.com/book
Git oﬃcial website
http://git-scm.com/
Video: James Bottomley’s tutorial on using Git
http://free-electrons.com/pub/video/2008/ols/
ols2008-james-bottomley-git.ogg

Practical lab - Going further: git
Get familiar with git by
contributing to a real project: the
Linux kernel
Send your patches to the
maintainers and mailing lists.

Linux Kernel and Driver Development Training

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to Linux Kernel and Driver Development Training (20)

More from Stephan Cadene (8)

Recently uploaded (20)

Linux Kernel and Driver Development Training