[ECOS] eCos in FOSDEM
Peter Vandenabeele
peter.vandenabeele@mind.be
Tue Jan 28 14:03:00 GMT 2003
On Tue, Jan 28, 2003 at 03:49:59PM +0200, Harri Siirtola wrote:
> Hi,
>
> I tried to find more info about when and where the eCos meeting is. I read
> somewhere that it's not 100% certain that the public meeting will take
> place, it depends on number of attendants. Is the schedule clear already?
Harri,
There are 2 elements:
1. below is the program and abstract list of the embedded & kernel track
that we organized (Nick will make an eCos presentation on Sunday).
2. there is an open eCos meeting planned by the eCos maintainers, and it
is suggested to ue the embedded room for that (e.g. during the
breakfasst we planned at 10:00 - 11:00 ?).. Jonathan will organize
that.
> I'm still not sure if it's useful for me to attend. Which kind of
> developers is it best for? Digging into the bowels of eCos or building
> applications (the latter is my case, no deep knowledge about eCos itself)?
Fosdem is D-2-D event (Developer to developer) and completely NON-commercial.
Entrance is entirely free. Organisation is a little bit ad-hoc, but it was
a succes recent years. This year, we have for the first time this embedded
and kernel development track. On Saturday we also have lectures of Rchard
Stallman and John Maddog Hall. More info on fosdem.org and subscrbe to the
mailing list.
> Does anyone know if any material will appear in the net or published otherwise?
- presentations and photos are typically published on fosdem.org and
associated sites.
> Thanks,
> Harri
+++++++++++++++++
Saturday 8/2/2003
=================
14:00 - 14:45 : Keynote
opencores.org : free hardware development
Speaker : Richard Herveille
14:45 - 15:30 : Adapting Debian for Embedded Use
Speaker : Martin Michlmayr
16:00 - 16:45 : Portable programming on complex systems
Speaker : Peter De Schrijver
16:45 - 17:30 : Wonka - a JVM for embedded systems
Speaker : Chris Gray
Sunday 9/2/2003
===============
10:00 - 11:00 : breakfast + BoF's
11:00 - 11:45 : Cross-compiling in embedded systems
Speaker : Eero Tamminen
11:45 - 12:30 : eCos
Speaker : Nick Garnett
14:00 - 14:45 : RTAI
Speaker : Philippe Gerum
14:45 - 15:30 : GNU Bayonne and real-time issues in freely licensed
telephony software
Speaker : David Sugar
15:30 - 16:15 : Using C++ for real-time extensions to the Linux kernel
Speaker : Peter Soetens
++++++++++++++++
List of abstracts for embedded and kernel development track at
fosdem 2003, Sat 8 and Sun 9 Feb 2003, Brussels
Version of 28 Jan 2003, changes in the program are still possible
Saturday 8 Feb 2003:
--------------------
KEYNOTE:
OPENCORES - A REPOSITORY OF OPEN SOURCE HARDWARE IP CORES
---------------------------------------------------------
Richard Herveille
www.opencores.org
OpenCores is a repository of open source freely available Intellectual
Property (IP) cores. It is the hardware counterpart of the free software
movement. Currently OpenCores hosts 185 projects, maintained by 657
developers. Many projects are still in the development stage and some
will never reach maturity. But 65 projects are currently mature enough
to be used in designs.
In order to start development access to tools is required. For software
development the GNU toolchain provides a solid foundation. For hardware
development freely available tools are either not (yet) available, or
simply not mature enough. Testing is an even bigger problem. Software
can be tested on the same system it is being developed, by simply running
the program. Hardware can be tested by running simulations, but testing
IP cores in real hardware always requires some form of programmable logic.
Fortunately FPGA vendors provide an array of test and development boards
for their devices. Also the major FPGA vendors are releasing 'lite'
versions of their synthesis and Place&Route tools. This proved to be a
major boost for open source hardware projects.
IP cores need to be interconnected via a common bus. OpenCores adopted the
'WISHBONE System-on-Chip (SoC) Interconnect Architecture for Portable IP
Cores', or WISHBONE bus for short. WISHBONE provides a common, logical
interface between IP cores. This improves portability and reliability of
the system, and results in faster time to market. Especially when combining
a multitude of cores from different designers from different backgrounds.
Currently OpenCores maintains stewardship over the WISHBONE standard.
To prove the concept of OpenCores and the quality of the cores available
from OpenCores we created a System-on-Chip (SoC). All IP cores needed to
build the SoC are available from OpenCores. The used cores include a 32bit
RISC (OpenRISC), JTAG Test Access Port for the OpenRISC, memory controller,
UART, 10/100 Ethernet MAC, VGA controller, PS/2 keyboard controller, and
an AC97 audio controller.
The 32bit OpenRISC 1200 forms the heart of the SoC. It is the first
available implementation of the OpenRISC 1000 architecture. The i
OpenRISC 1000 architecture is available under the GNU LGPL license.
The OR1200 main features are:
* Open and free architecture
* Linear 32bit logical address space
* Harvard architecture, separate data and instruction busses
* Simple and uniform length instruction set, 32bits opcode, 32/64bits data
* 32 entries general purpose register file
* 1 Branch delay slot
* 8 KByte 1-way direct mapped instruction cache
* 8 KByte 1-way direct mapped data cache
* 64 entry hash based 1-way direct mapped TLB Instruction Memory Management unit (IMMU)
* 64 entry hash based 1-way direct mapped TLB Data Memory Management unit (DMMU)
* Power management module
* JTAG Test Access Port debug module
* Tick timer module
* Programmable interrupt controller
Software development for the OpenRISC 1000 Architecture is supported by a
genuine GNU Toolchain Port and an advanced Architectural Simulator. A
uCLinux port is also available, RTEMS and Linux ports are under development.
To conclude the presentation a demo will be shown using the working
System-on-Chip.
ADAPTING DEBIAN FOR EMBEDDED USE
--------------------------------
Martin Michlmayr
Traditionally, it has been common to develop a home grown operation system
for embedded use. However, according to a study carried out by Evans Data
Corporation, there is a shift towards adopting standard systems, in
particular such based on embedded Linux. This is partly due to the fact
that embedded systems are nowadays as powerful as desktop systems were not
too long ago. While embedded systems still have unique requirements, they
are now more comparable to desktop systems than ever. This suggests the
possibility that a standard operating system and its applications can be
taken and adapted for embedded usage. The focus will be on the changes
required in Debian to be better suited for embedded use. In particular,
crucial changes to Debian's powerful package management system, dpkg, will
be discussed. For example, embedded systems must avoid installing files
which occupy storage but do not provide required functionality. This
includes documentation and translation files. The iPKG package manager,
closely related to dpkg, but developed specifically to meet the needs
to embedded systems, will be presented. Additionally, for embedded use,
large applications have to be replaced by smaller programs with similar
functionality. BusyBox, which combines tiny versions of many common UNIX
utilities into a single small executable, is important in this regard.
Also, it will be discussed what is needed to replace bash with dash, a
POSIX compliant shell much smaller than bash. Alternatives to the powerful
but big libc6 library, such as dietlibc, will be presented.
PORTABLE PROGRAMMING ON COMPLEX SYSTEMS
---------------------------------------
Peter De Schrijver
Portable programming is an important requirement for system level programming,
especially in embedded systems. However, as CPU and system architectures have
become more complex over the years, this is a growing challenge. System
programmers have to cope with such things as noncoherent I/O, out of order
memory accesses, various busses and address tranlations next to things like
endianess mismatches and alignment issues.
We will discuss these issues and give ideas on how to build a device
driver framework to allow for portable system programming.
WONKA - A JVM FOR EMBEDDED SYSTEMS
----------------------------------
Chris Gray
Wonka is a clean-room implementation of the Virtual Machine and class library
specifications for Java, developed by Belgian company Acunia. It was
developed specifically to meet Acunia's needs for Java in embedded and mobile
systems. Wonka is Java2 compliant, and supports the CDC/FC profile plus some
additional packages. A selective JIT compiler is currently under test.
Wonka is available under a BSD licence. In this presentation we will discuss
some of the challenges involved in implementing a high level of Java
functionality in an embedded context, and outline some current areas of
development and research.
Sunday 9 Feb 2003:
------------------
CROSS COMPILING IN EMBEDDED SYSTEMS
-----------------------------------
Eero Tamminen
Abstract not yet available.
eCos (EMBEDDED CONFIGURABLE OPERATING SYSTEM)
---------------------------------------------
Nick Garnett
eCos is a highly configurable, portable, open source, royalty free
Operating System for real time embedded applications. This presentation
provides a brief, mainly technical, overview of eCos: its history,
implementation, configuration system and future directions.
One of the key technological innovations in eCos is a highly
sophisticated configuration system. The configuration system allows the
application writer to impose their requirements on the run-time
components, both in terms of their functionality and implementation.
eCos provides both a graphical configuration tool that runs on both
Windows and Linux, and a command line tool that can be used to automate
the configuration process.
eCos is designed to be portable to a wide range of target architectures
and target platforms including 16, 32, and 64 bit architectures, MPUs,
MCUs and DSPs. The eCos kernel, libraries and runtime components are
layered on the Hardware Abstraction Layer (HAL), and thus will run on
any target once the HAL and relevant device drivers have been ported to
the target's processor architecture and board. Currently eCos supports a
large range of different target architectures (ARM, Intel StrongARM and
XScale, Hitachi SH3/4, Intel x86, MIPS, Matsushita AM3x, NEC V850,
Motorola PowerPC, SPARC and several others), including many of the
popular variants of these architectures and evaluation boards.
eCos has been designed to support applications with real-time
requirements, providing features such as full preemptability, minimal
interrupt latencies, and all the necessary synchronization primitives,
scheduling policies, and interrupt handling mechanisms needed for these
type of applications. eCos also provides all the functionality required
for general embedded application support including device drivers,
memory management, exception handling, networking, file systems,
standard C and math libraries, POSIX and uITRON compatibility etc.
To permit proprietary application code to be linked with eCos without
itself being forced to be released under the GPL, eCos is distributed
under the GPL license with an exception. It is also royalty and buyout
free, making it an attractive alternative to license inhibited, closed
source or commercial real-time operating systems.
RTAI
----
Philippe Gerum
RTAI is a GNU/Linux-based real-time interface for applications with
strict timing constraints which has recently been ported over Adeos, a
flexible environment for sharing hardware resources among concurrent
operating systems.
We will discuss the process of migrating the RTAI core over the Adeos
nanokernel, describing the pipelining scheme Adeos implements in order
to dispatch hardware and system events to RTAI, so that real-time
tasks have absolute priority over regular Linux activities.
We will conclude by considering the new paths this migration opens for
extending RTAI while still keeping a high degree of determinism.
GNU BAYONNE AND REAL-TIME ISSUES IN FREELY LICENSED TELEPHONY SOFTWARE
----------------------------------------------------------------------
David Sugar
http://www.gnu.org/software/bayonne
In understanding the needs of building current and next generation telephony
services we have to consider that in the past such services were often built from
proprietary realtime operating systems. Changes in the last 20 years in the power
and capabilities of commercial off the shelf commercial platforms and free oper-
ating systems have largely replaced the need for specialized proprietary realtime
systems, but not the realtime requirements that still exist today. This paper is meant
to provide an overview and introduction of the challenges and needs in providing
realtime Linux services to support current and next generation telephone networks,
but is not meant to be a rigorous analysis on the topic.
USING C++ FOR REAL-TIME EXTENSIONS TO THE LINUX KERNEL
------------------------------------------------------
Peter Soetens
The Linux kernel is well known for its portability and extendability.
One of the recent extensions is bringing realtime operating systems
under the kernel hood so that realtime software can run on the same
platform as the user(-space) programs. Examples are RTAI (www.rtai.org)
and RTLinux (www.fsmlabs.com). Since realtime programs introduce a
complexity far beyond device drivers, better programming methodologies
become a must to control the realtime softwares complexity. The C++
programming language facilitates object oriented development and design,
which can be used to manage complex programs. There are reports of
realtime operating systems being written entirely in C++ (e.g. eCos,
(http://sources.redhat.com/ecos/).
I will introduce how the Linux kernel can be extended to support
C++ to the extent that all C++ programs can be brought in the kernel,
except for exceptions. A second logical step is to bring the standard
C++ library into the kernel, which will also be explained.
--
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss
More information about the Ecos-discuss
mailing list